Blockchain Technologies

Keshav Kaushik
Ishu Sharma Editors

Next-Generation
Cybersecurity
AI, ML, and Blockchain
Blockchain Technologies

Series Editors
Dhananjay Singh , Department of Electronics Engineering, Hankuk University of
Foreign Studies, Yongin-si, Korea (Republic of)
Jong-Hoon Kim, Kent State University, Kent, OH, USA
Madhusudan Singh , Endicott College of International Studies, Woosong
University, Daejeon, Korea (Republic of)
This book series aims to provide details of blockchain implementation in technology and
interdisciplinary fields such as Medical Science, Applied Mathematics, Environmental Science,
Business Management, and Computer Science. It covers an in-depth knowledge of blockchain
technology for advance and emerging future technologies. It focuses on the Magnitude: scope,
scale & frequency, Risk: security, reliability trust, and accuracy, Time: latency & timelines,
utilization and implementation details of blockchain technologies. While Bitcoin and
cryptocurrency might have been the first widely known uses of blockchain technology, but
today, it has far many applications. In fact, blockchain is revolutionizing almost every industry.
Blockchain has emerged as a disruptive technology, which has not only laid the foundation for
all crypto-currencies, but also provides beneficial solutions in other fields of technologies. The
features of blockchain technology include decentralized and distributed secure ledgers,
recording transactions across a peer-to-peer network, creating the potential to remove
unintended errors by providing transparency as well as accountability. This could affect not only
the finance technology (crypto-currencies) sector, but also other fields such as:
Crypto-economics Blockchain
Enterprise Blockchain
Blockchain Travel Industry
Embedded Privacy Blockchain
Blockchain Industry 4.0
Blockchain Smart Cities,
Blockchain Future technologies,
Blockchain Fake news Detection,
Blockchain Technology and It’s Future Applications
Implications of Blockchain technology
Blockchain Privacy
Blockchain Mining and Use cases
Blockchain Network Applications
Blockchain Smart Contract
Blockchain Architecture
Blockchain Business Models
Blockchain Consensus
Bitcoin and Crypto currencies, and related fields

The initiatives in which the technology is used to distribute and trace the communication start
point, provide and manage privacy, and create trustworthy environment, are just a few
examples of the utility of blockchain technology, which also highlight the risks, such as
privacy protection. Opinion on the utility of blockchain technology has a mixed conception.
Some are enthusiastic; others believe that it is merely hyped. Blockchain has also entered the
sphere of humanitarian and development aids e.g. supply chain management, digital identity,
smart contracts and many more. This book series provides clear concepts and applications of
Blockchain technology and invites experts from research centers, academia, industry and
government to contribute to it.
If you are interested in contributing to this series, please contact [email protected] OR
[email protected]
Keshav Kaushik · Ishu Sharma
Editors

Next-Generation
Cybersecurity
AI, ML, and Blockchain
Editors
Keshav Kaushik Ishu Sharma
University of Petroleum and Energy Studies Chitkara University Institute of Engineering
Dehradun, Uttarakhand, India and Technology
Chitkara University
Rajpura, Punjab, India

ISSN 2661-8338 ISSN 2661-8346 (electronic)

Blockchain Technologies
ISBN 978-981-97-1248-9 ISBN 978-981-97-1249-6 (eBook)
https://doi.org/10.1007/978-981-97-1249-6

© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature
Singapore Pte Ltd. 2024

This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.

This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721,
Singapore

Paper in this product is recyclable.

This book is dedicated to my beloved
Parents—Sh. Vijay Kaushik, Smt. Saroj
Kaushik,
Wife—Priyanka, daughter Kashvi and son
Harshiv. May god always bless us, Har Har
Mahadev!!
Keshav Kaushik
This book is affectionately dedicated to my
constant wellspring of strength, Dr. Jagdeep
Sharma, my husband. His unwavering
support, wisdom, and love have been the
guiding lights that have illuminated my path.
This dedication is a small token of
appreciation for the endless moments of
encouragement and shared joys.
Dr. Ishu Sharma
Preface

In a time when the digital world is changing at a never-before-seen rate, cybersecurity

is a constantly changing battlefield. New frontiers are being opened by technological
breakthroughs, but they also bring obstacles that call for creative answers. In this ever-
changing world, “Next-Generation Cybersecurity—AI, ML, and Blockchain” shines
as a lighthouse, providing an in-depth examination of how Artificial Intelligence
(AI), Machine Learning (ML), and Blockchain technologies combine to strengthen
cybersecurity defenses.
Our journey in the book covers up to 19 chapters and each chapter, written by top
authorities in the subject, summarizes the most recent advancements and is comple-
mented with case studies, real-world examples, and useful applications. This book
presents a current and thorough assessment of the interaction between these trans-
formational technologies and the field of cybersecurity, aiming to close a significant
gap in the literature.
This book provides a route across the complex landscape of AI, ML, and
blockchain in cybersecurity, whether your goal is to start cutting-edge research,
strengthen your organization’s defenses, or get a deeper understanding of the field.

Dehradun, India Keshav Kaushik

Rajpura, India Ishu Sharma

vii
Contents

Introduction to Cybersecurity with AI, ML, and Blockchain . . . . . . . . . . . 1

Krishna Chaitanya Nosina and T. Swarna Latha
Opportunities and Challenges in New Generation Cyber Security
Applications Using Artificial Intelligence, Machine Learning
and Block Chain . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
Sarthak Malik, Praveen Kumar Malik, and Arshi Naim
AI and Blockchain for Secure Data Analytics . . . . . . . . . . . . . . . . . . . . . . . . 39
Shweta Mayor Sabharwal, Sonia Chhabra, and Manpreet Kaur Aiden
Synergizing Artificial Intelligence and Blockchain . . . . . . . . . . . . . . . . . . . . 83
Priyanka Tyagi, Neelam Shrivastava, Sakshi, and Vishal Jain
Blockchain-Based Smart Contracts: Technical and Usage Aspects . . . . . . 99
Gulbir Singh
An Impact of Cyber Security and Blockchain in Healthcare
Industry: An Implementation Through AI . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Monika Dandotiya and Indrajit Ghosal
Deep Learning and Blockchain Applications in Healthcare Sector
Using Imaging Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
Monika Sethi, Jatin Arora, Vidhu Baggan, Jyoti Verma,
and Manish Snehi
Healthcare Data Security Using AI and Blockchain: Safeguarding
Sensitive Information for a Safer Society . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 159
Jyoti Upadhyay, Sunil Kumar Singh, Naresh Kumar Kar,
Manoj Kumar Pandey, Priyanka Gupta, and Preeti Tiwari
Future of Electronic Healthcare Management: Blockchain
and Artificial Intelligence Integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
Parag Verma, Chakka Mohana Rao, Prudhvi Kumar Chapalamadugu,
Rajeev Tiwari, and Shuchi Upadhyay

ix
x Contents

Impact of Neural Network on Malware Detection . . . . . . . . . . . . . . . . . . . . . 219

Aaeen Alchi, Kiranbhai Dodiya, and Varayogula Sai Niveditha
Evaluating Different Malware Detection Neural Network
Architectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 243
Harinadh Varikuti and Valli Kumari Vatsavayi
Protecting Your Assets: Effective Use of Cybersecurity Measures
in Banking Industries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 265
Sulagna Das and Devraj Ganguly
Revolutionizing Banking with Blockchain: Opportunities
and Challenges Ahead . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 287
Shilpa Mahajan and Meghna Nanda
Leveraging AI and Blockchain for Enhanced IoT Cybersecurity . . . . . . . 305
Iraq Ahmad Reshi and Sahil Sholla
Dissecting the Role of Women in Cybersecurity and Information
Technology: A Medical Perspective . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
Wasswa Shafik
A Comprehensive Exploration of Advancements and Applications
of Digital Watermarking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 351
Ealuri Rohith, Buggaveeti Padmaja, and V. M. Manikandan
Optimizing Drug Discovery: Molecular Docking with Glow-Worm
Swarm Optimization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 369
Vijaya Sindhoori Kaza, P. R. Anisha, and C. Kishor Kumar Reddy
Bridging the Gap Between Learning and Security: An Investigation
of Blended Learning Systems and Ransomware Detection . . . . . . . . . . . . . 419
Vikas Pandey, Aman Jolly, Praveen Kumar Malik, and Shashikant
Ethical Considerations in AI-Based Cybersecurity . . . . . . . . . . . . . . . . . . . . 437
Keshav Kaushik, Aadil Khan, Ankita Kumari, Ishu Sharma,
and Rajat Dubey
About the Editors

Keshav Kaushik is an experienced educator with around ten years of teaching and
research experience in cybersecurity, digital forensics, and the Internet of Things.
He is working as an Assistant Professor (Selection Grade) in the systems cluster
under the School of Computer Science at the University of Petroleum and Energy
Studies, Dehradun, India. He has published 110+ research papers in International
Journals and has presented at reputed International Conferences. He is a Certified
Ethical Hacker (CEH) v11, CQI and IRCA Certified ISO/IEC 27001:2013 Lead
Auditor, Quick Heal Academy Certified Cyber Security Professional (QCSP), and
IBM Cybersecurity Analyst. He acted as a keynote speaker and delivered 50+
professional talks on various national and international platforms. He has edited
over twenty books with reputed international publishers like Springer, Taylor and
Francis, IGI Global, Bentham Science, etc. He has chaired various special sessions
at international conferences and also served as a reviewer in peer-reviewed journals
and conferences. Currently, he is also serving as a Vice Chairperson of the Meerut
ACM Professional Chapter and is also a brand ambassador for Bentham Science.
Moreover, he is also serving as a guest editor in the IEEE Journal of Biomedical
and Health Informatics (J-BHI) (IF:7.7).

Dr. Ishu Sharma is currently working as an Assistant Professor with Chitkara

University Research and Innovation Network (CURIN), Chitkara University, Punjab.
She has more than 14 years of experience in Academics and Management. Her
research area includes Cybersecurity, Blockchain, Wireless Ad hoc Network, Wire-
less Sensor Network, and Artificial Intelligence. She completed Ph.D. in Computer
Science and Engineering from Chitkara University, Punjab. She received a silver
medal in Master of Engineering. She cleared the examination for Associate Member
of the Institution of Engineers in the minimum duration. She was state-level topper
in Diploma from the State Board of Technical Education, Haryana. She is associated
with multiple EdTech Platforms and invited as a Guest Faculty in IIT Kanpur for

xi
xii About the Editors

e-Masters Course in Blockchain. She has published 40+ patents and 80+ Scopus-
indexed research papers in various international journals and conferences. She has
been invited as an Expert for Cybersecurity tool for various FDP and Guest Lectures
in academic institutions and industries. She is guiding multiple Ph.D. and M.Tech.
research scholars in the area of Cybersecurity and Blockchain.
Introduction to Cybersecurity with AI,
ML, and Blockchain

Krishna Chaitanya Nosina and T. Swarna Latha

Abstract This chapter provides an introduction of cybersecurity with the aim of

combining Artificial Intelligence (AI), along with Machine Learning (ML), as well
as Blockchain technologies. In today’s world, the need for robust cybersecurity
measures is of utmost importance. The traditional approaches of security methods
are often unable to defend the current evolving cyber threats, necessitating innovative
solutions. AI and ML have emerged as the best technology in order to enhance the
cybersecurity for processing of unlimited data, and also able to detect the regular
patterns, and to identify the potential threats. Applications of AI and ML in cyber-
security include threat detection and analysis, intrusion detection and prevention
systems, user and entity behavior analytics, and malware analysis. Blockchain, with
its decentralized ability offers additional security benefits such as data integrity and
secure identity management. With the proposed technologies, this chapter provides
an opportunity for building more robust and intelligent security systems. However,
the challenges such as privacy concerns and the evolving nature of cyber threats need
to be addressed. This chapter serves as a basic level for understanding the essence
of combining the technologies of AI, ML, and Blockchain in cybersecurity, setting
the stage for further exploration of practical implementation and best practices in
subsequent chapters.

Keywords Cybersecurity · Artificial intelligence (AI) · Machine learning (ML) ·

Blockchain · Threat detection · Data integrity · Security technologies

K. C. Nosina (B)
Department of ECE, RSR Engineering College, Kadanuthala, Kavali 524142, India
e-mail: [email protected]
T. Swarna Latha
Department of CSE, PBR Visvodaya Institute of Technology and Science, Kavali 524201, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 1
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_1
2 K. C. Nosina and T. Swarna Latha

1 Introduction

With the rise of digital technologies and the increasing sophistication of cyber threats,
organizations and individuals need robust security measures to protect their sensitive
data and systems. This chapter serves as an introduction to cybersecurity, wherein
it focuses by combining the technologies of AI, ML, with Blockchain as powerful
tools for enhancing security.

1.1 Understanding Cybersecurity

Cybersecurity [1] mainly aims to protect the computers, networks, as well as data
from the illicit access, chance of theft, as well as damage, or even disruption. In
the current internet world, where digital technologies [2] are deeply integrated into
various aspects of our lives, the need for the better cybersecurity actions [3] has
become very important.
A range of critical elements, as shown in Fig. 1, that are essential for the secure-
ment of Digital Assets and Data Against Cyber Threats will be covered by cyber
security. Human beings play a vital role in addressing and managing the cyber envi-
ronment, which requires ongoing user education so as to promote best practices and
awareness. Application security is concerned with the safety of software and systems
so as to avoid vulnerabilities and unauthorised access. The regulatory framework for
cyber security consists of legislation and policies to guide organizations in ensuring
their compliance with cybersecurity standards, as well as protecting user privacy.
Implementing measures to ensure the safe transmission of data, as well as protec-
tion against unauthorised access and breaches are part of network security. In view
of the cyber threat or disasters, disaster recovery and continuity planning ensure
that unexpected events can be prepared for as quickly as possible in order to enable
uninterrupted business operations. Together, these features contribute to creating an
effective cyber ecosystem that fosters a secure and stable digital environment.
The primary objective of cybersecurity is to provide the privacy [4], veracity, and
ease of use of digital assets. In particular, privacy safeguards important information.
Integrity [5] ensures that data remains intact, unaltered, and free from unautho-
rized modifications. Availability [6] ensures that systems and data are accessible and
usable when needed. Cyber security has a wide variety of strategies, technologies
and processes. It involves identifying and assessing [7] vulnerabilities in computer
systems and networks, implementing protective measures to prevent attacks, and
developing incident response plans to effectively mitigate and recover from security
incidents. Cyber threats exist in various forms, and they can have a serious impact.
Malware [8, 9], such as viruses, worms, and ransomware, is designed to infiltrate
and disrupt computer systems. To gain access to sensitive information, phishing
attacks target individuals via misleading email or web sites. Social engineering
[10, 11] exploits human vulnerabilities to manipulate those into disclosing secret
Introduction to Cybersecurity with AI, ML, and Blockchain 3

Fig. 1 Components in cybersecurity

information or compromise unlawful access. Other threats [12] such as denial-of-

service (DoS) attacks, where attackers devastate a system with data transfer to make
it unapproachable, and insider threats, where trusted individuals misuse their access
privileges.
Cyber security professionals use a variety of techniques and technologies in order
to effectively counter such threats [12]. The deployment of a firewall, intrusion detec-
tion system and antivirus software to eliminate the unauthorised access as well as
detecting and combating any illegal activities [13–15] should also be undertaken in
this context. In order to ensure confidentiality and security of this data, encryption
shall be applied during transmission and storage. To verify the identity of individuals
accessing systems [16] and data so that unauthorised access is prevented, access
controls and authentication mechanisms are used. Cybersecurity is a constantly
evolving field due to the rapid change with respect to the technology [17–19] and
appearance of new threats. Cyber security experts need to keep up with the evolving
attackers and constantly refresh their knowledge and skills in order to stay ahead.
This includes staying informed about the latest threats, vulnerabilities, and security
best practices, as well as engaging in regular training and professional development
[20].
4 K. C. Nosina and T. Swarna Latha

1.2 AI and ML in Cybersecurity

AI and ML have ruled as most effective tools [21–23] in the field of cybersecurity,
revolutionizing the way threats are detected, analyzed, and mitigated. These tech-
nologies [24] enable cybersecurity professionals to effectively address the growing
complexity and volume of cyber threats in today’s interconnected world.
Artificial intelligence means the creation of machines capable of performing a
job [25], such as learning to solve problems, making decisions and recognizing
patterns, that requires high levels of Human Intelligence. The portion of AI dealing
with algorithm [26] design is called ML. To increase the detection, analysis and
response capabilities of threats, artificial intelligence and machine learning are being
incorporated into cyber security. These technologies [27] can verify large amounts
of information, capable of identifying the patterns, and to detect the incidents where
human analysts miss. By continuously learning and adapting, AI and ML systems
can improve their accuracy and also efficiency over time.
One of the significant contribution [28] of AI and ML in cybersecurity is detection
of severe threats as well as analyze them efficiently. AI and ML algorithms have the
ability to read the network traffic, to log files, and possess the behavior of the user
to find suspicious activities and potential threats. They can detect known malware
signatures [29] and also identify previously unknown threats using anomaly detec-
tion techniques. This helps in early detection of potential cyber-attacks and enables
organizations to take proactive measures to prevent or mitigate the impact of such
attacks. AI and ML techniques are also used in Intrusion Detection and Prevention
Systems (IDPS). The systems [30] monitor network traffic in real time to detect and
block malicious activities. Artificial Intelligence and Machine Learning algorithms
can, by analysing patterns and behaviors, find out what’s going on in the network as
well as which kind of threat is likely to attack them; they can automatically perform
mitigation measures.
User and Entity Behavior Analytics (UEBA) [31] is another area where AI and
ML are extensively utilized. UEBA systems establish baseline behavior patterns for
users and entities within an organization’s network. They monitor and analyze user
activities [32], network access, and data interactions to detect anomalous behavior
of compromised user accounts. By identifying such threats in real-time [33], orga-
nizations can proactively respond and prevent potential data breaches. AI and ML
also play a crucial role in malware analysis and antivirus solutions. Malware analysis
involves studying malicious software [34] to understand its behavior, capabilities,
and potential impact. This analysis may be automated by the use of artificial intel-
ligence and machine learning techniques, helping to identify and classify threats
much more quickly. By increasing the detection capabilities of ML models, they are
continuously adapting to new threats and improving the effectiveness of antivirus
solutions.
AI and ML have become obligatory in the field of cybersecurity [35]. These
technologies offer superior capabilities for identification of threats, analysis, and
response, allowing organizations to enhance their security posture in the face of
Introduction to Cybersecurity with AI, ML, and Blockchain 5

evolving cyber threats. In today’s increasingly complex digital environment, cyber

security experts can use AI and machine learning to detect and mitigate risks more
effectively, reduce the response times and remain at least one step ahead of those
responsible for malicious activities.

1.3 Applications of AI, ML and Block Chain in Cybersecurity

These technologies have numerous applications across various cybersecurity

domains, enabling organizations to enhance their security posture and effectively
combat evolving cyber threats [36].
(a) Threat Detection and Analysis:
The threat detection and analysis is done by using AI with ML [37]. It is implemented
with continuously learning from historical data, these algorithms can improve their
accuracy and efficiency over time, enabling the early detection of cyber-attacks [38].
Using an artificial intelligence and blockchain methodology, Fig. 2 shows the
detection and analysis of threats. The way that organizations approach the detection
and defence of cyber threats has been transformed by these cutting edge technolo-
gies. Cybersecurity professionals can enhance their ability to detect threats, react
more quickly and build resilience defenses through the unique capabilities of Artifi-
cial Intelligence, Machine Learning or Blockchain. Below, we delve into how each
technology contributes to the process of threat detection and analysis:

(i). Artificial Intelligence (AI) in Threat Detection: AI in cybersecurity involves

the application of intelligent algorithms that mimic human cognition to process
vast amounts of data and detect for potential threats. The various purposes in
threat detection using AI are [37].

Fig. 2 Threat detection and analysis

6 K. C. Nosina and T. Swarna Latha

• Behavior Analysis: AI-powered systems can establish a baseline of normal

behavior for users, systems, and networks. Any deviation from this baseline
can be flagged as suspicious, potentially indicating an ongoing attack.
• Anomaly Detection: AI algorithms can identify abnormal patterns or activi-
ties that do not match known signatures or established behavior. This helps
detect zero-day attacks and novel threats that traditional signature-based
systems may miss.
• Predictive Analytics: The technology can analyze chronological data to
predict potential future threats and vulnerabilities, enabling organizations
to take proactive measures to prevent attacks.
• Natural Language Processing (NLP): This can be employed to analyze
unstructured data.
(ii). Machine Learning (ML) in Threat Detection: The training algorithms are
learned from the available data and also its performance is improved with ML.
ML techniques play a vital task in cybersecurity threat detection as well as
analysis [37]:
• Signature-less Detection: ML algorithms can identify unknown threats
by learning from past incidents and recognizing patterns that indicate
malicious behavior.
• Classification and Clustering: ML models can categorize threats into
different classes based on their attributes, helping security teams prioritize
and respond to incidents more effectively.
• Automated Response: ML-powered systems can autonomously respond to
low-level threats, reducing the burden on security personnel and enabling
faster incident resolution.
• Data Preprocessing: ML can assist in data cleaning, transformation, and
feature extraction, which are crucial for preparing data for analysis.
(iii). Blockchain in Threat Detection: Blockchain technology, although initially
associated with cryptocurrency, holds promise for enhancing threat detection
and analysis in cybersecurity [38]:
• Immutable Logging: Blockchain’s decentralized and tamper-resistant
nature can be utilized for secure and reliable logging of cybersecurity events
and incidents.
• Distributed Threat Intelligence: Blockchain can facilitate secure sharing of
threat intelligence among organizations, enabling collective defense against
cyber threats.
• Secure Identity Management: Blockchain’s cryptographic techniques can
enhance identity and access management, reducing the risk of unauthorized
access and identity-based attacks.
(iv). Integration of AI, ML, and Blockchain: The true power of threat detection
and analysis in cybersecurity lies in the integration of AI, ML, and Blockchain
technologies [38]:
Introduction to Cybersecurity with AI, ML, and Blockchain 7

• AI and ML can be combined to build dynamic and adaptive threat detection

systems that learn from new data and adjust their models to stay ahead of
evolving threats.
• AI-driven anomaly detection can be applied to Blockchain networks to iden-
tify suspicious activities, ensuring the integrity and security of distributed
ledgers.
• Blockchain can be utilized to store and manage large-scale threat intelli-
gence databases securely, while AI and ML can process and analyze this
data for valuable insights.
• Smart contracts on a blockchain can autonomously trigger incident response
actions based on predefined criteria, reducing response time and minimizing
human error.

Cybersecurity practices have become radically altered by the integration of artifi-

cial intelligence, machine learning and Blockchain technology into threat detection
and analysis. The convergence presents an opportunity for organizations to set up
more responsive, adaptive and robust defense mechanisms in the face of rapidly
emerging cyber threats which enable them to enhance their Digital Asset Protec-
tion. In the future, there is a promise that more advanced and efficient detection
and analysis of threats will be possible, thanks to continuous advances in these
technologies.
(b) Intrusion Detection and Prevention Systems:
IDPS is controlled [39] by AI and ML techniques to monitor network traffic in real-
time. These systems can identify and block malicious activities by analyzing network
packets and comparing them against known attack patterns [40]. ML models can also
detect previously unseen attacks using anomaly detection algorithms, identifying
deviations from normal behavior and raising alerts for further investigation.
Intrusion Detection and Prevention Systems (IDPS) play a crucial role in cyber-
security by monitoring network traffic and are shown in Fig. 3 and systems for signs
of malicious activities and responding to potential threats. This technology in IDPS
enhances their capabilities, making them more efficient, accurate, and resistant to
tampering. Let’s explore how these technologies contribute to IDPS:

(i) AI in IDPS: AI technologies provide IDPS with the ability to process large
volumes of data and make intelligent decisions based on patterns and trends.
Some ways where AI is utilized in IDPS [39] include:
• Pattern Recognition: AI can identify complex patterns of network behavior
associated with various types of attacks, enabling rapid threat detection.
• Predictive Analytics: AI can analyze historical data and predict potential
cyber threats, helping organizations take proactive measures to prevent
attacks.
• Dynamic Rule Generation: AI can autonomously generate and modify
detection rules based on real-time data, adapting to new threats without
manual intervention.
8 K. C. Nosina and T. Swarna Latha

Fig. 3 Intruder detection system

(ii) ML in IDPS: ML complements AI in IDPS by enabling systems to learn from

existing information and rise the performance w.r.t time. Some applications of
ML in IDPS include [39]:
• Anomaly Detection: ML algorithms can identify abnormal network activi-
ties that deviate from normal behavior, detecting zero-day attacks and novel
threats.
• Traffic Analysis: ML can classify network traffic and distinguish between
legitimate traffic and suspicious or malicious activities.
• Signature-less Detection: ML can detect threats without relying on pre-
defined signatures, making IDPS more adaptive to new and unknown threats.
• Incident Severity Prediction: ML models can predict the severity of detected
incidents, allowing security teams to prioritize responses efficiently.
(iii) Blockchain in IDPS: Blockchain technology enhances the security and
integrity of IDPS by providing a decentralized, immutable, and tamper-resistant
platform. Key applications of Blockchain in IDPS include [39]:
• Secure Logging: Blockchain’s distributed and append-only nature ensures
that logs of security events and incidents remain tamper-proof and trust-
worthy.
• Threat Intelligence Sharing: Blockchain enables secure sharing of threat
intelligence among different entities, fostering collaborative defense against
cyber threats.
• Identity and Access Management: Blockchain’s cryptographic features can
enhance identity management, reducing the risk of unauthorized access and
insider threats.
• Smart Contract-based Response: Blockchain’s smart contracts can auto-
mate predefined responses to specific security events, improving incident
response time and accuracy.
Introduction to Cybersecurity with AI, ML, and Blockchain 9

(iv) Integration of AI, ML, and Blockchain in IDPS: The integration of AI, ML,
and Blockchain in IDPS creates a powerful defense system that is capable of
providing more robust protection against cyber threats [40]:
• AI and ML can process large volumes of network data, identify patterns,
and detect anomalies, enabling more accurate and timely threat detection.
• Blockchain ensures the integrity of IDPS logs, preventing attackers from
tampering with critical data and covering their tracks.
• AI-driven anomaly detection and ML-based classification in IDPS can
benefit from the immutability of Blockchain, which safeguards the integrity
of the system’s decision-making process.
• Threat intelligence sharing among different organizations via Blockchain
can enhance the collective knowledge and improve the overall effectiveness
of IDPS.
In order to counter sophisticated cyber threats, the integration of Artificial Intelli-
gence, Machine Learning and Blockchain into intrusion detection and prevention
systems such as IDNPS strengthens their capabilities. These technologies allow
the Integrated Data Protection System to detect and respond more effectively to
security incidents, provide accurate threat analyses as well as improve cooperation
between cybersecurity stakeholders. By continually advancing these integrated solu-
tions, organizations can bolster their cybersecurity posture and safeguard their digital
assets in an ever-evolving threat landscape.
(c) User and Entity Behavior Analytics (UEBA):
UEBA is another important application of AI and ML in cybersecurity [39]. UEBA
systems establish baseline behavior patterns for users and entities within an orga-
nization’s network. By monitoring and analyzing user activities, network access,
and data interactions, AI and ML algorithms can detect anomalous behavior of user
accounts [41]. It helps organizations identify suspicious activities and take proactive
measures to prevent data breaches or unauthorized access.
UEBA is a cybersecurity approach that leverages the power of the technolo-
gies to detect and respond to insider threats, advanced persistent threats (APTs),
and other sophisticated cyber-attacks. UEBA focuses on monitoring and analyzing
the behavior of users and entities within an organization’s network to identify
abnormal or malicious activities. The integration of AI, ML, and Blockchain in
UEBA enhances its capabilities, making it more effective and secure. Let’s explore
how these technologies contribute to UEBA:
(i) AI in UEBA: AI technologies in UEBA enable the system to process large
data for patterns detection with anomalies in user and entity behavior. Some
key applications of AI in UEBA include [40, 41]:
• Anomaly Detection: AI algorithms can identify unusual activities or devi-
ations from normal behavior, helping to identify potential insider threats or
compromised accounts.
10 K. C. Nosina and T. Swarna Latha

• Contextual Analysis: AI can analyze the context of user actions and inter-
actions with the network to distinguish between legitimate activities and
suspicious behaviors.
• Dynamic Baselines: AI can create dynamic baselines of user and entity
behavior, adapting to changes in behavior patterns over time.
• Real-time Decision Making: AI-driven UEBA systems can make real-time
decisions, providing immediate alerts or responses to potential security
incidents.
(ii) ML in UEBA: ML complements AI in UEBA by enabling the system to learn
from data and improve its detection capabilities. Some applications of ML in
UEBA include [41]:
• User Profiling: ML algorithms can create profiles of user behavior based
on historical data, helping to identify deviations from normal behavior.
• Pattern Recognition: ML can recognize patterns associated with known
attack methods, enhancing the system’s ability to detect specific threats.
• Risk Scoring: ML models can assign risk scores to users and entities based
on their behavior, facilitating the prioritization of potential threats.
• Predictive Analytics: ML can predict future behaviors based on historical
data, allowing security teams to proactively address potential security risks.
(iii) Blockchain in UEBA: Blockchain technology strengthens UEBA by providing
a secure and tamper-resistant policy for storing & sharing sensitive data. Key
applications of Blockchain in UEBA include [40, 41]:
• Secure Data Storage: Blockchain’s decentralized and immutable nature
ensures the integrity and security of user and entity behavior data.
• Identity Verification: Blockchain’s cryptographic features can enhance
identity management, verifying the authenticity of user identities and
reducing the risk of impersonation.
• Threat Intelligence Sharing: Blockchain enables secure and anonymous
sharing of threat intelligence among different organizations, enhancing the
collective defense against cyber threats.
(iv) Integration of AI, ML, and Blockchain in UEBA: The integration of AI,
ML, and Blockchain in UEBA creates a robust and efficient system to detect
insider threats and other cyber-attacks [41]:
• AI and ML enable UEBA to process and analyze vast amounts of behavioral
data, detecting subtle patterns and anomalies that may indicate potential
threats.
• Blockchain ensures the integrity and security of user behavior data,
preventing unauthorized access or modification of sensitive information.
• AI-driven UEBA systems can benefit from the tamper-resistant nature
of Blockchain, safeguarding the accuracy and reliability of the system’s
decision-making process.
Introduction to Cybersecurity with AI, ML, and Blockchain 11

• Secure threat intelligence sharing via Blockchain enables organizations to

gain insights into emerging threats and strengthen their overall cybersecurity
posture.
These technologies, provide organizations with advanced capabilities to detect
and respond to insider threats and sophisticated cyber-attacks. By continuously
improving and advancing these integrated solutions, organizations can stay ahead
of evolving threats and protect their critical assets from illicit access activities.
(d) Malware Analysis and Antivirus Solutions:
In analyzing malware and antivirus solutions, artificial intelligence and machine
learning have also a key role to play. Malware analysis [42] is the study of malicious
software, with a view to understanding its behavior, capabilities and potential impact.
Such analysis is automated through the use of artificial intelligence and machine
learning technologies [43], enabling faster detection and classification of computer
viruses than conventional methods are able to do today. ML models can continu-
ously learn from new threats, improving the effectiveness of antivirus solutions by
enhancing their detection capabilities [44]. Additionally, AI and ML can be used in
security information and event management (SIEM) [45] systems to comprehensive
and investigate security logs. By applying AI and ML algorithms to SIEM data,
organizations can gain insights into potential security incidents, identify patterns
of attacks, and prioritize response efforts. Their applications span various domains,
including threat detection and analysis, intrusion detection and prevention [46]. By
leveraging AI and ML [47], organizations can bolster their cybersecurity defenses,
detect and mitigate risks more effectively, and stay ahead of rapidly evolving cyber
threats.

1.4 Blockchain and Cybersecurity

Blockchain technology [48], best known as the essential infrastructure for cryptocur-
rencies like Bitcoin for its potential to enhance cybersecurity. Blockchain’s decen-
tralized and immutable nature offers several security benefits, making it an attractive
solution for various cybersecurity applications. The basic diagram for block chain in
cyber security is shown in Fig. 4.

(a) Data Integrity and Immutable Audit Trails:

One of the key advantages [49] of Blockchain in cybersecurity is data integrity.

Blockchain creates a tamper-proof and transparent ledger where data can be recorded
in a decentralized manner. Every block consists of a cryptographic hash that links
it to the prior block, in order to create a secure and verifiable record. This provides
[50] assurance of data integrity. The diagram for blockchain audit trail mechanism
is shown in Fig. 5.
12 K. C. Nosina and T. Swarna Latha

Fig. 4 Block chain in cyber security

Fig. 5 Blockchain audit trail mechanism

Introduction to Cybersecurity with AI, ML, and Blockchain 13

Fig. 6 Blockchain based secure identity and access management

(b) Secure Identity and Access Management:

In the area of security for identity and access management [51], bitcoin can also
play an important role. A breach or identity theft is likely to occur in traditional
centralized ID systems. Decentralized authentication mechanisms for the purpose
of safeguarding identities are provided in Blockchain Based Identity Management
Systems. With Blockchain, users can control their digital identities, avoids the risk
of unauthorized access and theft of uniqueness [52] and is shown in Fig. 6.
(c) Supply Chain Security:
Furthermore, Blockchain technology can enhance supply chain security [52]. By
leveraging Blockchain, organizations can establish transparent and verifiable tracking
of goods and transactions. The diagram for supply chain security is shown in Fig. 7.
In terms of data sharing and verification [53], Blockchain provides a secure
and transparent infrastructure. Blockchain-based smart contracts can enforce access
control policies, facilitating secure data sharing among parties. Smart contracts
ensure that data is accessed and shared based on predefined rules and conditions,
avoiding the peril [54] of unauthorized information access and manipulation.
Blockchain technology also presents its own set of challenges [55, 56], including
scalability, privacy concerns, and the potential for new vulnerabilities. Organizations
must carefully evaluate the suitability of Blockchain for their specific cybersecurity
needs and consider the trade-offs associated with its implementation [57].
Blockchain technology has the potential to enhance cybersecurity by providing
data integrity [58], secure identity management, supply chain security, and trans-
parent data sharing. It has a unique security advantage, which makes it an attrac-
tive solution for various cybersecurity applications because of its decentralized and
immutable nature. However, organizations [59] must consider the specific require-
ments and challenges associated with Blockchain implementation to ensure its
effective integration into their cybersecurity strategies.
14 K. C. Nosina and T. Swarna Latha

Fig. 7 Blockchain in supply chain security

1.5 Convergence of AI, ML, and Blockchain

in Cybersecurity:

The convergence of these technologies [60] holds significant promise for strength-
ening cybersecurity. Each of these technologies brings unique capabilities to the table,
and their integration offers new opportunities for enhancing security, detecting and
mitigating threats, and ensuring data integrity. In this section, we will see the sights
of the convergence of AI, ML, and Blockchain in the cybersecurity and discuss how
they can complement and reinforce each other’s strengths [61] as shown in Fig. 8.
AI and ML algorithms can play a crucial role [62] in analyzing the vast amount
of data generated by Blockchain networks. Blockchain, with its decentralized and
immutable nature, provides a transparent and secure infrastructure for storing data.
One area where the convergence of [63] AI, ML, and Blockchain can be particu-
larly impactful is in threat detection and analysis. AI and ML algorithms can analyze
data stored on the Blockchain, such as transaction records and network activity, to
identify patterns and detect potential threats. By continuously learning from histor-
ical data and identifying abnormal behaviors or suspicious activities, AI and ML
algorithms can provide early warnings and enhance threat detection capabilities.
This can enable organizations to take proactive measures to prevent or mitigate
the impact of cyber attacks. Blockchain provides a decentralized and secure [62]
platform for managing digital identities, while AI and ML techniques can analyze
user behavior and access patterns to detect unauthorized access attempts or unusual
Introduction to Cybersecurity with AI, ML, and Blockchain 15

Fig. 8 Linking AI and blockchain

activities. In addition, organizations can build strong authentication mechanisms

and detect potential fraud or misuse of access credentials through a combination of
Blockchain’s secure identity management with artificial intelligence and machine
learning algorithms.
Additionally, the convergence of AI, ML, and Blockchain can improve the secu-
rity and privacy of data sharing and collaboration [64]. AI and ML algorithms can
analyze data shared through smart contracts to identify potential security risks or
anomalies. By leveraging Blockchain’s transparency and immutability, organiza-
tions can ensure the integrity and authenticity of shared data, while AI and ML
algorithms can enhance data privacy by anonymizing or encrypting sensitive infor-
mation. Blockchain can also endow with sheltered and transparent communications
[65] for storing AI and ML models. AI and ML models can be vulnerable to attacks
or tampering, as their accuracy and effectiveness rely on the integrity and confiden-
tiality of the underlying data and algorithms. By storing AI and ML models on the
Blockchain, organizations can ensure the immutability and integrity of these models,
reducing the risk of unauthorized modifications or tampering. This can enhance the
trustworthiness and reliability of AI and ML-based cybersecurity systems.
There are still many challenges to be tackled, despite the considerable opportu-
nities for convergence of Artificial Intelligence, Machine Learning and Blockchain
in cyber security. The privacy of the data stored in Blockchains is a particular chal-
lenge [66]. While it provides transparency and immutableness, the risk of disclo-
sure of sensitive information is also raised by Blockchain. The privacy implications
of protecting data need to be properly considered by enterprises and appropriate
encryption or anonymization techniques should be implemented.
16 K. C. Nosina and T. Swarna Latha

Another challenge is the need for robust governance frameworks and standards
[67] for AI, ML, and Blockchain integration in cybersecurity. As such technologies
evolve, it is essential to develop guidelines and best practices aimed at ethics and
responsibility in their use. This includes addressing bias in AI algorithms, ensuring
transparency in decision-making processes, and establishing accountability for the
outcomes of AI and ML-based cybersecurity systems.
The convergence of these technologies in cybersecurity offers exciting opportu-
nities [68] for enhancing security, threat detection, and data integrity. By leveraging
the strengths of these technologies, organizations can build more resilient and intel-
ligent security systems. However, addressing privacy concerns, establishing gover-
nance frameworks, and promoting responsible use are crucial for harnessing the full
potential of the convergence of AI, ML, and Blockchain in cybersecurity.

1.6 Challenges and Future Directions

1.6.1 Challenges

The convergence of AI, ML, and Blockchain in cybersecurity presents several chal-
lenges that need to be addressed for successful implementation [69, 70]. These
challenges include:
(a) Privacy and Confidentiality Concerns: The transparent and immutable nature
of Blockchain raises concerns [71] about the exposure of sensitive information.
Protecting privacy and ensuring confidentiality of encryption or anonymization
techniques.
(b) Ethical Considerations and Bias in AI: Ensuring ethical use of AI and
ML algorithms is crucial [72]. Bias in AI algorithms, whether due to biased
training data or algorithmic design, can have serious implications. Organiza-
tions must address biases, establish governance frameworks, and adopt ethical
AI principles to ensure fairness, transparency, and accountability.
(c) Scalability and Performance: As the volume of data continues to grow, orga-
nizations must ensure their infrastructure [73] can handle the computational
demands of AI and ML algorithms. As far as transaction speed and duration
are concerned, a blockchain network is facing challenges. In order to overcome
these problems, R&D efforts should be focused on the development of scalable
architectures and optimization algorithms.
(d) Standardization and Interoperability: Establishing standards and protocols to
integrate Artificial Intelligence, Machine Learning or Blockchain into cyberse-
curity is essential in order for these technologies to be widely adopted and inter-
operable [74]. To facilitate the integration and ensure a seamless interchange
between different systems, common frameworks, data formats and interoperable
APIs are available.
Introduction to Cybersecurity with AI, ML, and Blockchain 17

(e) Evolving Threat Landscape: The threat landscape in cybersecurity [74] is

dynamic, with new attack vectors and techniques emerging constantly. Research
and development efforts should focus on staying ahead of emerging threats and
developing proactive AI and ML algorithms that can detect and mitigate novel
attack patterns.
(f) Resilience and Adversarial Attacks: Adversarial attacks pose a significant
challenge [73] in the application of AI and ML in cybersecurity. Building robust
and resilient AI models that can detect and defend against such attacks requires
research into adversarial training, model interpretability, and anomaly detection
techniques.

1.6.2 Future Directions

In addition to addressing the challenges, several future directions can enhance the
convergence of AI, ML, and Blockchain in cybersecurity. These directions include:
(a) Advancing Privacy-Preserving Techniques: Research should focus on devel-
oping more sophisticated privacy-preserving techniques [75] to ensure confi-
dentiality while leveraging the benefits of AI, ML, and Blockchain.
(b) Trustworthy AI Systems: Future efforts should prioritize the development
of trustworthy AI systems [76] that uphold ethical considerations, fairness,
transparency, and accountability in their operations.
(c) Scalability and Performance Optimization: Continued research is needed
to enhance the scalability and performance [77] of AI, ML, and Blockchain
systems to handle the increasing volume of data and transactions effectively.
(d) Establishment of Standards and Protocols: Industry-wide collaboration is
necessary to establish standards and protocols [77] that enable interoper-
ability and seamless integration of AI, ML, and Blockchain technologies across
different cybersecurity systems.
(e) Proactive Threat Mitigation: Research should focus on proactively identifying
and mitigating emerging threats [78] through advanced AI and ML algorithms
that can detect and respond to novel attack patterns effectively.
(f) Defense Against Adversarial Attacks: Future directions should include
strengthening defenses against adversarial attacks through robust AI models
[79], adversarial training, model interpretability, and anomaly detection tech-
niques.

2 Conclusion

We conclude here with the use of the above discussed technologies in cybersecurity
holds immense potential for enhancing security, threat detection, and data integrity.
With the strengths of these technologies and addressing the associated challenges,
18 K. C. Nosina and T. Swarna Latha

organizations can build robust and intelligent cybersecurity systems. AI and ML

algorithms are capable of processing large volumes of data to identify patterns,
anomalies, and potential threats. By continuously learning from historical data, these
algorithms can improve their accuracy and efficiency, providing early warnings and
enhancing threat detection capabilities. Blockchain technology offers unique security
benefits, such as data integrity and tamper-proof records. This provides a transparent
and secure infrastructure for various cybersecurity applications, including secure
identity management, supply chain security, and transparent data sharing.
However, privacy and confidentiality concerns must be carefully considered,
and appropriate encryption or anonymization techniques should be implemented
to protect sensitive data. Ethical considerations, such as addressing bias in AI algo-
rithms and ensuring transparency and accountability, are crucial for responsible AI
usage.
Scalability and performance optimization are key areas of focus to handle the
increasing volume of data and transactions in AI, ML, and Blockchain systems.
Standardization and interoperability efforts are necessary to establish common frame-
works and protocols for seamless integration across different cybersecurity systems.
Looking ahead, future research and development efforts should aim to enhance
privacy-preserving techniques, develop trustworthy AI systems, optimize scala-
bility and performance, establish industry-wide standards and protocols, proactively
mitigate emerging threats, and strengthen defenses against adversarial attacks.
By investing in these areas and exploring new avenues for innovation, organi-
zations can yoke the full potential of these technologies in cybersecurity. This will
enable them to build resilient, intelligent, and secure systems that effectively address
the evolving cyber threat landscape. The integration of AI, ML, and Blockchain is
not a standalone solution, but a powerful combination that can significantly enhance
cybersecurity and pave the way for a more secure digital future.

References

1. Anderson R (2018) Security engineering: a guide to building dependable distributed systems.

Wiley
2. Bishop M (2021) Computer security: art and science. Addison-Wesley
3. Schneier B (2015) Data and goliath: the hidden battles to collect your data and control your
world. W. W. Norton & Company
4. Zhang H, Zhu S, Ghosh A (eds) (2021) Handbook of blockchain security: a technical guide to
secure distributed ledgers. Springer
5. Buchanan W, Blyth A (eds) (2020) Cyber security handbook. Springer.
6. Cavoukian A, Castro J (2019) The rise of blockchain technology in securing the internet of
things. Eur Data Prot Law Rev 5(2):226–241
7. Dhillon G, Moores T (2019) Principles of information security. Cengage Learning
8. Ford R, Sorell M, Traynor P (2019) Cybersecurity: protecting critical infrastructures from cyber
attack and cyber warfare. Wiley.
9. Goodrich MT, Tamassia R (2021) Introduction to computer security. Pearson.
10. Janczewski L, Colarik A (2019) Cyber warfare and cyber terrorism. Springer
Introduction to Cybersecurity with AI, ML, and Blockchain 19

11. Kim D (2018) Cybersecurity and blockchain technology. In: The international conference on
information networking (ICOIN).
12. Kouicem DE, Guesmi R (2019) Blockchain technology for secure cyber physical systems: a
survey. Futur Gener Comput Syst 97:512–533
13. Liao CH, Lu YC, Huang CY (2020) Enhancing privacy and security in cloud-based medical
systems using blockchain. Futur Gener Comput Syst 105:368–377
14. Mitnick KD, Simon WL (2017) The art of invisibility: the world’s most famous hacker teaches
you how to be safe in the age of big brother and big data. Little, Brown and Company.
15. Ross SM (2019) Computer security fundamentals. Pearson.
16. Samaniego M, Larrabeiti D, Díaz V (2019) Security in IoT communications based on
blockchain technology. Sensors 19(6):1402
17. Schreiber P, Omerzu M, Kompara M (2020) Blockchain in cybersecurity: a systematic mapping
study. Security and Communication Networks
18. Shariq S, Ahmad I (2021) Machine learning in cybersecurity: a survey. In: Machine learning
for cybersecurity. Springer, pp 1–32
19. Vacca JR (2019) Computer and information security handbook. Morgan Kaufmann
20. Yampolskiy M (2018) Artificial superintelligence: a futuristic approach. CRC Press.
21. Chen M, Hao Y, Ma X (2020) Applications of machine learning in cybersecurity. IEEE Trans
Netw Sci Eng 7(2):998–1010
22. Yadav P, Agarwal A (2020) A Comprehensive survey of deep learning in cybersecurity. J Inf
Secur Appl 50:102419
23. Zhao Y, Ge L (2020) Machine learning-based cybersecurity solutions: a survey. Comput Secur
88:101670
24. Akhtar Z, Khan A, Mehdi S (2020) Applications of artificial intelligence in cybersecurity: a
review. J Cybersecur 6(1):tyaa009.
25. Kambourakis G, Maragoudakis M (eds) (2020) Artificial intelligence applications in cyber
security. Springer.
26. Thomas D (2021) Artificial intelligence in cybersecurity: a guide to detection, response, and
prevention. Apress.
27. Nasseh S, Moallem A (2020) An overview of machine learning applications in cybersecurity. In:
Proceedings of the 10th international conference on web intelligence, Mining and Semantics.
pp 1–6.
28. Qin C, Cheng X (2020) Machine learning methods for cybersecurity. J Cybersecur 6(1):tyy027
29. Gandomi A, Haider M (2019) Beyond the hype: big data concepts, methods, and analytics. Int
J Inf Manage 35(2):137–144
30. Rong C, Huang W (2018) A survey of advances in deep learning for cybersecurity. Inf Fusion
46:35–49
31. Qu Z, Zhang R (2020) Machine learning in cybersecurity: a comprehensive survey. IEEE
Commun Surv & Tutor 22(3):1842–1872
32. Sitnikova E, Tulumenkov A, Ometov A (2019) Machine learning in network security: a
comprehensive survey. IEEE Commun Surv & Tutor 21(3):2702–2733
33. Zhang X, Zhang Z, Wang H (2020) Machine Learning-based cybersecurity systems: a survey.
Futur Gener Comput Syst 107:776–797
34. Sun Y, Liu X, Du X (2019) A survey of machine learning techniques applied to network security.
Secur Commun Netw 2019:1–21
35. Bolon-Canedo V, Sanchez-Maroño N, Alonso-Betanzos A (2018) A review of feature selection
methods in medical applications. Comput Biol Med 2018:279–291
36. Barbosa MA, de Souza RF, da Silva EP (2021) Artificial intelligence and machine learning for
cyber security. Comput Secur 113:102363. https://doi.org/10.1016/j.cose.2021.102363
37. Choudhary KS, Kaushik A (2022) Artificial intelligence in cyber security: a review. ACM
Comput Surv (CSUR) 55(3):1–33. https://doi.org/10.1145/3507490
38. Dash S, Mishra P, Rathore S (2022) Artificial intelligence in cyber security: a comprehensive
survey. J Netw Comput Appl 186:103090. https://doi.org/10.1016/j.jnca.2022.103090
20 K. C. Nosina and T. Swarna Latha

39. Dwivedi A, Jain R (2022) Artificial intelligence-based security solutions for cloud computing.
Comput Secur 114:102469. https://doi.org/10.1016/j.cose.2022.102469
40. Garcia JC, Martinez JJ (2022) Artificial intelligence in cyber security: a systematic review.
Comput Secur 115:102504. https://doi.org/10.1016/j.cose.2022.102504
41. Kaur A, Singh A (2022) Artificial intelligence and machine learning in cyber security: a review
of literature. Secur J 35(1):145–166. https://doi.org/10.1057/s41284-021-00312-0
42. Kumar V, Kaur I (2022) Artificial intelligence and machine learning for cyber security: a
state-of-the-art survey. Security and Privacy 10(1):3–19. https://doi.org/10.1002/sp.2168
43. Lakshmi V, Kumar S (2022) Artificial intelligence for cyber security: a survey. Comput Secur
116:102545. https://doi.org/10.1016/j.cose.2022.102545
44. Mishra S, Singh S, Sharma AK (2022) Artificial intelligence in cyber security: a comprehensive
review. J Inf Secur 13(1):25–50. https://doi.org/10.1016/j.jis.2022.01.001
45. Patel M, Patel K (2022) Artificial intelligence in cyber security: a review. Int J Inf Secur
21(2):407–427. https://doi.org/10.1007/s10207-021-00463-y
46. Ranganathan S, Manjunath GK (2022) Artificial intelligence for cyber security: a survey and
research directions. ACM Comput Surv (CSUR) 55(4):1–41. https://doi.org/10.1145/3507495
47. Saxena A, Jain R (2022) Artificial intelligence in cyber security: a state-of-the-art survey. IEEE
Access 10:78105–78129. https://doi.org/10.1109/ACCESS.2022.3152332
48. Alam M, Ahad MA, Hossain MA (2022) Blockchain-based security solutions: a survey. Comput
Secur 115:102505. https://doi.org/10.1016/j.cose.2022.102505
49. Bartoletti M, Lamberti F, Pompianu L, Satta G (2022) Blockchain security: a survey of current
research. IEEE Commun Surv & Tutor 24(2):1603–1634. https://doi.org/10.1109/COMST.
2021.3059155
50. Chen WY, Chen P, Lin LC (2022) A survey of blockchain security. ACM Comput Surv (CSUR)
55(3):1–33. https://doi.org/10.1145/3507489
51. Ellul J, Karame GO (2022) Blockchain security: a survey of current research and trends. Front
Blockchain 5:68. https://doi.org/10.3389/fbloc.2022.787270
52. Han Z, Wang X, Zhou K, Wang L (2022) Blockchain security: a survey. AIEEE Access
10:6977–6992. https://doi.org/10.1109/ACCESS.2022.3152331
53. Jiang L, Zhang C, Wang X (2022) A survey on blockchain security: current status, challenges
and opportunities. Secur Priv 10(1):20–36. https://doi.org/10.1002/sp.2170
54. Kamble VS, Patil S, Patil S (2022) Blockchain security: a comprehensive review. J Inf Secur
13(1):51–78. https://doi.org/10.1016/j.jis.2022.01.002
55. Khodaei A, Dehghantanha A (2022) A survey on blockchain security: challenges, solutions,
and open problems. Comput Secur 116:102546. https://doi.org/10.1016/j.cose.2022.102546
56. Kumar V, Kaur I (2022) Blockchain security for industrial IoT: a systematic literature review.
IEEE Syst J 16(2):815–832. https://doi.org/10.1109/JSYST.2021.3083220
57. Liu Y, Zhang W, Li R (2022) A survey on blockchain security: current status, challenges and
opportunities. Secur Priv 10(1):37–52. https://doi.org/10.1002/sp.2169
58. Mishra S, Singh R (2022) Blockchain security: a comprehensive review. J Inf Secur 13(1):79–
100. https://doi.org/10.1016/j.jis.2022.01.003
59. Nguyen TD, Chang H (2022) Blockchain security: a survey of current research and trends.
IEEE Access 10:6969–6976. https://doi.org/10.1109/ACCESS.2022.3152330
60. Anoop R, Rathee A (2023) Blockchain and artificial intelligence: the convergence of two
powerful technologies. Int J Inf Secur 22(1):3–19. https://doi.org/10.1007/s10207-022-005
35-3
61. Bhardwaj A, Gupta A (2022) Convergence of artificial intelligence, machine learning, and
blockchain: a survey of emerging security solutions. Comput Secur 116:102547. https://doi.
org/10.1016/j.cose.2022.102547
62. Chang H, Lee J (2022) The convergence of artificial intelligence, machine learning, and
blockchain for cybersecurity. IEEE Access 10:78129–78144. https://doi.org/10.1109/ACC
ESS.2022.3152333
63. Dwivedi A, Jain R (2022) Artificial intelligence and blockchain: a convergence for next-
generation cybersecurity. Comput Secur 115:102503. https://doi.org/10.1016/j.cose.2022.
102503
Introduction to Cybersecurity with AI, ML, and Blockchain 21

64. Gupta R, Kumar S (2022) Artificial intelligence, machine learning, and blockchain: conver-
gence and implications for cybersecurity. IEEE Syst J 16(3):1668–1680. https://doi.org/10.
1109/JSYST.2022.3189369
65. Kumar V, Kaur I (2022) Convergence of artificial intelligence, machine learning, and blockchain
for cybersecurity: a systematic literature review. Inf Sci 565:1314–1336. https://doi.org/10.
1016/j.ins.2022.01.068
66. Mishra S, Saini A, Singh S (2023) Convergence of artificial intelligence, machine learning, and
blockchain for cybersecurity: a systematic review. J Netw Comput Appl 189:103173. https://
doi.org/10.1016/j.jnca.2022.103173
67. Rathee A, Anoop R (2022) The convergence of artificial intelligence, machine learning, and
blockchain: a survey of security challenges and solutions. Comput Secur 115:102502. https://
doi.org/10.1016/j.cose.2022.102502
68. Sharma AK, Mishra S, Singh S (2022) Convergence of artificial intelligence, machine learning,
and blockchain: a systematic review of security and privacy challenges. Secur Priv 10(1):1–24.
https://doi.org/10.1002/sp.2171
69. Singh M, Kaur J (2022) Convergence of artificial intelligence, machine learning, and
blockchain: a survey of security and privacy challenges. Secur J 35(2):299–319. https://doi.
org/10.1057/s41284-022-00313-1
70. Alotaibi B, Alghamdi A, Yaseen M (2022) Security challenges and future directions in the
internet of things. Comput Secur 116:102548. https://doi.org/10.1016/j.cose.2022.102548
71. Anoop R, Rathee A (2023) Challenges and future directions in blockchain security. Int J Inf
Secur 22(2):20–41. https://doi.org/10.1007/s10207-022-00536-2
72. Chang H, Lee J (2022) Challenges and future directions in artificial intelligence-based
cybersecurity. IEEE Access 10:78145–78160. https://doi.org/10.1109/ACCESS.2022.3152334
73. Dwivedi A, Jain R (2023) Challenges and future directions in cybersecurity for cloud
computing. Comput Secur 116:102549. https://doi.org/10.1016/j.cose.2022.102549
74. Gupta R, Kumar S (2022) Challenges and future directions in cybersecurity for internet of
things. IEEE Syst J 16(3):1681–1692. https://doi.org/10.1109/JSYST.2022.3189370
75. Kumar V, Kaur I (2022) Challenges and future directions in cybersecurity for artificial
intelligence. Inf Sci 565:1337–1359. https://doi.org/10.1016/j.ins.2022.01.069
76. Mishra S, Singh S (2023) Challenges and future directions in cybersecurity for blockchain. J
Netw Comput Appl 189:103174. https://doi.org/10.1016/j.jnca.2022.103174
77. Rathee A, Anoop R (2022) Challenges and future directions in cybersecurity for artificial
intelligence, machine learning, and blockchain. Comput Secur 115:102501. https://doi.org/10.
1016/j.cose.2022.102501
78. Sharma AK, Mishra S, Singh S (2022) Challenges and future directions in cybersecurity
for artificial intelligence, machine learning, and blockchain: a systematic review. Secur Priv
10(1):25–50. https://doi.org/10.1002/sp.2170
79. Singh M, Kaur J (2022) Challenges and future directions in cybersecurity for artificial intelli-
gence, machine learning, and blockchain: a survey and research directions. ACM Comput Surv
(CSUR) 55(4):1–41. https://doi.org/10.1145/3507495
Opportunities and Challenges in New
Generation Cyber Security Applications
Using Artificial Intelligence, Machine
Learning and Block Chain

Sarthak Malik, Praveen Kumar Malik, and Arshi Naim

Abstract This chapter explores the dynamic landscape of new-generation cyber

security applications, delving into the integration of Artificial Intelligence (AI),
Machine Learning (ML), and Block chain technologies. The opportunities presented
by this amalgamation are significant, with AI and ML offering unparalleled capa-
bilities in adaptive threat detection, predictive analysis, and automated response
mechanisms. The decentralized and tamper-resistant nature of Block chain adds
transparency and integrity to data security and transaction verification. However,
these opportunities are accompanied by challenges, including the imperative for
continuous innovation to combat evolving cyber threats, ethical considerations in
technology utilization, and the need to strike a balance between security and privacy.
This abstract underscores the transformative potential of AI, ML, and Block chain
in reshaping cyber security paradigms, emphasizing the ongoing necessity for the
cyber security community to address challenges and harness the synergies of these
technologies responsibly and effectively.

Keywords Artificial learning · Challenges · Cyber security · Machine learning ·

Next generation

S. Malik
School of Computer Science Engineering, Lovely Professional University, Phagwara, Punjab,
India
e-mail: [email protected]
P. K. Malik (B)
School of Electronics and Electrical Engineering, Lovely Professional University, Jalandhar,
Punjab, India
e-mail: [email protected]
A. Naim
Business Management, King Khalid University, AlSamer, University Campus, Aseer, Abha, KSA,
Saudi Arabia
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 23
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_2
24 S. Malik et al.

1 Introduction

The landscape of cybersecurity is undergoing a profound transformation with

the integration of cutting-edge technologies such as Artificial Intelligence (AI),
Machine Learning (ML), and Blockchain. This introduction explores the oppor-
tunities and challenges that arise from the synergies of these new-generation cyber-
security applications [1]. The advent of AI and ML introduces adaptive capabili-
ties for threat detection, predictive analysis, and automated responses, promising
a paradigm shift in cybersecurity resilience. Concurrently, the decentralized and
tamper-resistant attributes of Blockchain offer unprecedented transparency and
integrity to data security and transaction verification. However, with these oppor-
tunities come multifaceted challenges, encompassing the imperative for continuous
innovation to counter ever-evolving cyber threats, ethical considerations in the appli-
cation of advanced technologies, and the delicate balance between security imper-
atives and privacy concerns. As we embark on this exploration, it becomes evident
that the convergence of AI, ML, and Blockchain stands at the forefront of reshaping
cybersecurity, requiring a nuanced understanding of the opportunities and challenges
that accompany this transformative journey. In an era where the digital landscape is
evolving at an unprecedented pace, the integration of AI, ML, and Blockchain into
cybersecurity represents a frontier of innovation and adaptation. AI and ML technolo-
gies empower cybersecurity systems to dynamically learn and respond to emerging
threats, offering a proactive defense mechanism that goes beyond traditional reac-
tive approaches. Simultaneously, Blockchain’s decentralized architecture addresses
the vulnerabilities associated with centralized systems, ensuring data integrity and
authentication in an increasingly interconnected world. As organizations strive to
harness the advantages of these technologies, they are confronted with challenges
ranging from the need for skilled expertise in managing sophisticated systems to
ethical dilemmas surrounding privacy and security trade-offs. This exploration delves
into the intricate interplay between opportunities and challenges, shedding light on
the transformative potential of AI, ML, and Blockchain in shaping the future of cyber-
security. Figure 1 illustrates various resources that can be employed for cyber-security
within the AI-powered metaverse.
It is possible to use block chain to move data between parties in a secure manner,
which can help reduce the likelihood of data theft. In addition, these technologies
can be utilised to automate a large number of the processes and procedures that are
presently controlled manually [3–5]. As a result, the workload of security teams can
be lightened, and resources can be reallocated to other projects. Artificial intelligence
(AI) and machine learning (ML) systems are only as successful as the data they are
trained on, which means that they can be deceived by malicious actors. However,
the adoption of these technologies also brings with it a variety of obstacles. It is
also possible for hostile actors to get access to the ledger and read or edit data
using block chain technology; hence the system is not entirely safe. In addition,
these technologies are still in their infancy, so their efficacy is still being assessed.
Because of this, businesses need to be ready to adjust as the technology advances
Opportunities and Challenges in New Generation Cyber Security … 25

Fig. 1 Assets which can be used in cyber-security in the AI-based metaverse [2]

[6, 7]. The implementation of next-generation applications for cyber security, such
as those that make use of artificial intelligence, machine learning, and block chain,
presents organisations with opportunities as well as obstacles which can also be seen
in Table 1.
It is possible to use these technologies in order to increase cyber security; but,
in order to ensure that they are effective; they need to be properly installed and
monitored. It is important for organisations to take the time to carefully consider
the potential drawbacks and advantages of using these technologies before deciding
whether or not they should use them [8, 9].

Table 1 Relation between

Cyber security apps AI ML Block chain
different entities of cyber
security applications using Antivirus software ✓ ✓
artificial intelligence, Firewalls ✓ ✓
machine learning and block
chain Intrusion detection ✓ ✓
Data encryption ✓ ✓
Password management ✓ ✓
DDoS protection ✓ ✓
Vulnerability scan ✓ ✓
Security audits ✓ ✓
26 S. Malik et al.

2 Understanding the Basics of Cyber Security

Protecting computer networks, computer systems, computer programmes, and data

from being attacked digitally is the practise known as cyber security. Understanding
the basics of cybersecurity is paramount in navigating the increasingly complex
digital landscape. At its core, cybersecurity is a discipline dedicated to safeguarding
digital systems, networks, and data from unauthorized access, attacks, and damage.
It involves the implementation of robust measures to ensure confidentiality, integrity,
and availability of information. Key components of cybersecurity include firewalls,
antivirus software, encryption, and secure authentication methods. A fundamental
aspect is user awareness, emphasizing the importance of recognizing and mitigating
potential threats such as phishing, malware, and social engineering. As technology
advances, so do cyber threats, making continuous learning and adaptation integral
to effective cybersecurity practices. Ultimately, a strong foundation in the basics of
cybersecurity is crucial for individuals and organizations alike, forming the front-
line defense against an ever-evolving array of digital risks. In addition to the tech-
nical aspects, a comprehensive understanding of cybersecurity involves recognizing
the human element as a crucial factor in maintaining a secure environment. This
includes promoting a culture of cybersecurity awareness and education within orga-
nizations, ensuring that employees are informed about best practices and potential
risks. Regular updates and patches for software and systems play a pivotal role in
closing vulnerabilities that cyber attackers may exploit. Additionally, cybersecurity
measures extend beyond the confines of personal computers to encompass mobile
devices, IoT (Internet of Things) devices, and cloud-based services. Collaborative
efforts between individuals, businesses, and governments are essential for creating a
resilient cybersecurity ecosystem. By grasping the basics of cybersecurity, individ-
uals can contribute to a collective defense against cyber threats and play an active role
in fostering a secure digital environment for themselves and the broader community
[10]. It is a process, not a product, and in order for it to be effective, it requires a
combination of different technologies, practises, and other procedures. Organisations
are able to guarantee the safety of their data and systems if they have a fundamental
awareness of, and adherence to, cyber security best practises. Following are some of
the important aspects of cyber securities (Fig. 2).

2.1 Understand the Importance of Cyber Security

Protecting computer networks, computer systems, and computer programmes from

being attacked digitally is the practise known as cyber security. The goal of these
attacks is to acquire access to sensitive information, disrupt or destroy that informa-
tion, extort money from an organisation or individual, or cause harm in some other
way. As both the number and the level of sophistication of online attacks continue to
rise, it is essential for businesses to incorporate cyber security into their overall risk
Opportunities and Challenges in New Generation Cyber Security … 27

Fig. 2 Importance and understanding of cyber-security

management plan. It is imperative for businesses to take a preventative approach to

controlling their cyber security risk by putting in place the necessary safeguards to
secure their data, systems, and networks. This involves doing things like encrypting
data and routinely patching and updating software, as well as implementing robust
authentication techniques. In addition, companies should devise acceptable use stan-
dards for their computer networks and systems, make sure staffs receives frequent
training on the latest cyber security best practises, and make sure the policies are
enforced [11].
In addition, companies and other organisations must to have a thorough inci-
dent response plan in place, which will assist them in swiftly reacting to any cyber
security events that may arise. This includes the establishment of distinct roles and
duties for responding to incidents, identifying and analysing risks, and designing and
putting into action strategies for remediation. In the modern digital world, when cyber
dangers are becoming more frequent, having adequate cyber protection is absolutely
necessary. Organisations are obligated to protect their networks, systems, and data
by taking preventative measures and developing detailed incident response plans that
can be implemented in the event that an incident occurs [12]. By doing so, organisa-
tions will be able to reduce their exposure to risk, safeguard their assets, and ensure
the continuation of their operations.
28 S. Malik et al.

2.2 Know the Different Types of Cyber Threats

Understanding the various types of cyber threats is imperative in building a robust

defense against potential security breaches. Cyber threats encompass a broad spec-
trum of malicious activities designed to compromise digital systems and data.
Common types include malware, which includes viruses, worms, and ransomware,
designed to infiltrate and damage systems. Phishing attacks involve deceptive tactics
to trick individuals into revealing sensitive information, often through fake emails or
websites. Distributed Denial of Service (DDoS) attacks aim to overwhelm a target’s
online services, rendering them inaccessible. Man-in-the-Middle (MitM) attacks
involve intercepting communication between two parties, posing a significant risk to
data integrity. Additionally, SQL injection, cross-site scripting, and zero-day exploits
target vulnerabilities in software and websites. A comprehensive awareness of these
cyber threats enables individuals and organizations to implement tailored security
measures and proactive strategies to mitigate risks effectively [13].

2.2.1 Malware

It is a type of harmful software that is meant to cause damage to or collect infor-

mation from a computer system. Malware also goes by the name “spyware.” Email
attachments, software download, and malicious websites are all potential vectors for
the propagation of malware. Malware can take many different forms, some of which
are viruses, worms, Trojan horses, rootkits, ransom ware, and spyware.

2.2.2 Phishing

Phishing is a type of cyber-attack in which an attacker tries to deceive a victim

into disclosing sensitive information such as passwords, credit card details, or
bank account numbers. These assaults are frequently carried out through the use
of malicious websites or emails.

2.2.3 Social Engineering

Creating a fake identity or using social media to win someone’s trust in order to
obtain sensitive information is a common tactic utilised in these types of attacks.
Opportunities and Challenges in New Generation Cyber Security … 29

2.2.4 Attacks Using DDoS

DDoS is an abbreviation that stands for distributed denial of service. This kind
of assault inundates a server with requests, which eventually leads to the server
becoming overloaded and crashing.

2.3 Understand the Roles and Responsibilities

of a Cyber-Security Professional

Comprehending the roles and responsibilities of a cybersecurity professional is

crucial in addressing the escalating challenges of the digital landscape. These
professionals are tasked with safeguarding sensitive information, networks, and
systems from a diverse array of cyber threats. Their responsibilities often include
implementing security measures such as firewalls, encryption, and intrusion detec-
tion systems to fortify digital infrastructure. Regularly assessing vulnerabilities,
conducting penetration testing, and staying abreast of the latest cyber threats are
integral components of their duties. Cybersecurity professionals also play a pivotal
role in educating end-users and colleagues about best practices and potential risks.
Moreover, they are at the forefront of incident response, swiftly identifying and
mitigating security breaches when they occur. The evolving nature of cyber threats
demands continuous learning, adaptability, and a commitment to staying ahead of
emerging risks. In essence, the role of a cybersecurity professional is multifaceted,
requiring a holistic approach to protect digital assets and ensure the integrity of infor-
mation in an ever-changing technological landscape. They have to be able to strike
a balance between the many privacy laws and regulations and the security require-
ments. A cyber security professional’s responsibilities may extend to advising staff
members and end users on the most effective ways to safeguard their organization’s
information systems and data in some businesses. This involves the provision of
training on subjects such as the management of passwords and the secure use of the
internet. In the modern digital world, the function of a cyber-security professional
is an extremely important one. Because of the vast amounts of data and informa-
tion that are kept in digital form, organisations need to make sure that they have the
appropriate expertise on staff in order to keep their networks and systems secure
[14].

2.4 Understand the Different Cyber Security Tools Available

Software applications that assist safeguard networks, systems, and data against cyber-
attacks, cyber criminals, and other types of cyber dangers are known as cyber security
tools. These instruments can perform a wide range of security functions, such as
30 S. Malik et al.

malware identification, vulnerability scanning, patch update administration, network

monitoring, intrusion prevention, and data protection. A firewall is one of the most
important parts of a strong cyber security system. By restricting the types of data
traffic that are permitted to pass through the firewall, firewalls serve as a barrier
between an internal network and the outside world. Firewalls can be configured to
block malicious traffic, including that generated by viruses, malware, and phishing
scams. Antivirus software can assist in the detection, prevention, and removal of
malicious software, also referred to as malware, from a system.
It can also detect unusual activities, such as attempts to access restricted portions
of the system, which may indicate that the user is trying to gain unauthorised access.
Intrusion detection and prevention systems, often known as IDS/IPS, are able to
identify any attempts made to obtain unauthorised access to a system or network.
IDS and IPS systems have the ability to identify potentially malicious behaviour,
such as port scans and attempts to exploit vulnerabilities, and then take preventative
measures to thwart an attack [15].
Spotting vulnerabilities in a network or computer system is the primary purpose
of vulnerability scanners. These scanners are able to identify software flaws such as
vulnerabilities, misconfigurations, and missing patches, as well as provide recom-
mendations on how to fix the problems they find. Tools for encrypting data are used
to protect the information that is saved on computers and servers. Encryption algo-
rithms transform data into a form that can only be read and understood by those
who have been granted permission to do so. Users are assisted in the creation of,
storage of, and administration of secure passwords via tools for password manage-
ment. These solutions have the ability to generate robust passwords, store them in a
secure location, and control access permissions. These are only a few of the many
cyber security products that are currently accessible. Because each tool possesses a
distinct set of features and capabilities, it is essential to select the tool that is most
suited to meet the requirements for security imposed by your organisation [16].

2.5 Understand Data Privacy and Protection Regulations

Understanding data privacy and protection regulations is paramount in the contem-

porary digital landscape, where the handling of personal information is ubiquitous.
These regulations, such as the General Data Protection Regulation (GDPR) in Europe
or the California Consumer Privacy Act (CCPA) in the United States, are designed
to safeguard individuals’ privacy rights and impose responsibilities on organizations
that collect, process, and store personal data. The regulations typically outline strict
guidelines on consent mechanisms, data breach notifications, and the rights of indi-
viduals to access and control their data. Compliance with these regulations is not
only a legal requirement but also a fundamental ethical obligation. It involves imple-
menting robust data protection measures, including encryption, access controls, and
secure storage practices. Understanding and adhering to data privacy regulations not
Opportunities and Challenges in New Generation Cyber Security … 31

only mitigate legal risks for organizations but also foster trust with users by demon-
strating a commitment to respecting their privacy and safeguarding their sensitive
information. Regulations designed to protect individuals’ privacy, as well as the data
they voluntarily contribute to businesses, are critically vital. In order to guarantee that
they are processing data in a responsible and secure manner, businesses are required
to comply with certain requirements [17, 18].

2.6 Recognize the Signs of a Cyber-attack

Attacks in cyberspace can take many different forms, and it can be challenging to
spot the tell-tale symptoms of an impending attack. On the other hand, there are
specific markers that can assist in determining whether or not an attack is taking
place.
1. If you suddenly notice that your network is performing more slowly than usual,
this could be an indication that an attack is taking place. Attackers in the digital
realm frequently deploy malicious malware to bog down networks and make
them run more slowly.
2. Unusual emails or website visits: You should be on the lookout for symptoms of
a cyber-attack if you start receiving emails from senders you aren’t acquainted
with or if you start visiting websites that you wouldn’t typically go to.
3. Account access without authorization: If you find that someone has accessed
your account without your authorization, this may be a sign of a cyberattack.
4. Unexpected file modifications: If you discover that files have been altered without
your knowledge, this could be a sign that your system is under attack.
5. Unusual activity in your log files: If you observe unusual activity in your log
files, it may indicate that you are the victim of a cyberattack.
If you can identify the telltale signs of a cyberattack and take appropriate action,
you will be able to protect your data and systems from malicious activity. To ensure the
security of your computer systems, you must remain vigilant and take the necessary
precautions [19, 20].

2.7 Understand the Basics of Cryptography

The most prevalent form of cryptography is called symmetric cryptography, and it

entails using a single cryptographic key to both encrypt and decrypt data. Symmetric
cryptography is also the most secure form of cryptography. In asymmetric cryptog-
raphy, each step—encryption, decryption, and verification—requires the use of a
distinct key. In addition, cryptography encompasses the utilisation of digital signa-
tures, which are used to validate the genuineness of data or messages. The authen-
ticity of a document can be validated with the help of something called a digital
32 S. Malik et al.

signature, which is a form of cryptographic hash. A private key, which is used to

sign the document, and a public key, which is used to verify the signature, are both
incorporated into its generation. The private key is used to sign the document. In
today’s digital age, one of the most important tools for protecting one’s data and
one’s communications is cryptography. Its purpose is to protect the privacy of data
and communications while also guaranteeing their integrity and accessibility.

2.8 Learn How to Create an Incident Response Plan

Every strong cyber-security programme needs to have a solid incident response plan
that can handle any situation that may arise. It specifies the procedures that need
to be taken in order to respond to a security incident in a manner that guarantees
the response will be effective, uniform, and delivered in a timely manner. The first
thing that has to be done when building a plan for responding to incidents is to
determine the different kinds of occurrences that could arise. This should cover
both harmful and non-malicious incidents, such as data breaches, system failures,
and unauthorised access, among other things. Following the identification of these
incidents, the response plan should detail the measures to take when responding
to each distinct type of incident. Additionally, the plan ought to establish who is
accountable for providing a response to each incident. This should include both
internal staff and external resources, such as law enforcement and cyber-security
experts, among other types of professionals. In addition to this, the plan should
detail the roles and duties of each member of the team, in addition to providing a
timeframe for responding to each incident.

3 The Benefits of Artificial Intelligence in Cyber Security

Artificial Intelligence (AI) has emerged as a crucial ally in the realm of cybersecurity,
offering a myriad of benefits that significantly enhance our ability to safeguard digital
assets. One of the key advantages lies in AI’s capacity to swiftly analyze vast amounts
of data in real-time, enabling the rapid detection of anomalies and potential security
threats. Machine learning algorithms, a subset of AI, excel at recognizing patterns and
adapting to evolving cyber threats, thereby fortifying defense mechanisms against
sophisticated attacks. Additionally, AI facilitates predictive analysis, allowing cyber-
security systems to anticipate potential vulnerabilities and proactively address them
before they can be exploited. Moreover, AI-powered automation streamlines routine
security tasks, reducing the burden on human operators and minimizing the risk of
human error. This not only enhances the overall efficiency of cybersecurity protocols
but also ensures a quicker response to emerging threats. Ultimately, the integration
of AI in cybersecurity not only fortifies our digital defenses but also empowers
Opportunities and Challenges in New Generation Cyber Security … 33

Fig. 3 Artificial intelligence in cyber security, benefits and advantages

organizations to stay one step ahead in an increasingly complex and dynamic threat
landscape.
Following are some of the benefits of AI in cyber security as shown in Fig. 3 also:
• Enhanced Detection: Compared to more conventional approaches, artificial
intelligence can detect potential dangers both more quickly and more correctly.
• Automated Response: Artificial intelligence has the ability to automatically
respond to cyber-attacks, which can significantly cut down on the amount of
time and effort spent on manual responses.
• Enhanced Cyber security: AI may be used to detect and prevent unwanted
activity, such as malicious network invasions, hacking attempts, and phishing.
This can be a significant benefit to organisations concerned about their online
safety.
• Improved Data Analysis: AI may be used to analyse vast volumes of data in
order to spot patterns and abnormalities that may point to a possible breach in
security. This can be done in order to reduce the likelihood of a breach occurring.
• Enhanced Productivity: Artificial Intelligence has the potential to cut down on
the amount of time and resources needed to keep networks under surveillance and
protect them.
34 S. Malik et al.

4 Machine Learning and Cyber Security

Machine Learning (ML) has become a linchpin in the field of cybersecurity, revolu-
tionizing the way we detect, prevent, and respond to evolving cyber threats. At its
core, machine learning equips cybersecurity systems with the ability to learn from
data patterns and adapt in real-time, offering a dynamic and proactive defense mech-
anism. ML algorithms excel in identifying anomalies and unusual patterns within
vast datasets, enabling the swift detection of potential security breaches that may go
unnoticed by traditional methods. By continuously analyzing and learning from new
data, machine learning models can evolve to recognize emerging threats, providing
a robust defense against sophisticated and constantly evolving cyber attacks. The
predictive capabilities of ML contribute significantly to cybersecurity by allowing for
the identification of vulnerabilities before they are exploited. This proactive approach
enhances overall resilience and responsiveness, as security measures can be adjusted
in anticipation of potential threats. In essence, the integration of machine learning
into cybersecurity not only fortifies digital defenses but also represents a critical
paradigm shift in adapting to the ever-changing landscape of cyber threats.
It is possible for businesses to increase the efficiency of their security operations
and cut down on the amount of time and resources needed to detect and respond to
cyber threats by making use of the power of machine learning (ML). Following are
the main connection between machine learning and cyber security as shown in Fig. 4
also:
• Identifying malicious activity can be accomplished with the use of machine
learning by analysing user behaviour, system logs, and network traffic for
recurring trends.

Fig. 4 Cyber security and machine learning connections

Opportunities and Challenges in New Generation Cyber Security … 35

• Classifying malicious files and identifying malicious code patterns can both be
accomplished through the application of machine learning.
• Vulnerabilities in software, networks, and other systems can be discovered with
the help of machine learning.
• Zero-day vulnerabilities can be identified and protected against with the use of
machine learning.
• Monitoring and analysing huge datasets for suspicious behaviour can be accom-
plished with the help of machine learning.
• Detection and warning of potential harmful risks can be achieved through the
application of machine learning.
• Identifying malicious actors and their strategies can be accomplished through the
use of machine learning.
• Dangerous websites and networks can be identified and blocked using machine
learning, which can also be used to identify dangerous websites.
• Phishing attempts and malicious email campaigns can be uncovered with the use
of machine learning techniques.
• Both detecting and preventing data breaches can be accomplished with the help
of machine learning.

5 The Potential of Block Chain in Cyber Security

The world is getting more digitalized at a rapid pace, and as a result, the demand
for a safe and secure online environment is higher than it has ever been. Blockchain
technology holds immense potential in reshaping the landscape of cybersecurity,
offering innovative solutions to address longstanding challenges in securing digital
transactions and data. One of the key strengths of blockchain lies in its decentralized
and tamper-resistant nature. By utilizing a distributed ledger, blockchain ensures that
once data is recorded, it becomes virtually immutable, reducing the risk of unautho-
rized alterations or malicious tampering. This characteristic enhances the integrity
and transparency of digital records, providing a robust foundation for secure and
verifiable transactions. Additionally, blockchain’s consensus mechanisms and smart
contracts contribute to the automation of security processes, reducing the reliance
on centralized authorities. This decentralized approach not only mitigates the risk
of a single point of failure but also enhances the resilience of cyber security infras-
tructure against potential attacks. As block chain continues to mature, its potential
applications in securing identity management, protecting sensitive information, and
enabling secure peer-to-peer transactions underscore its pivotal role in fortifying the
foundations of cyber security for the digital age.
36 S. Malik et al.

6 Conclusion

In conclusion, the fusion of Artificial Intelligence (AI), Machine Learning (ML), and
Block chain in the realm of cyber security presents a landscape rich with opportunities
and challenges. The integration of AI and ML brings forth the promise of unparalleled
threat detection capabilities, enabling systems to adapt and evolve in real-time. The
predictive analysis and automated response mechanisms enhance overall cyber secu-
rity resilience. Meanwhile, Block chain’s decentralized and tamper-resistant nature
adds a layer of transparency and integrity, revolutionizing data security and transac-
tion verification. However, with these opportunities come challenges, such as the need
for continuous innovation to keep pace with evolving cyber threats and the poten-
tial ethical considerations surrounding the use of advanced technologies. Striking
the right balance between security and privacy, ensuring regulatory compliance, and
addressing the dynamic nature of cyber threats are ongoing challenges that must
be navigated. As we navigate this complex terrain, it is evident that the synergies
between AI, ML, and Block chain hold the key to shaping the future of cyber security,
requiring a concerted effort from the cyber security community to unlock their full
potential while mitigating associated risks.

References

1. Lee S, Kim S (2022) Blockchain as a cyber defense: opportunities, applications, and challenges.
IEEE Access 10:2602–2618. https://doi.org/10.1109/ACCESS.2021.3136328
2. Pooyandeh M, Han K-J, Sohn I (2022) Cyber security in the AI-based metaverse: a survey.
Appl Sci 12:12993. https://doi.org/10.3390/app122412993
3. Zhang H, Li P, Du Z, Dou W (2020) Risk entropy modeling of surveillance camera for public
security application. IEEE Access 8:45343–45355. https://doi.org/10.1109/ACCESS.2020.297
8247
4. Akram SV, Alshamrani SS, Singh R, Rashid M, Gehlot A, AlGhamdi AS, Prashar D (2021)
Blockchain enabled automatic reward system in solid waste management. Secur Commun
Netw
5. Sun N et al (2022) Defining security requirements with the common criteria: applications,
adoptions, and challenges. IEEE Access 10:44756–44777. https://doi.org/10.1109/ACCESS.
2022.3168716
6. Zhang Z, Hamadi HA, Damiani E, Yeun CY, Taher F (2022) Explainable artificial intelli-
gence applications in cyber security: state-of-the-art in research. IEEE Access 10:93104–93139.
https://doi.org/10.1109/ACCESS.2022.3204051
7. Bhatt H, Bahuguna R, Singh R, Gehlot A, Akram SV, Priyadarshi N, Twala B (2022) Artificial
intelligence and robotics led technological tremors: a seismic shift towards digitizing the legal
ecosystem. https://doi.org/10.3390/app122211687
8. Choudhury S, Singh R, Gehlot A, Kuchhal P, Akram SV, Priyadarshi N, Khan B (2022) Agri-
culture field automation and digitization using internet of things and machine learning. https://
doi.org/10.1155/2022/9042382
9. Elsisi M, Tran M-Q, Mahmoud K, Mansour D-EA, Lehtonen M, Darwish MMF (2021) Towards
secured online monitoring for digitalized GIS against cyber-attacks based on IoT and machine
learning. IEEE Access 9:78415–78427. https://doi.org/10.1109/ACCESS.2021.3083499
Opportunities and Challenges in New Generation Cyber Security … 37

10. Shaukat K, Luo S, Varadharajan V, Hameed IA, Xu M (2020) A survey on machine learning
techniques for cyber security in the last decade. IEEE Access 8:222310–222354. https://doi.
org/10.1109/ACCESS.2020.3041951
11. Dua S, Kumar SS, Albagory Y, Ramalingam R, Dumka A, Singh R, Rashid M, Gehlot A,
Alshamrani SS, Alghamdi AS (2022) Developing a speech recognition system for recognizing
tonal speech signals using a convolutional neural network. Appl Sci (Switzerland) 12(12).
https://doi.org/10.3390/app12126223
12. Gehlot A, Malik PK, Singh R, Akram SV, Alsuwian T (2022) Dairy 4.0: intelligent communi-
cation ecosystem for the cattle animal welfare with blockchain and IoT enabled technologies.
Appl Sci (Switzerland) 12(14). https://doi.org/10.3390/app12147316
13. Ferrag MA, Friha O, Hamouda D, Maglaras L, Janicke H (2022) Edge-IIoTset: a new compre-
hensive realistic cyber security dataset of IoT and IIoT applications for centralized and federated
learning. IEEE Access 10:40281–40306. https://doi.org/10.1109/ACCESS.2022.3165809
14. Kumar SC, Thakur AK, Aseer JR, Natarajan SK, Singh R, Priyadarshi N, Twala B (2022) An
experimental analysis and ANN based parameter optimization of the influence of microalgae
spirulina blends on CI engine attributes. Energies 15(17). https://doi.org/10.3390/en15176158
15. Ferrag MA, Friha O, Maglaras L, Janicke H, Shu L (2021) Federated deep learning for cyber
security in the internet of things: concepts, applications, and experimental analysis. IEEE
Access 9:138509–138542. https://doi.org/10.1109/ACCESS.2021.3118642
16. Madan P, Singh V, Chaudhari V, Albagory Y, Dumka A, Singh R, Gehlot A, Rashid M, Alsham-
rani SS, Alghamdi AS (2022) An optimization-based diabetes prediction model using CNN
and Bi-directional LSTM in real-time environment. Appl Sci (Switzerland) 12(8). https://doi.
org/10.3390/app12083989
17. Liang F, Hatcher WG, Liao W, Gao W, Yu W (2019) Machine learning for security and the
internet of things: the good, the bad, and the ugly. IEEE Access 7:158126–158147. https://doi.
org/10.1109/ACCESS.2019.2948912
18. Malik P, Gehlot A, Singh R, Gupta LR, Thakur AK (2022) A review on ANN based model
for solar radiation and wind speed prediction with real-time data. Arch Comput Methods Eng.
https://doi.org/10.1007/s11831-021-09687-3
19. Hussain F, Hussain R, Hassan SA, Hossain E (2020) Machine learning in IoT security: current
solutions and future challenges. IEEE Commun Surv & Tutor 22(3):1686–1721. thirdquarter
2020. https://doi.org/10.1109/COMST.2020.2986444
20. Verma P, Dumka A, Singh R, Ashok A, Gehlot A, Malik PK, Gaba GS, Hedabou M (2021)
A novel intrusion detection approach using machine learning ensemble for iot environments.
Appl Sci (Switzerland) 11(21). https://doi.org/10.3390/app112110268
AI and Blockchain for Secure Data
Analytics

Shweta Mayor Sabharwal, Sonia Chhabra, and Manpreet Kaur Aiden

Abstract In the age of big data and sophisticated analytics, organizations experi-
ence major barriers to ensuring the privacy, security, and dependability associated
with their data-driven projects. This chapter examines how integrated blockchain and
machine learning might effectively handle these issues and improve the reliability of
data analytics procedures. In addition to providing users with tokenized benefits, the
artificial intelligence and blockchain integration offer distinctive characteristics to
protect the confidentiality of data, assure data integrity, and allow the safe exchange
of information and partnership. This chapter emphasizes the crucial relevance of
the security and confidentiality of data. It examines how blockchain’s decentralized
and irreversible characteristics can offer a solid platform for safeguarding private
information, preventing unauthorized access, and reducing the danger of security
breaches. In order to demonstrate how the privacy of information can be protected
while utilizing intelligence that is distributed, the idea of federated learning, where
artificial intelligence models undergo training on decentralized data sources, is inves-
tigated in regards to the blockchain. This chapter also discusses how the safe sharing
of data and teamwork could be enabled by blockchain technology and blockchain-
based smart contract systems, and access restrictions allow businesses to specify
data usage commitments, guaranteeing that data is distributed and retrieved following
established guidelines and privileges. Lastly, this chapter discusses the technological
issues, like scalability, interoperability, and processing cost, that emerge when inte-
grating AI and blockchain. It additionally emphasizes how crucial it is to take legal
and regulatory considerations into account, especially when it comes to safeguarding
information and adhering to standards, to achieve a strong and ethical execution. In
general, this chapter provides a thorough review of the connections among blockchain

S. M. Sabharwal (B)
Galgotia’s University, Greater Noida, India
e-mail: [email protected]
S. Chhabra
Sharda University, Greater Noida, India
M. K. Aiden
Maharishi University, Noida, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 39
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_3
40 S. M. Sabharwal et al.

and AI for reliable data analytics. This integration helps in increasing the confiden-
tiality of data, integrity, and partnership while highlighting the necessity of careful
planning and execution to reap its full rewards.

Keywords Artificial Intelligence · Blockchain · Security · Data analytics ·

Federated learning · Privacy

1 Introduction

Organizations confront considerable challenges in maintaining the confidentiality,

safety, and reliability connected to their initiatives based on data in the era of big data
and smart analytics. The significance of protecting private information is increasing
as a result of a sharp rise in the amount of data and developments in analytics
and machine learning [1]. These issues are frequently not adequately addressed
by conventional centralized storage and retrieval methods, which leaves potential
for data breaches, unauthorized use, and damaged integrity. The combination of
blockchain with artificial intelligence appears as a potent approach to increase the
accuracy of the data analytics process as a response to these problems [2, 3].
Organizations have become increasingly reliant on decisions based on data, which
has forced them to gather, store, and analyse huge volumes of data. While this
development opens up a lot of doors for new ideas and innovations, it also raises
serious privacy and security issues. Alarm bells have been rung by prominent data
theft and misuse instances, forcing organizations to review their information security
plans [4].
Due to its decentralized and irreversible characteristics, blockchain technology—
the foundation of cryptocurrency like Bitcoin—has attracted a lot of attention. Every
block in a blockchain, which forms a series of data that is linked, includes a secure
hash of the block before it, making the record of the transaction or information
unbreakable by design [5, 6]. Data security and reliability are solidly supported by
this innovative design, which makes sure that once information is stored on the
blockchain, information cannot be accessed or changed.
In contrast, algorithms for machine learning and AI have proven to have extraor-
dinary talents in gleaning worthwhile insights from huge datasets. Privacy issues
arise, especially when confidential user data is involved because these algorithms
frequently need access to large volumes of data for proper training [7].
A potential answer to these problems is the combination of blockchain technology
and artificial intelligence. Blockchain’s decentralized structure improves privacy and
security of information, whereas AI expands the potential of data analytics [8, 9].
A more dependable and inclusive data analytics environment is facilitated by this
synergy, which enables reliable information sharing, teamwork, and the creation of
tokenized incentive.
With particular objectives in consideration, this chapter intends to provide a thor-
ough analysis of the use of artificial intelligence and blockchain in reliable analytics.
AI and Blockchain for Secure Data Analytics 41

It explores the difficulties businesses encounter when it comes to maintaining the

accuracy, protection, and confidentiality of their data. The chapter discusses how
blockchain technology relates to data analytics by examining its core ideas and
characteristics [10]. To improve the security and confidentiality of information, the
advantages and implications of combining machine learning and artificial intelli-
gence with blockchain are examined. Additionally, the idea of federated learning is
examined to see if it may protect the confidentiality of data in widespread training
for artificial intelligence environments. The chapter explores how the blockchain
relies on smart contract platforms to enable safe data exchange and teamwork while
assuring compliance with privacy contracts [11, 12].
It also investigates the use of tokenized rewards and the use of tokens in rewarding
data producers, intelligence designers, and validators. The chapter emphasizes the
significance of constitutional and regulatory factors, particularly with regard to data
protection and social execution, while discussing technical difficulties including
flexibility, connectivity, and data processing costs [13, 14]. This chapter aims to
provide an in-depth knowledge of how blockchain technology and artificial intelli-
gence collaboration might revolutionize data analytics, ensuring increased security,
privacy, and reliability in the age of data. It does this by offering suggestions for
potential futures and new developments.

2 The Era of Big Data

Authorities, businesses, and advertisers all have access to a growing amount of data
about areas of our daily lives that, in earlier times, we would have assumed to be
relatively confidential. This is the Big Data age. While the processing capacity needed
to manage such information is always growing, technology that extracts, gathers,
keeps, and analyses data is getting less expensive and quicker. The ‘datafication’
of the community, which impacts all spheres of life, has been made feasible by
electronic devices [15, 16]. It is undeniable that data is becoming more and more
important for business and the community, and this trend will continue.
So what does the term “Big Data” indicate? Despite being often used, the phrase
has no established meaning. It is typically related to complicated and enormous
datasets, whereby specialized techniques and instruments are used to carry out oper-
ations to extract valuable data to make better decisions. Big Data isn’t merely about
the amount of data that is available, though; it also includes fresh approaches to data
analysis and the creation of novel information [17]. The phrase is frequently used in
the media to describe the growing abundance of data, the scale of information sets,
the expansion of digital data, as well as emerging or different data sources.
42 S. M. Sabharwal et al.

2.1 Features of Big Data

Big Data includes five critical features considering a more specialized technological
standpoint.

2.1.1 Volume

The amount of data produced and preserved, especially in terms of its size. The worth
and possibility of understanding the data are determined by its amount [18]. Big Data
requires an enormous amount to be able to exist.

2.1.2 Variety

The variety and type of information, in addition to how it was organized. Big Data
can incorporate text, images, audio, and video clips, and data synthesis can fill in the
gaps. It can also be organized, semi-organized, or unorganized. Data can be gathered
from a wide range of sources, and its relevance changes based on the type of analysis
being done, from online communities to internal devices to cell phone navigational
systems [19]. The multiple levels and forms of big data might also vary.

2.1.3 Velocity

The duration required to produce and deal with data. In a corporate setting, rapid
transmission can undoubtedly provide an edge over others, thus data must flow fast
and as close to real-time as feasible [20].

2.1.4 Veracity

Data accuracy and dependability; it’s crucial to have methods for spotting and fixing
any fake, erroneous, or inaccurate information.

2.1.5 Value

The evaluation of trustworthy data enhances knowledge between and within

specialties and contexts. The creation of helpful data results in wealth.
AI and Blockchain for Secure Data Analytics 43

2.2 Ethical Challenges of Big Data

Big Data is becoming more widely acknowledged as a catalyst for change in the
private and public sectors today. With advantages spanning from financial to medical,
meteorological to genome research, medical or ecological study to analytics and
commerce, databases are enabling broad changes in society that are progressively
becoming an integral component of our daily lives [21].
The techniques of manufacturing, utilization, and general living will be revolu-
tionized by data. Its benefits will penetrate every aspect of our living and include
more knowledge regarding energy consumption, greater product, substances, and
agricultural accountability, as well as better medical care and wellness. Data serves
as the building block for novel goods and services, fostering improved output and
resource effectiveness throughout every industry. Data is the essence of commer-
cial advancement. It enables more personalized products, improved legislation, and
improved governmental services [22].
Due to the quick transition of goods and services through recognizing patterns
and insight creation to more complex forecasting approaches and, consequently,
enhanced decision-making, the accessibility of data is crucial when developing
machine learning systems. To address social, environmental, and climate concerns
and promote happier, wealthier, as well as more equitable societies, additional infor-
mation must be made accessible, and data utilization must be improved. To accom-
plish the goals of the European Green Deal, for instance, it will result in improved
legislation [23, 24].
Making choices, client interaction, demand-side prediction, brand and market-
place growth, and operational effectiveness are just a few of the areas where big data
may be used to great effect. Automotive dominates all other industries in keeping
records, according to a McKinsey & Company analysis, rendering Big Data an essen-
tial part of the latest industrial revolution, sometimes known as “Industry 4.0.” With
improved handling of supply chains and improved mitigation of risks mechanisms,
led by smart choices, this renaissance has the potential to increase efficiency. Addi-
tionally, Industry 4.0 emphasizes developing sophisticated goods that can collect and
send massive volumes of data across the course of their manufacture and consump-
tion lifetimes. In order to identify consumer needs and influence next items, such
information must be acquired and evaluated in real-time. The enormous adoption of
transformational practices, like the usage of computerized doubles in production, is
also anticipated to be fuelled by data [25, 26].
As previously indicated, Big Data also adds value in a wide range of other indus-
tries, such as government functions, learning, and the medical field. The implemen-
tation of democratic government and openness principles is anticipated to improve
many facets of citizens’ life. Along with perhaps more obvious applications, like
better preventive medicine in the medical field or monitoring their own in the educa-
tional sector, this may end up resulting in the creation of increasingly fair and
participatory societies [27]. These advantageous outcomes must be balanced off
against intricate and multifaceted problems, though. Some worries in the medical
44 S. M. Sabharwal et al.

sector, which stand to gain greatly from Big Data remedies, pertain, for example,
to the challenge of upholding rules of ethics regarding confidential information,
where the huge amount of data may make it difficult to obtain the updated and
particular consent necessary before every instance of handling it requires place. A
further instance comes from the danger that learners in educational institutions feel
constantly watched due to the ongoing gathering and analysis of their information,
which could reduce their ability to innovate and/or increase their anxiety levels [28].
The discussion of Big Data must emphasize the various potentially moral and
social factors that may arise and examine the related legal, sociological, and moral
difficulties. Here, it is necessary to develop an ecological and moral structure for
defending rights for individuals, reducing risks, and making sure that ethical princi-
ples and actions are consistently in line [29]. A structure like this ought to be able to
boost public and corporate trust in big data as well as the data revolution. Enormous
data involves enormous responsibility, so as the European Data Protection Supervisor
(EDPS) stated, “suitable data security measures need to be in existence.”
Modern ethical conversations have centred on issues relating to trust beyond all
else, confidentiality, anonymity, encoding, and monitoring. As innovation improves,
the discussion is shifting more and more towards self-driving technologies and
machine learning. Future hazards of novel kinds may also be recognized and
discussed when technology develops even more, it is probable [30].

2.3 Research Ethics

Considering the advancements in heavily data-driven study, it is important to consider

how well ethical standards created in the fields of psychology and sociology using
fundamental ideas like informed consent, threat, harm, possession, etc. can be
instantly implemented in the field of data science or whether they need to be supple-
mented with other ideas aimed at safeguarding “living subjects” in heavy data inves-
tigations. When data may easily be linked to a person who faces the possibility of
harm in daily life, personal objects’ safety is typically applied. Big Data, however,
may put a significant barrier between how sensitive information is used in daily
life. Because of the numerous, intricate justifications for collaborating, utilizing, and
disseminating findings from studies, it is difficult to forecast the kinds of potential
consequences to people if technological safeguards are insufficient and fail to avoid
the reuse of confidential information across multiple repositories [31].
We must re-evaluate the conventional perspectives if these obstacles are too great
to be overcome within the current frameworks of research morality. In this section,
an innovative model of scientific ethics designed for data science might be developed
to better place the “individual” at the core of the discussion. The growing body of
research on confidentiality as well as other rights of citizens supports the notion that
the moral consequences of big data are taking centre stage in the European Union
(EU) discussion, and that everyone is working towards finding workable solutions
that strike a balance between maximizing the worth of big data despite threatening
basic human rights [32].
AI and Blockchain for Secure Data Analytics 45

2.4 Big Data and Its Impact on Privacy

Big Data’s widespread gathering, preservation, and evaluation of enormous amounts

of personal data have a significant influence on privacy. As additional information
is collected and acquired from many sources, worries about privacy violations and
potential exploitation of confidential information have grown significantly. A few
significant ways that Big Data impacts privacy are listed below.

2.4.1 ‘Creep Factor’ of Big Data

A crucial development for the European economy is the utilization of Big Data,
contemporary surveillance technologies, as well as data collection methods. In virtue
of the updated legislative structure, it also presents substantial legal issues from a
data safety standpoint. While the information is frequently utilized and repurposed
in ways that were practically unthinkable when the data were obtained, conven-
tional processes and conceptions of confidentiality protections (such as informed
permission approaches) may be insufficient in some cases under the Big Data model
[33].
As noted through the EDPS, the regard for a person’s right to confidentiality and
the freedom to the safeguarding of private information is inextricably linked with
the regard for the worth of others. The European Charter of Fundamental Rights
recognizes that one’s right to respect for humanity is unalienable. This fundamental
right may be violated by acts like dehumanization, which take place when a person is
considered as an object used for another person’s benefit (European Data Protection
Supervisor, Opinion 4/2015).
Big Data innovations’ effects on confidentiality span from data inequality and
automated choice to category confidentiality and sophisticated segmentation. It is
much more important if individuals share sensitive data online while experiencing
the major stages of their lives and do so with varying degrees of knowledge. Indi-
viduals can frequently almost disappear from view here for data mining applications
that utilize publicly available information from social media platforms and other
information connected to an Internet Protocol (IP) address for analytical reasons
[34].
Because of unethical and intentional practices, Big Data has a “creep factor” that
defeats the purpose of confidentiality laws. Such practices, which frequently have the
end goal of addressing and categorizing clients, are made possible by developments
in the analysis and use of Big Data.
The potential for retrieval of the information’s source following the anonymiza-
tion procedure is a further concern in regard to big data. The growing processing
capacity of contemporary personal computers has made de-anonymization technolo-
gies accessible, allowing a track directly to the initial private information. Traditional
anonymization methods, which make every data access non-identifiable by elim-
inating distinctively recognizable details, do have limitations. For example, even
46 S. M. Sabharwal et al.

when personal data submitted by users is substituted in a dataset, de-anonymization

can be accomplished in a relatively short amount of time through straightforward
connections among such unidentified databases, other databases, and personal data.
Reconstruction of the information source may result from the potent insights gained
through the joining of various, particular datasets from various sources. This could
make it possible for those with an interest to definitively pin down certain anatomical
individuals or a restricted number of individuals [35].
Data and new identification raise major issues regarding privacy since once
anonymized (or pseudo-anonymized), the data can be treated outside the subject’s
permission at any time before the individual is re-identified. Lack of proper openness
about the utilization of Big Data makes the circumstance worse since it impacts a
data the subject’s capacity to manage the exposure of their personal details to third
parties along with access to those data, which also has an effect on privacy rights.
It is advised that businesses interested in using big data create open processes and
make sure the general public can comprehend them. In this approach, development
and boundary-setting could be genuinely driven by ethical considerations, effectively
taking into consideration the requirement for confidentiality and freedom of choice
on the part of an individual.

2.4.2 Stigmatization and Manipulation

There is a change in the confidentiality right. Primarily developing as a privilege

to privacy and exclusivity, it has evolved through time into the right to regulate
personal data and is currently heading even farther in the path of enhanced control.
The present movement is approaching the right to control identification and the
statistical picture developed by outside parties that choose the pertinent trends to
be taken into account in metadata. The third stage focuses more on knowledge that
can be used to identify specific behavioural characteristics than it does on data that
can be used to identify specific physical individuals, such as health information,
purchasing choices, medical history, sleep schedules, patterns of movement, access
to the internet, connections, etc. Organizations have grown more open regardless of
the fact that the data is unidentified; in fact, removing all characteristics of group
feeling connected from that data would obstruct the gathering of its substance and,
thus, render it useless [35].
Big Data information can be utilized to specifically influence how individuals
conduct or what they buy. Targeted advertising is one instance, but additional strate-
gies that depend on Big Data’s capacity to uncover concealed relationships and on
the assumed choices and circumstances of a particular group, might be embraced to
promote or dissuade a particular behaviour with rewards whose objectives are less
open to scrutiny.
Stigmatization may also take on new forms, such as when it comes to a group’s
purchasing habits or other private information. Aspects that discriminate are probably
going to occur, particularly as the categories get narrower. Big Data technology may
supersede steadfast fundamental rights safeguards in this regard [36].
AI and Blockchain for Secure Data Analytics 47

The associated massive Big Data gathering and accumulation, as well as the
qualitative evaluation and organized information produced for this purpose but
not subjected to the implementation of the present privacy regulations, constitute
what raises ethical concerns. Therefore, new and creative approaches to considering
the safety of citizens are required, capable of providing sufficient and complete
protection.

3 The Convergence of AI and Blockchain

Before delving into the convergence of blockchain and AI, it is essential to establish
a basic comprehension of both technologies and their extensive array of potential
applications.

3.1 Blockchain Technology

As was already said, the blockchain is regarded as a cutting-edge technology that has
the potential to revolutionize human transactions. The blockchain is a configuration
of distributed ledger technology (DLT) that transmits digital data securely to all nodes
linked in a peer-to-peer (P2P) network before storing it. Blockchain-related technolo-
gies that increase trust, data security, and transparency include shared ledgers and
cryptography. The records of open transactions that are carried out upon authentica-
tion are stored as blocks in the blockchain. The linked user or group of users must
agree to any changes made to the records that have been recorded in this manner
[37].

3.1.1 Blockchain Components

Blockchain technology comprises several essential elements, including below.

Blocks

These are the fundamental units of data that contain a set of transactions. Each block
is linked to the previous one, forming a chain of blocks, hence the term “blockchain.”
48 S. M. Sabharwal et al.

Transactions

Transactions represent the exchange of data or assets between participants on the

blockchain network. These can include financial transactions, contracts, or any other
data that needs to be recorded and verified [33].

Ledgers

The blockchain operates as a decentralized and distributed ledger, meaning that all
participants on the network maintain a copy of the entire transaction history. This
shared ledger ensures transparency and immutability of the recorded data.

Asymmetric or Public-Key Cryptography

This cryptographic method uses a pair of keys, a public key, and a private key, to
secure transactions and ensure that only the intended recipient can access the data.
The public key is used for encryption, while the private key is used for decryption
[34].

Cryptographic Hash Functions

These functions play a crucial role in securing the integrity of data within each
block. They convert data of variable sizes into fixed-length hash codes, making it
challenging for any malicious party to alter the content of a block without detection.

3.1.2 Blockchain Categories

According to Fig. 1, various categories of blockchain exist. Based on data accessi-

bility, blockchain can be classified into the following four types.

Public Blockchain

In this type, anyone can access and submit transactions to the blockchain network.

Private Blockchain

Only a designated group of individuals or entities has the authority to access and
submit transactions on this type of blockchain.
AI and Blockchain for Secure Data Analytics 49

Blockchain

Data Permission Level Functionality and

Accessibility Support

1. Permissionless
1. Public Blockchain 1. Stateless
Blockchain
2. Private Blockchain Blockchain
2. Permissioned
3. Community/Consor 2. Stateful
Blockchain
tium Blockchain Blockchain
3. Hybrid
4. Hybrid Blockchain
Blockchain

Fig. 1 Categories of blockchain

Community/Consortium Blockchain

Multiple groups are allowed to access and submit transactions on this blockchain
category.

Hybrid Blockchain

This type combines different blockchain models to facilitate transactions.

Additionally, blockchain can be categorized based on permission levels.

Permissionless Blockchain

No advance permission is required for participants to access and engage in

transactions on this type of blockchain.
50 S. M. Sabharwal et al.

Permissioned Blockchain

Advance permission is necessary for users to access and participate in transactions

within this blockchain network.

Hybrid Blockchain

In this case, there is a possibility of switching between permissionless and

permissioned blockchains as needed.
These categorizations help define the structure and access controls of blockchain
networks, offering various options for specific use cases and requirements [35].

3.1.3 Blockchain Characteristics

The blockchain technology encompasses several distinct characteristics, including

below.

Decentralization

The data and control in a blockchain network are distributed across multiple nodes,
ensuring that no single entity has complete control, enhancing security and resilience.

Continuous Data Flow

Transactions are continuously added to the blockchain, creating an immutable and

chronological record of all activities on the network.

Validated Transactions

Transactions on the blockchain are verified and validated by consensus mechanisms,

ensuring the integrity and accuracy of the recorded data [35].

Transparency

The blockchain’s public nature allows all participants to view and audit the entire
transaction history, promoting trust and accountability.
AI and Blockchain for Secure Data Analytics 51

Anonymity

While transactions are transparent, participants’ identities are often represented by

cryptographic keys, providing a certain level of anonymity.

Autonomy

The blockchain operates based on pre-defined rules and smart contracts, enabling
automatic execution without the need for intermediaries.

3.2 Power of AI

Indeed, decentralization, continuous data flow, validated transactions, transparency,

privacy, and autonomy are key characteristics of blockchain technology. Figure 2
illustrates a resource management framework leveraging blockchain technology. This
framework comprises various elements and relies on three primary resources: energy
and infrastructure, communication, and computing [36]. The blockchain functions
as a central hub for data exchange and also serves as a regionally distributed system
for resource management.
By implementing such a system, an open and transparent market is established,
facilitating seamless transactions between different entities. The use of blockchain
transactions further enhances the efficient utilization of available resources, opti-
mizing their allocation and usage. This combination of characteristics and the integra-
tion of blockchain technology offers promising solutions for resource management
and transactional efficiency.

3.2.1 Blockchain and AI Applications

The vast potential of intelligent computers equipped with machine learning capa-
bilities has sparked a remarkable surge in the scope of AI applications, affecting
corporations, governments, and society as a whole [37]. As the manuscript proceeds,
it offers further background information on both blockchain and AI, with a primary
focus on the following aspects as shown in Fig. 2.

Supply Chain Management

With the help of blockchain, every transaction and operation in a supply chain can
be recorded transparently and irrevocably. The procedure for supply chains may be
enhanced by using AI to analyse this data and provide insights into inconsistencies,
identifying fraud, and statistical analysis [38].
52 S. M. Sabharwal et al.

Healthcare Internet of
Things

Identity Smart
Management Contracts

Supply Chain Blockchain

and AI Decentralized
Management AI Models
Applications

Intellectual Autonomous
Property Rights Vehicles

Finance and Gaming

Payments

Fig. 2 Blockchain and AI applications

Identity Management

Blockchain can be used to verify identities in a decentralised and safe manner. The
preciseness and dependability of identity verification processes can be improved by
using AI-powered identification of faces and biometric identification.

Healthcare

Blockchain can enable safe and accessible medical records, enabling smooth access
to data and sharing for both patients and healthcare professionals. Huge volumes of
health information may be analysed by AI to help with diagnosis, therapy suggestions,
and medication discovery [39].
AI and Blockchain for Secure Data Analytics 53

Internet of Things (IoT)

By integrating AI and blockchain, IoT systems and networks may be made more
secure and private. While AI can evaluate data produced by IoT devices to arrive
at wise judgements, blockchain can offer a decentralised framework for device
connectivity and verification of transactions.

Smart Contracts

These self-executing agreements have specific terms and constraints incorpo-

rated directly into the programme code. Smart contracts can incorporate AI to
autonomously initiate particular operations based on predetermined conditions
without the involvement of a human.

Decentralised AI Models

Blockchain can make it easier for AI models to be created and shared decentralized.
Smart contracts enable data owners to monetize the use of their data while maintaining
control over access, promoting fair reimbursement and security [40].

Autonomous Vehicles

Blockchain is capable of being used to safely store and distribute information about
the operation, upkeep, and accidents of self-driving automobiles. This data can be
processed by AI to enhance the driving skills and general safety of the cars.

Intellectual Property Rights

Blockchain can create a verifiable and unchangeable record of rights to intellec-

tual property. By examining a large amount of online content, AI can help with
safeguarding intellectual property and infringement of copyright identification.

Finance and Payments

Blockchain-based digital currencies and smart contracts have the potential to revo-
lutionise financial interactions and make international payments easier and safer. In
order to identify fraud, evaluate credit risk, and improve strategies for investment,
AI can analyse financial information [41].
54 S. M. Sabharwal et al.

Gaming
Blockchain can be used in gaming to support asset ownership, allowing gamers
to truly own the characters and objects they use in-game. AI can improve game
experiences by creating dynamic and customised content.
By addressing a variety of issues with confidence, openness, and efficiency across
a variety of disciplines, integrating artificial intelligence with blockchain has the
ability to open up new possibilities and upend established business models. It’s
crucial to remember that implementing these technologies comes with its own set of
difficulties and concerns, including flexibility, data protection, and compliance with
laws and regulations.

3.3 The Convergence of AI and Blockchain for Data

Analytics

The fusion of blockchain with AI has shown to be a formidable and revolutionary

combination that has the ability to change data analytics. The combination of each of
these technologies offers special capabilities to improve the safety, confidentiality,
and dependability of data-driven operations as organizations progressively depend
on data to gather insightful knowledge and make knowledgeable decisions [42].
AI has made strides in predictive modelling, machine learning, machine vision,
as well as other applications of data analytics by handling and analysing enormous
quantities of data. However, as AI algorithms frequently need access to large volumes
of data, issues about security and confidentiality of information are raised.
Such problems are addressed by the use of blockchain technology, which offers a
safe and open framework for storage of information and administration. Blockchain
technology is recognized for its decentralized and irreversible ledger. Data integrity
is improved and unauthorized access is prevented through blockchain, which guar-
antees that information cannot be changed or tampered therewith once it has been
recorded [43].

3.3.1 Advantages

The convergence of AI and blockchain brings several advantages to data analytics.

Secure Data Sharing

Organizations have the ability to safely share data with a variety of partners using
the decentralized architecture of the blockchain while maintaining authority over
its accessibility and utilization. Smart contracts make it possible to automate and
permissioned share data, protecting the safety and confidentiality of such data.
AI and Blockchain for Secure Data Analytics 55

Data Lineage and Auditing

The immutability of blockchains makes it possible to track the origins of data in

an open and accountable manner. This characteristic increases the credibility of AI-
generated findings by enabling partners to confirm the reliability of information
sources and the analytical techniques [44].

Analytics that Protect Privacy

By using artificial intelligence algorithms to secure information that is kept on the

blockchain, businesses may run data analytics without disclosing the unprocessed
information. Data privacy and security are ensured by this method.

Tokenized Incentives

The capacity of blockchain to produce electronic currencies paves the way for incen-
tive structures for data producers, artificial intelligence (AI) model creators, and
validators [45]. For the contributions they make, people or organizations might
receive tokens, encouraging a cooperative data analytics ecosystem.
Blockchain may facilitate federated learning, in which artificial intelligence (AI)
models undergo training using global sources of data instead of centralized data
storage. This strategy protects confidentiality of information and promotes data
exchange for better analytics results.
Although the fusion of blockchain with AI offers exciting possibilities for
analytics of data, it also comes with a number of distinct difficulties, such as flexi-
bility, interoperability, and processing overhead. In order to guarantee ethical deploy-
ment, legislative and regulatory considerations like confidentiality of information
compliance and rights to intellectual property have to be properly considered [46].
In the end, the fusion of AI with blockchain results in an effective blend
that strengthens safety, confidentiality, and confidence in data-driven operations,
improving data analytics [47]. Organizations may open up new avenues for safe
and dependable data analytics, resulting in more knowledgeable choices and
game-changing insights, by combining the benefits of the two platforms.

4 Enhancing Data Privacy with Blockchain

One of the key advantages of combining these technologies is the enhancement

of data privacy through blockchain. Blockchain’s decentralized and impenetrable
design offers a solid framework for protecting sensitive data and guaranteeing user
privacy. The following are some ways that blockchain improves data privacy.
56 S. M. Sabharwal et al.

4.1 Decentralization and Immutability

The two core characteristics of blockchain technology—decentralization and

immutability—play a key role in increasing data privacy.

4.1.1 Decentralization

Blockchain runs on a dispersed system of nodes, as opposed to conventional central-

ized systems, where information is held and managed by an individual body. Every
member of the group has an exact replica of the complete blockchain ledger. By
doing away with the requirement for a centralized middleman, this decentralized
architecture lowers the possibility of one single point of breakdown or unauthorized
use of data [48].
Information is not entirely managed by any one organization in a decentralized
blockchain network, which renders it more resistant to breaches and ensures no one
organization has complete authority over the information. As individuals are able
to independently verify the veracity and reliability of data without depending on a
centralized authority, decentralization promotes confidence among users.

4.1.2 Immutability

The failure to change or remove data once it has been stored on a blockchain is referred
to as immutability. The blockchain is a network of data links where each block is an
encrypted version of the one before it. Any effort to change data would necessitate
altering the information in following blocks, which is practically impossible because
of the consensus technique employed by blockchain systems [49].
When data is stored on the distributed ledger, its immutability guarantees that
it will remain accurate and open during its entire existence. An accountable and
open record of data operations and modifications is provided by this attribute, which
improves the integrity of data and minimizes unauthorized updates.
Blockchain technology establishes a framework for private data that is safe and
reliable by fusing decentralization and immutability. Stakeholders can confidently
communicate confidential information because they understand that the data is trans-
ferred reliably via a network along with any modifications to the information have
been openly documented and verified. These characteristics are especially useful
in applications in which security and confidentiality of information are essential,
such as healthcare, financial services, distribution chains, and identity administra-
tion, enabling organizations and individuals to preserve authority over personal data
and protecting data confidentiality in a quickly changing environment.
AI and Blockchain for Secure Data Analytics 57

4.2 Preventing Unauthorized Access

4.2.1 Blockchain networks use access controls and encryption techniques to handle
issues with data security. Operations on open blockchains like Bitcoins are fictitious,
which means that user identities are hidden behind encrypted addresses. Whereas
secure blockchains can add more severe access restrictions to limit visibility of
information to authorized parties, public ones prioritize openness [50].
4.2.2 Data access control at the granular level is made possible by the use of
smart contracts, executable code on the distributed ledger. The smart contract can
set access rights that permit particular individuals or organizations to see or alter
data in accordance with established regulations. This strategy minimizes the risk of
unauthorized data breaches by ensuring that only authorized stakeholders can access
confidential data [51].

4.3 Mitigating Security Breach Risks

In data-driven initiatives, reducing the likelihood of privacy thefts is of the utmost

significance. Blockchain technology offers a number of techniques to improve secu-
rity of information and reduce the chance of breaches [52]. The following are
important tactics to use blockchain to reduce the risk of privacy theft [53].

4.3.1 Single Point of Failure

Blockchain is based on a distributed system of nodes, which eliminates single points

of failures. Such attacks are less damaging when a single node has a security breach
because the whole network is not compromised [54].

4.3.2 Consensus Procedure

Blockchain networks use consensus procedures to verify and concur on the accuracy
of data. By preventing hostile actors from changing data without agreement, the
blockchain’s entries are very difficult to tamper with [55].

4.3.3 Immutability

Data that has been stored on the distributed ledger is no longer able to be changed or
erased. By offering an impermeable to record of auditing that discourages possible
attackers, this function protects the reliability and validity of data [56].
58 S. M. Sabharwal et al.

4.3.4 Encryption

Blockchain networks are capable of safeguarding data throughout its transfer and
preservation by utilizing cutting-edge encryption methods. By encrypting data, you
can make certain that even if unauthorized individuals access the distributed ledger,
the data will still be encrypted and impossible to decrypt without the right keys for
decryption [56].

4.3.5 Identity Management

Cryptographic keys are used in blockchain to identify users, ensuring safe authenti-
cation. The risk of unauthorized utilization of a network or confidential information
is decreased by effective management of identities [57].

4.3.6 Access Controls

The blockchain’s smart contracts allow for the definition of granular controls on
access for data. The danger of hacking into data can be reduced by including
access authorizations in the smart contract, which will ensure that only authorized
individuals can access particular data [58].

4.3.7 Regular Monitoring and Periodic Inspections

Regular monitoring and periodic inspections of the distributed ledger system help
quickly identify unusual activity and possible breaches of confidentiality. Early
identification enables rapid responses and mitigation measures.

4.3.8 Bug-Free Smart Contracts

To prevent weaknesses, smart contracts must be carefully developed and audited.

Prioritizing secure coding practices is crucial since smart contract flaws or bugs
might result in data loss or security violations [59].

4.3.9 Blockchains with Access Restrictions

Using a blockchain with restricted access is recommended for scenarios where confi-
dentiality of information is of utmost importance. As a result, there is less chance of
confidential data being accessed by unauthorized people [60].
AI and Blockchain for Secure Data Analytics 59

4.3.10 Data Backups and Redundancy

Keeping blockchain’s private data redundant and regularly backing it up reduces the
consequences of possible breaches of security by guaranteeing accessibility in the
event of infrastructure failures or crimes [61].
Organizations may considerably improve security, lower the danger of data theft,
and guarantee the accuracy of data in initiatives that are data-driven by putting these
techniques into practice. It is vital to understand that no system is completely imper-
vious to privacy hazards, and that constant enhancement and continued diligence are
required to keep on top of emerging threats.

5 Secure Data Sharing and Collaboration

The exchange and cooperation of protected data are essential components of data-
driven initiatives, and the use of blockchain technology is vital in strengthening
the safety and confidentiality of these procedures. Historically, security breaches,
unauthorized use, and a lack of user confidence have presented problems for sharing
information and engagement [62]. The issues can be addressed through the utilization
of decentralized, immutable, and transparent blockchain technology.

5.1 Securing Remedies

To guarantee that people and organizations have the opportunity to seek fair and
reasonable solutions whenever challenged with legal conflicts or unfair practices,
it is essential to secure remedies [63]. Remedies are used to remedy harm, right
injustices, and defend the rule of law, whether in terms of legal systems, economic
dealings, or societal challenges.

5.1.1 Decentralization and Immutability

Blockchain’s decentralized structure of nodes and immutability make it possible

to function without a governing body. Data is spread across numerous nodes in
the network, increasing its resistance to single points of failures and lowering the
possibility of unauthorized access [64]. Data that has been stored on the blockchain
remains immutable, which means that it cannot be changed or removed without the
entire network’s approval. These characteristics foster trust between stakeholders by
ensuring the validity and integrity of provided data.
60 S. M. Sabharwal et al.

5.1.2 Consensus Mechanisms

Consensus techniques are used by blockchain systems to verify and consent to the
legitimacy of data contributed to the blockchain. These procedures, such as Proof of
Work (Pow) or Proof of Stake (PoS), guarantee that data is validated and approved
by a significant number of network members prior to being added to the network. By
preventing malevolent actors from altering or interfering with data, this consensus
method improves data security [65].

5.1.3 Encrypted Data Sharing

Blockchain can help with confidential information sharing, allowing for the safe
transmission and storage of confidential information. Information that is shared is
further protected by encryption, which guarantees that only authorized individuals
with the necessary decryption credentials are allowed to view the data [66].

5.1.4 Smart Contracts

Self-executing code known as “smart contracts” is kept on the distributed ledger and
automatically executes specified actions when certain criteria are satisfied. Commu-
nication contracts can be implemented via smart contracts, which guarantee that
information can only be used and utilized in line with predetermined rules and rights.
As a result, there are no longer any intermediaries required in the processes for sharing
data, which lowers costs and minimizes security risks [67].

5.1.5 Tokenized Incentives

Blockchain enables the development of digital currencies that can be utilized for
monetary engagement and the exchange of data. Tokens may be awarded to users
for exchanging data, assisting with analytics procedures, or offering insightful
commentary. Such tokenized incentives promote information exchange and build
a cooperative ecology [68].
Reliable collaboration and the sharing of data are made possible by utilizing
blockchain’s characteristics without sacrificing security or confidentiality of infor-
mation. The use of distributed ledgers in data-driven projects encourages member
accountability, openness, and confidence, which improves data sharing procedures
across a range of businesses.
Blockchain-based smart contract systems are a foundational element in ensuring
secure data sharing and collaboration. Smart contracts are self-executing contracts
with predefined rules and conditions written in code. They reside on the blockchain
and are automatically executed when specific conditions are met, without the need for
intermediaries. These contracts enable trust less interactions, as all parties involved
AI and Blockchain for Secure Data Analytics 61

can rely on the transparent and immutable nature of the blockchain to enforce the
agreed-upon terms.

5.2 Advantages

In the context of data sharing and collaboration, blockchain-based smart contract

systems offer several key advantages [69].

5.2.1 Transparency and Trust

The blockchain makes smart contracts accessible to all users, fostering faith in the
sharing information process. There is not a requirement to depend on a single source
for validation because anyone may examine the agreement’s logic and check that it
is being executed.

5.2.2 Autonomy and Decentralization

Once implemented, smart contracts run independently without outside assistance.

This decentralized structure reduces the possibility of human mistake or manipulation
and eliminates the likelihood of single points of failure.

5.2.3 Data Security and Privacy

Smart contracts provide the ability to impose access controls and define who has
access to and is allowed to use the shared data. Security and confidentiality of infor-
mation can be ensured by storing encrypted data on a distributed ledger and limiting
access to critical information using secret keys [70].

5.2.4 Immutable Records

On the distributed ledger, smart contracts keep track of all communications and data
transfers, establishing a lasting record of every transaction. By offering an auditable
track, this feature improves data reliability and transparency.
62 S. M. Sabharwal et al.

5.2.5 Automatic Execution

Smart contracts speed up data-sharing procedures by controlling the carrying out of

agreement clauses. This effectiveness lowers expenses while accelerating member
engagement.
Blockchain coins can be used by smart contracts as rewards for data exchange and
cooperation. Tokens can be given to members in exchange for their data, knowledge,
or computing power, promoting a cooperative and financially motivated ecosystem.
However, as defects or weaknesses in the code might have serious repercussions,
it is crucial to carefully create smart contracts and properly analyse them for possible
weaknesses. To guarantee adherence to the relevant rules and laws, legislative and
governmental considerations must also be made while creating agreements [71].
In general, by offering a safe, open, and self-governing framework for data-driven
conversations based on blockchain technology, smart contract networks play a critical
part in improving data sharing and cooperation. Smart contract incorporation in data-
driven initiatives enables members to participate in secure and reliable data transfers,
fostering innovation and boosting productivity in a variety of industries.

5.3 Data Usage Agreements

A key step in guaranteeing secure data exchange and teamwork within the frame-
work of the blockchain system is the definition of use of information contracts.
These contracts specify the circumstances under which parties can obtain, utilize,
and exchange data [72]. Organizations can increase confidence and dependability
across providers of data, customers, and various other partners by implementing
simple and open use of data contracts.

5.3.1 Data Usage Agreement Element

Key elements in defining data usage agreements on the blockchain include below.

Data Use Agreements

These agreements define who is granted access to the data that is shared and what
rights they have. Varying roles or entities associated with the partnership may have
varying access privileges. For example, although some individuals may just have
permission to peruse the data, individuals might be given the authority to edit or
analyse it [73].
AI and Blockchain for Secure Data Analytics 63

Restrictions on Data Usage

Limitations on the use of data may be outlined in Data Usage Agreements. This
may include limitations on the utilization of the data for business reasons, beyond
the parameters of the partnership, or even dissemination. These limitations assist in
making certain that the information is utilized sensibly and for the purpose it was
created.

Intellectual Property Rights and Data Ownership

The contract should specify the ownership of data responsibilities and handle any
issues with intellectual property. Either data producers keep control over their data or
provide that ownership to the collaborative organization should be made clear [74].

Safety and Protection

Clauses addressing the safety and privacy of data should be included in data usage
contracts. This involves verifying adherence to pertinent data protection laws and
defining safeguards to guard against unauthorized access, manipulation, or violations
of data security.

Lifespan and Termination

The contract should outline the collaboration’s lifespan as well as the circumstances
under which it may be terminated. It should specify how information will be managed
after the cooperation is finished, like by removal or return to the original data
providers [75].

Consent and Ethical Considerations

Data usage contracts should include provisions addressing consent specifications,

particularly for critical or private information. User consent for sharing information
must be obtained in writing, and participants must be made aware of how their
information will be utilized.

Dispute Resolution

In the event of conflicts or contract violations, the contract may include procedures
for resolving disputes. By establishing an accountable record of conversations, the
openness and inviolability of blockchain technology can help resolve conflicts.
64 S. M. Sabharwal et al.

Participants can gain from an automatic and impenetrable system that upholds
the agreed-upon rules by putting such information usage contracts on the distributed
ledger using smart contracts. The conditions are open, accountable, and imple-
mented independently because of the implementation of blockchain technology,
which reduces the need for faith in a centralized authority and improves security.

5.4 Safe Data Exchange and Collaboration

Data-driven initiatives must enable secure data interchange and cooperation, and
blockchain technology provides helpful solutions to improve trust and safety in
these procedures [76]. Several major advantages result from incorporating blockchain
technology into data collaboration and exchange.

5.4.1 Decentralization and Security

The decentralized nature of blockchain eliminates the requirement for a centralized

organization to promote interaction and cooperation. Because each member of the
system has an exact replica of the blockchain, redundancy of information is ensured
and the likelihood of a single point of failures is decreased [77]. By reducing depen-
dence on a single body for handling data, its decentralized nature improves data
security.

5.4.2 Immutable Data Files

Information transmitted and stored on the blockchain is inviolable, which means

that once it has been done so, it cannot be changed or removed without the entire
network’s approval. Through the provision of an accountable record of all information
exchanges and activities operations, this feature protects the reliability and legitimacy
of shared data.

5.4.3 Encrypted Security

To protect data, blockchain uses cutting-edge encryption techniques. Because trans-

actions on the distributed ledger are by encryption hashed and connected, it is very
challenging for attackers to covertly change data. Using both public and private key
encryption, only authorized parties may safely access and share data.
AI and Blockchain for Secure Data Analytics 65

5.4.4 Using Smart Contracts to Facilitate Automated Collaboration

The blockchain’s smart contract structures, which are pieces of executable code,
allow for autonomous and automatic collaboration. These agreements ensure that
collaboration and sharing of data take place in accordance with predetermined terms
by automatically enforcing specified norms and restrictions [78]. This decreases the
possibility of human error and does away with the necessity for middlemen.

5.4.5 Access to Permissioned Data

Blockchain networks may be created as permissioned systems, in which users must

obtain permission before they can participate in the system to utilize its shared data.
This makes it possible to have exact control about data access and guarantees that
only parties with permission can work together to contribute to the system [79].
Blockchain currencies can be utilized as rewards for secure data sharing and
teamwork. Tokens may be given to members in exchange for their data, knowledge,
or computational assets. The collaborative and rewarded sharing information ecology
is fostered by these tokenized incentives, which promote active engagement.

5.4.6 Superior Data Privacy

Sensitive data is safely exchanged and safeguarded on the blockchain, facilitating

secure information sharing. An additional degree of security and confidentiality
is provided by encryption, which makes sure that only people with the necessary
permissions and keys for decryption can access the data.
Organizations can establish a safe, open, and cooperative platform for sharing
information by utilizing blockchain technology. By incorporating blockchain tech-
nology into data-driven initiatives, participants are more likely to feel trusted, data
security is improved, and the collaboration process is streamlined, resulting in
insights based on data and developments that are more dependable and effective.

6 Federated Learning and Data Privacy

Organizations can leverage federated learning to fully realize the benefits of cooper-
ative analysis of data and artificial intelligence while maintaining the confidentiality
of data. This strategy is particularly pertinent to sectors like medical care, banking,
and communications that have strict regulations on the confidentiality of informa-
tion [80]. Federated learning is constantly changing as advancements in technology
occur, and continuing research intends to further enhance its secure characteristics,
making it an important tool for safe and considerate data analytics.
66 S. M. Sabharwal et al.

6.1 The Concept of Federated Learning

A decentralized machine learning method called federated learning enables different

devices or organizations to cooperatively train a single network without exchanging
their unprocessed information with a centralized server. When using confidential
information, security and confidentiality problems may arise since in conventional
artificial intelligence, data is gathered and centralized in one place for model training
[81].
Federated learning allays these worries by limiting updates to models to being
sent to a centralized server while maintaining data on personal servers or endpoints
[82]. The following steps are often included in the federated learning procedure, as
shown in Fig. 3.
Organizations can leverage federated learning to fully realize the benefits of coop-
erative analysis of data and artificial intelligence while maintaining the confidentiality
of data. This strategy is particularly pertinent to sectors like medical care, banking,
and communications that have strict regulations on the confidentiality of informa-
tion [80]. Federated learning is constantly changing as advancements in technology
occur, and continuing research intends to further enhance its secure characteristics,
making it an important tool for safe and considerate data analytics.

Fig. 3 Federated learning

Initiali zation

Distribution

Local Model Training

Model Aggregation

Model Improvement

Reiterate
AI and Blockchain for Secure Data Analytics 67

6.1.1 Initialization

At the central server’s level, the worldwide model is constructed and initialized.

6.1.2 Distribution

Each device or network holding local information receives the worldwide model.

6.1.3 Local Model Training

Despite exchanging the unprocessed information with the centralized server, every
device independently develops the worldwide model with its own individual data
[83]. The efficiency of the model may be enhanced through this training procedure
over several cycles.

6.1.4 Model Aggregation

Every equipment transmits its modified model parameters—not unprocessed data—

back to the centralized server after regional training.

6.1.5 Aggregated Model Improvement

Using the information gleaned from multiple data sources, the central server collects
the model upgrades across every device.

6.1.6 Reiterate

To further enhance the worldwide model, the repeated localized training, model
accumulation, and model updating processes are carried out.

6.2 Advantages

Federated learning offers several advantages.

68 S. M. Sabharwal et al.

6.2.1 Data Security

By keeping data decentralized, federated learning lowers the possibility of data

disclosure, compromises, or unauthorized access. To improve the security and confi-
dentiality of data, confidential information is kept on devices that are local, and just
latest models are transferred [84].

6.2.2 Efficiency

For huge datasets, centralized training of models may not be feasible because it
involves data transport and can be computationally challenging. Federated learning
divides up the training procedure, minimizing the demand for substantial transfer of
information and conserving computer resources.

6.2.3 Flexibility

Federated learning is perfect for situations with a variety of data distributions since
it enables diverse devices and information sources to take a role in the process of
training models.

6.2.4 Edge Computing

Federated learning is particularly appropriate for edge computing situations, wherein

endpoints have restricted processing capacity or connectivity, as information is
handled and taught directly on devices [85].
In many fields where confidentiality of information, safety, and communication
are crucial, such as healthcare, banking, mobile technology, the Internet of Things
(IoT), and federated analysis, federated learning is used. While federated learning
overcomes issues with data privacy, it also raises issues with model integration,
communication effectiveness, and managing uneven distributions of data among
devices. These issues will be addressed, and the entire potential of federated learning
for cooperative and private machine learning will be realized, thanks to continuous
studies and developments in the field.

6.3 Preserving Data Privacy with AI and Blockchain

Integration

An effective way to protect confidentiality of information in projects based on data

is through the combination of artificial intelligence and blockchain. Businesses may
AI and Blockchain for Secure Data Analytics 69

improve security and confidentiality of information while allowing sophisticated

machine learning and analytics capabilities by integrating the best features of the
two technologies [86]. The combination protects the privacy of data in the following
significant ways.

6.3.1 Decentralized Data Preservation

The decentralized nature of the blockchain reduces the danger of one point of fail-
ures and unauthorized access by allowing data to be disseminated throughout the
network. Highly confidential information can be less exposed since artificial intel-
ligence algorithms can work instantly on data that is kept in the blockchain rather
than having to send it to a centralized server [87].

6.3.2 Protected Information Sharing

Blockchain permits the storing and transfer of encrypted data, guaranteeing that data
is kept private and secure throughout sharing and interaction. Further preserving
data privacy, artificial intelligence algorithms are capable of handling data that is
encrypted without entirely decrypting it.

6.3.3 Smart Contracts with Data Access Management

Smart contracts on the blockchain let you control data access in very specific ways.
Data confidentiality agreements can be rigorously enforced by organizations by
defining who has permission to see what information and under what circumstances.

6.3.4 Tokenized Data Accessibility

With blockchain currencies, information use may be restricted as well as information

sharing can be rewarded. The owners of data can create tokens that allow permis-
sion to their information, and currencies can be awarded to artificial intelligence
algorithms for analysing information without actually gaining access to it [88].

6.3.5 Auditable Data Provenance

The transparency and immutability of blockchain technology create a verifiable

record of data exchanges and exchanges. Data owners can monitor how their
information is used, increasing responsibility as well as openness.
70 S. M. Sabharwal et al.

6.3.6 Data Privacy and De-identification

When analysing and interpreting data, artificial intelligence approaches like different
levels of privacy and the data anonymity abilities of blockchains can be utilized to
secure people’s identities as well as confidential information [89].

6.3.7 Regulation Adherence

Blockchain’s secure nature and AI’s capacity to implement data privacy standards
ensure adherence to data protection laws like GDPR, HIPAA, and more.

6.3.8 User-Controlled Data Exchange

By combining blockchain and artificial intelligence, consumers can take more owner-
ship over their information, choose who has permission to use it and how, and
maintain their personal information security choices.
Organizations may create a safe and confidentiality data environment by utilizing
the convergence of artificial intelligence and blockchain. Researchers can maintain
authority over what they share, and artificial intelligence algorithms can learn from
a large body of data without jeopardizing the privacy of individual users. Through
this combination, people can collaborate and innovate in a protection-conscious way
while also improving the security of their data.

7 Tokenized Incentives for Users

The process for launching tokenized rewards for individuals includes giving users
digital tokens as prizes or rewards for their involvement, engagement, or certain
acts inside an environment or ecosystem. These tokens have value and may be
traded, redeemed, or applied in a variety of ways throughout the system [90]. In
blockchain-based initiatives and decentralized platforms, tokenized rewards have
become popular as a means of promoting user engagement, adherence, and contri-
butions. By making tokens available, companies can build a community in which
users are encouraged to take an active role in the development and accomplishments
of the platform.

7.1 Introducing Token Economies

A decentralized system based on the blockchain system called a “token economy”

makes use of tokens as virtual assets that symbolize value and reward different acts or
AI and Blockchain for Secure Data Analytics 71

behaviours inside a network. Smart contracts, which serve as autonomous agreements

on the distributed ledger, are used for manufacturing and distributing tokens. These
tokens of value have uses beyond typical cryptocurrency and can be utilized as a sort
of virtual currency inside the overall system [91].
Token economies are intended to increase participation among users, encourage
desirable behaviours, and build a network that can support itself. In accordance with
the guidelines and norms outlined in the smart contracts, network users can acquire,
trade, and use tokens for particular tasks, products, or benefits [92].
Key elements and characteristics of token-based economies are as follows.

7.1.1 Token Creation

Tokens frequently arise through token generation events (TGEs) or initial coin offers,
in which a particular number of tokens are distributed and sold to the attendees or
shareholders. The token economy can be developed and expanded using the money
earned from such occurrences.

7.1.2 Token Utility

In the economy of tokens, tokens may serve a variety of purposes, including granting
users’ access to products and services, the right to vote, or giving them incentives
for making contributions. The value of the token’s layout plays a crucial role in
encouraging desirable behaviours and increasing engagement among users.

7.1.3 Incentives and Rewards

For certain behaviours or services to the ecosystem, users are rewarded and incen-
tivized with tokens. For instance, users might be compensated with cryptocurrencies
for data communication, transaction validation, platform creation work, or customer
referrals [93].

7.1.4 Transparency and Decentralization

Token economies run on decentralized blockchain systems, which have both of

these qualities. On the blockchain, every transaction and token transfer are tracked,
promoting openness and accountability.
72 S. M. Sabharwal et al.

7.1.5 Interoperability

Users may exchange or transfer tokens for other digital currencies or fiat money on
a variety of cryptocurrency trading platforms. The flexibility and increased value of
the token result from this interconnectivity.

7.1.6 Governance

Decentralized governance methods may be implemented in token-based economies,

allowing users of tokens to take part in making decisions and cast votes on crucial
matters affecting the growth and functioning of the ecosystem.

7.1.7 Trust and Security

The immutability and safety of token exchanges are guaranteed by the use of
blockchain technology, lowering the possibility of theft or tampering. Members can
rely on the token economic system’s regulations and procedures to be followed as
specified in smart contract agreements.
Numerous industries, like banking, gaming, supply chain management, and decen-
tralized applications (DApps), have adopted token economies [94]. They provide a
new framework for exchanges of value and are driven by communities’ ecosystems,
as well as enabling creative business strategies. However, the success of a token
economic system depends on careful planning, a large member base, and ongoing
governance to guarantee justice and stability.

7.2 Rewarding Data Contributors and Validators

Token economies can compensate Data Contributors and Validators for their
assistance to the ecosystem of actors in initiatives that are driven by data.

7.2.1 Data Contributors

As a reward for providing their data, people or organizations that contribute relevant
information to the network may be given tokens. These tokens serve to acknowledge
the value of the data and to motivate its owners to actively engage with the ecosystem.
Token economies encourage collaboration and data sharing by compensating data
contributors [95].
AI and Blockchain for Secure Data Analytics 73

7.2.2 Validators

Validators are essential to maintaining the system’s data reliability and accuracy. They
check and confirm the accuracy of data changes and update the model. Validators
in a federated learning environment could be in charge of collecting model updates
from various devices [96].
Smart contracts that automatically carry out token transfers depending on specified
requirements, such as the quantity and quality of information given or the accuracy
of validation of models, can be used to disperse token incentives following prede-
fined regulations. The system of rewards is made more transparent and equitable
thanks to this automated procedure, which also lessens the demand for centralized
administration [97].
Data-driven initiatives can encourage active involvement, enhance the accuracy
of data, and foster innovation inside the ecosystem by offering tokenized rewards
for users. By bringing members’ interests into line with the network’s achievement,
token-based economies foster engagement and value generation through fostering a
win–win situation.

8 Technical Challenges of AI and Blockchain Integration

To fully realize the promise of this strong combination, a number of technolog-

ical obstacles relating to the combination of artificial intelligence and blockchain
technology must be resolved [98]. Among the principal difficulties are below.

8.1 Scalability Issues

Due to the fundamental structure and properties of both methods, combining artifi-
cial intelligence with blockchain poses serious scalability issues. Due to the require-
ment for every network node to handle and maintain each transaction and piece of
information on the chain, blockchain technology faces a scalability problem. The
blockchain’s size continuously expands with the number of users and operations,
requiring more storage space and increasing the duration of processing [99].
The resource-intensive characteristics of sophisticated machine learning algo-
rithms present scaling issues for artificial intelligence, particularly when handling
enormous quantities of data. Large artificial intelligence models demand a lot of
compute and memory, which can be taxing on blockchain nodes’ resources and slow
down the whole network.
The difficulties with scaling are exacerbated when artificial intelligence and
blockchain are combined because both technologies demand a lot of resources. These
74 S. M. Sabharwal et al.

problems are currently being worked on, and efforts include investigating outside-of-
the-chain remedies, sharding strategies, and resource-saving AI algorithm optimiza-
tion [100]. To increase scalability without sacrificing the safety and decentralization
features of the blockchain, layer-two alternatives like secondary chains and state
networks are also being investigated.

8.2 Interoperability Challenges

Interoperability is the efficient communication and interaction between various

blockchain-based networks and artificial intelligence systems. Data exchange, the
implementation of smart contracts, or transactions across chains may all be a part
of the artificial intelligence and blockchain incorporation process. Interoperability
is difficult to establish since different blockchain systems have different protocols,
consensus procedures, and data structures [101].
When combining artificial intelligence models developed on several blockchain
systems or exchanging data among different blockchain platforms, interoperability
issues may appear. Standardization of data standards and methods of communication
may also be necessary to ensure effective interaction between applications based on
artificial intelligence and smart contracts.
The creation of bridging methods, communication across chain regulations,
and compatible protocols are necessary to tackle interoperability difficulties. The
blockchain ecosystem is working to develop interoperability options that enable the
seamless fusion of blockchain with artificial intelligence systems.

8.3 Processing Costs and Efficiency

Blockchain technology and artificial intelligence both entail relatively expensive

procedures that might result in high expenses for processing and decreased efficiency
as a whole. As a result, whenever AI-related operations are done on the blockchain,
fees for transactions may increase. Models based on deep learning, in particular,
require a lot of processing power.
Additionally, the Proof of Work (Pow) and Proof of Stake (PoS) consensus tech-
niques used by many blockchains consume a significant amount of energy, resulting
in computational inefficiencies and problems with the environment [102].
Research focuses on increasing the computing efficiency of AI systems to increase
processing effectiveness and decrease expenses. In order to save energy and boost
transaction performance, blockchain-based systems are also investigating different
consensus procedures, including Proof of Authority (PoA) and Delegated Proof of
Stake (DPoS).
The combination of AI with blockchain can open up new opportunities and appli-
cations in a variety of sectors, including healthcare, banking, supply chain, and IoT,
AI and Blockchain for Secure Data Analytics 75

by overcoming these technical obstacles. To get over these challenges and fully
realize the endless possibilities of this revolutionary integration, further study and
development are essential.

9 Legal and Regulatory Considerations

To maintain compliance and promote appropriate use of these technologies, it

is important to properly handle the legal and regulatory issues that arise when
combining blockchain and artificial intelligence technologies. Among the most
important legal and regulatory factors are below.

9.1 Data Privacy Compliance

Organizations must solve complicated data privacy issues to meet legal and regu-
latory obligations when integrating AI and blockchain. Organizations must abide
by pertinent data protection rules, such as the General Data Protection Regulation
(GDPR) of the European Union or other local data privacy legislation because these
technologies entail the handling and retention of confidential information.
Data minimization techniques, the right to be forgotten (data erasure), explicit
user authorization for the processing of data, and confidentiality and security of data
are all part of data privacy compliance. Moreover, companies must notify customers
in an open way about how they use, store, and share their data [103].
The decentralized characteristics of the blockchain system can make it difficult
to comply with data privacy laws because data stored there is frequently immutable.
To guarantee that private and sensitive information is properly safeguarded, and that
the privacy of data subjects may be upheld, organizations must carefully build their
blockchain applications.
Conducting privacy impact analyses, putting in place strict access restrictions, and
adhering to best practices for encryption of data and pseudonymization are all things
that organizations should do to ensure data security compliance in the combination
of artificial intelligence and blockchain [104]. To manage the complexity of data
protection rules and ensure that the combination complies with regulatory standards,
interactions with legal specialists are also crucial.

9.2 Ensuring Ethical Implementation

To preserve public faith and confidence, ethical issues related to the combination of
artificial intelligence and blockchain technology must be addressed. Providing that
76 S. M. Sabharwal et al.

the algorithms used by AI are impartial, fair, and open, as well as that they don’t
support or infringe against human rights, requires ethical execution [105].
Organizations should be open and honest about how artificial intelligence (AI)
algorithms are developed, the data utilized for education, and any potential reper-
cussions of the applications powered by AI when combining AI with blockchain.
Understanding how decisions are made and seeing possible biases can be accom-
plished by incorporating explainability and interpretability methods into artificial
intelligence models [106].
Adhering to principles like informed consent, user freedom, and responsibility is
part of ethical execution, along with taking technological factors into account. Data
should be under people’s authority, and they should be aware of how their information
is used in the blockchain and AI ecosystems.
Additionally, organizations should think about how integrating AI and blockchain
would affect society as a whole. Evaluating potential risks, unforeseen effects, and
making sure the integration is consistent with moral principles and social conventions
are all included in this.
Collaboration between ethicists, sociologists, and all other appropriate experts is
crucial for promoting ethical execution. Businesses may address moral issues and
make sure that blockchain and artificial intelligence deployment are done responsibly
by engaging in a candid conversation with stakeholders, such as users and groups
affected by the connection.
Finally, it should be noted that merging blockchain with AI requires careful atten-
tion to legal, regulatory, and ethical issues. Organizations may foster trust, safeguard
user confidentiality, and ensure the moral and responsible use of these innovative
innovations by actively dealing with these issues.

10 Conclusion

The combination of artificial intelligence with blockchain enabling data analytics

presents a viable approach to the problems of data confidentiality, safety, and cooper-
ation. This chapter has examined the difficulties organizations confront in protecting
the safety and confidentiality of data when working on data-driven initiatives. We
have examined the foundational ideas, characteristics, and applicability of the tech-
nology of blockchain to data analytics. Additionally, we have emphasized how
combining blockchain with AI and machine intelligence may improve data confi-
dentiality and security. In distributed AI training environments, the idea of federated
training has been touted as a potent method to protect data privacy. Furthermore, we
have looked at how the blockchain relies on smart contract solutions that allow for
safe data exchange and teamwork while enforcing access restrictions as well as data
usage contracts. Additionally, the use of token-based economies as a way to motivate
and recognize data providers and validators inside the blockchain ecosystems has
been discussed. Nevertheless, this integration has several technological difficulties
that must be successfully controlled, such as scalability problems, difficulties with
AI and Blockchain for Secure Data Analytics 77

interoperability, and processing expenses. The significance of legal and governmental

issues, such as information security compliance and moral execution, has also been
emphasized.
The combined use of artificial intelligence and blockchain for data analytics
has bright future possibilities and many facets. We may anticipate advances in
decentralized federated learning, increased data privacy, scalability remedies, token-
based innovation, ethical artificial intelligence governance, acceptance in a variety
of sectors and laws and regulations as these innovations continue to advance. The
combination of artificial intelligence and blockchain technology will provide effec-
tive privacy-preserving methods that will ensure confidential sharing of information
while retaining security. Artificial intelligence models may be trained effectively
on dispersed devices thanks to decentralized federated learning. Enhancing inter-
operability will enable easy data sharing between multiple blockchain networks. In
order to support massive data analysis applications, solutions for scaling will solve
processing constraints. A new token economy will reward user engagement and
promote the expansion of blockchain ecosystems. Openness, equity, and responsi-
bility will be given top priority in ethical AI governance. Blockchain technology and
artificial intelligence (AI) will be used in a variety of businesses, revolutionizing data
analysis and decision-making. Regulations will strike a balance between user safety
and creativity.
In the end, the fusion of blockchain with AI has tremendous potential to revolu-
tionize data analytics by improving security, confidentiality, and teamwork. These
innovations will influence the development of data-driven initiatives as studies and
innovations proceed, opening up new opportunities for both businesses and people.
The ethical, moral, and technical facets of the data analytics environment must
be carefully taken into account in order for artificial intelligence and blockchain
technology to be integrated responsibly and sustainably.

References

1. Xu L, Wang C, Chen X, Shen X (2020) A survey on blockchain-based data security for

industrial Internet of Things. IEEE Commun Surv Tutor
2. Bernardino M, Ferreto TC et al (2019) Blockchain for AI: review and open research challenges.
IEEE Access
3. Shakir MZ, Saad W et al (2020) Blockchain and AI for federated learning in 6G networks:
opportunities, challenges, and solutions. IEEE Open J Commun Soc
4. Puthal D et al (2019) Secure data sharing in cloud computing using blockchain and attribute-
based encryption. Future Gener Comput Syst
5. Li R, Li S, Jiang T (2020) Secure data analytics in federated cloud-edge computing with
blockchain. IEEE Trans Ind Inform
6. Gondal MW et al (2019) AI-driven blockchain for secure and privacy-preserving decentralized
applications. Sensors
7. Khan SU et al (2019) Securing data sharing with blockchain and attribute-based encryption
for IoT-enabled smart grids. J Netw Comput Appl
8. Anwar F et al (2020) Blockchain and AI-based secure and privacy-preserving data sharing
framework for industrial Internet of Things. Comput Electr Eng
78 S. M. Sabharwal et al.

9. Zhang Y, Liu L, Liu Y (2019) Blockchain-based secure data sharing scheme with fine-grained
access control in cloud storage. Comput Electr Eng 76:66–77
10. Li J, Wu Y, Fan P (2020) A secure data storage scheme using blockchain and attribute-based
encryption for industrial Internet of Things. Comput Electr Eng 85:106658
11. Zhang L, Zhao Y, Liu J (2019) Blockchain-based secure and privacy-preserving content
sharing scheme for industrial Internet of Things. IEEE Trans Industr Inf 15(3):1798–1807
12. Gao F, Qu M, Sun X (2020) Secure data sharing for smart city using blockchain and attribute-
based encryption. Comput Electr Eng 84:106616
13. Yu Y, Zhang T, Xu W (2019) Blockchain-based secure data sharing scheme with user
revocation in cloud storage. Clust Comput 22(S1):1465–1476
14. Cao Y, Wang X (2020) A secure data sharing scheme based on blockchain and smart contracts
for industrial Internet of Things. Secur Commun Netw 2020:1–13
15. Sobrinho OG, Bernucci LLM, Corrêa PLP, Motta RS, Machicao J, Junqueira AS, Lopes F,
Cassaro L, Oliveira L (2021) Big data analytics in support of the under-rail maintenance
management at Vitória—Minas Railway. In: Proceedings of the 2021 IEEE international
conference on big data (big data), pp 126–133
16. Cuzzocrea A, Leung CK, Hajian M, Jackson MD (2022) Effectively and efficiently
supporting predictive big data analytics over open big data in the transportation sector: a
Bayesian network framework. In: Proceedings of the 2022 IEEE international conference on
dependable, autonomic and secure computing, international conference on pervasive intel-
ligence and computing, international conference on cloud and big data computing, inter-
national conference on cyber science and technology congress (DASC/PiCom/CBDCom/
CyberSciTech)
17. Rawat DB, Doku R, Garuba M (2021) Cybersecurity in big data era: from securing big data
to data-driven security. IEEE Trans Serv Comput 14(6):2055–2072
18. Wang F, Wang H, Chen X (2023) Research on access control technology of big data cloud
computing. In: Proceedings of the 2023 IEEE 3rd international conference on information
technology, big data and artificial intelligence (ICIBA), p 10165326. https://doi.org/10.1109/
ICIBA56860.2023
19. Mayer-Schönberger V, Cukier K (2013) Big data: a revolution that will transform how we
live, work, and think
20. Mayer-Schönberger V, Cukier K (2013) Big data: a twenty-first century arms race. Foreign
Affairs
21. Smolan R, Erwitt J (2012) The human face of big data
22. Marz N, Warren J (2015) Big data: principles and best practices of scalable real-time data
systems
23. Chen M, Mao S, Liu Y (2014) Big data analytics: a survey. Mob Netw Appl
24. Heeks R (2017) Big data for development: challenges and opportunities. Inf Technol Dev
25. David RC, Fajar JK, Siahaan ML (2019) Big data: new oil or just semantics? A systematic
literature review of big data concepts in marketing. Commun Comput Inf Sci
26. Bollier D, Bollier K (2010) The promise and peril of big data. The Aspen Institute
27. Zheng Z, Zhao Y, Wei J (2013) Big data and its technical challenges. J Softw Eng Appl
28. Lin J, Dyer C (2010) Data-intensive text processing with MapReduce
29. Andrejevic M (2014) The big data divide. Int J Commun 8:1673–1689
30. COM (2019) 168 final “Building trust in human-centric artificial intelligence”. https://eur-
lex.europa.eu/legal-content/EN/ALL/?uri=CELEX:52019DC0168. Accessed 26 July 2021
31. COM (2020) 65 final “White paper on artificial intelligence—a European approach to
excellence and trust”. https://ec.europa.eu/info/sites/default/files/commission-white-paper-
artificial-intelligence-feb2020_en.pdf. Accessed 26 July 2021
32. COM (2020) 66 final “A European strategy for data”. https://eur-lex.europa.eu/legal-content/
EN/TXT/?uri=CELEX%3A52020DC0066. Accessed 26 July 2021
33. COM (2020c) 767 final “Proposal for a regulation of the European Parliament and of the
Council on European data governance (Data Governance Act)”. https://eur-lex.europa.eu/
legal-content/EN/TXT/?uri=CELEX%3A52020PC0767. Accessed 26 July 2021
AI and Blockchain for Secure Data Analytics 79

34. ENISA (2015) Privacy by design in big data. An overview of privacy enhancing technologies
in the era of big data analytics. www.enisa.europa.eu. Accessed 26 July 2021
35. Favaretto M, De Clercq E, Elger BS (2019) Big data and discrimination: perils, promises and
solutions. A systematic review. J Big Data 6(1):12
36. van Brakel R (2016) Pre-emptive big data surveillance and its (dis)empowering consequences:
the case of predictive policing. In: van der Sloot B, Broeders D, Schrijvers E (eds) Exploring
the boundaries of big data. Amsterdam University, Amsterdam, pp 117–141
37. Gupta M, Dey S, Sharma SK, Singh J, Chang V (2020) AI and blockchain-enabled privacy-
preserving big data analytics for smart healthcare. J Big Data 7(1):1–21
38. Cai Q, Zhu L, Peng W (2020) Blockchain and AI in edge computing: opportunities and
challenges. IEEE Netw 34(5):126–133
39. Nguyen DT, Kim YH, Kim S (2019) AI-based blockchain technology for secure IoT
communications. IEEE Access 7:101553–101562
40. Lemieux VL (2020) Democratizing AI and blockchain: are we headed towards a distributed
utopia or digital feudalism? Proc IEEE 108(3):440–451
41. Suo H, Wan J, Zhang C, Liu J (2020) A survey on blockchain and AI technologies for
healthcare. IEEE Access 8:121034–121054
42. Yazdavar AH, Li S, Fathy M (2019) Blockchain for AI-enabled manufacturing. Procedia
CIRP 84:830–835
43. Park S, Yang JY, Lee J (2021) Blockchain meets AI: challenges and opportunities. J Commun
Netw 23(1):1–11
44. Talwar R, Chatterjee S, Rana S, Raghavendra R (2020) Towards an AI and blockchain enabled
integrated edge-cloud computing framework. Futur Gener Comput Syst 111:1045–1059
45. Sang J, Wan J, Chen C, Wang S (2020) A review of blockchain technologies for different IoT
applications. J Netw Comput Appl 168:102675
46. Lam WWY, Lam JSL (2020) Blockchain and AI-enabled healthcare applications: a systematic
review. Healthcare 8(3):232
47. Maheshwari H, Chandra U, Yadav D, Gupta A, Kaur R (2023) Machine learning and
blockchain: a promising future. In: Proceedings of the 2023 4th international conference
on intelligent engineering and management (ICIEM)
48. Nakamoto S (2008) Bitcoin: a peer-to-peer electronic cash system. https://bitcoin.org/bitcoin.
pdf
49. Swan M (2015) Blockchain: blueprint for a new economy. O’Reilly Media
50. Buterin V (2013) Ethereum white paper: a next-generation smart contract and decentralized
application platform. https://github.com/ethereum/wiki/wiki/White-Paper
51. Narayanan A, Bonneau J, Felten E, Miller A, Goldfeder S (2016) Bitcoin and cryptocurrency
technologies: a comprehensive introduction. Princeton University Press
52. Yli-Huumo J, Ko D, Choi S, Park S, Smolander K (2016) Where is current research on
blockchain technology? A systematic review. PLoS ONE 11(10):e0163477
53. Bechini A, Cioni A, Gori A (2019) Blockchain as a service for data privacy management in
IoT scenarios. IEEE Internet Things J 6(2):2625–2637
54. Qin J, Strinati EC (2020) Toward user-centric identity management systems with blockchain.
IEEE Trans Dependable Secure Comput 17(6):1339–1353
55. Sharma PK, Zhang H, Li Z (2019) A privacy-preserving framework for EHR using blockchain
and IPFS. J Med Syst 43(5):133
56. Pawani H, Desai A, Wang S (2019) A blockchain based data privacy and integrity preservation
framework for smart grid communication networks. J Netw Comput Appl 126:85–97
57. Garg D, Chaudhary P, Sharma P (2020) Blockchain based privacy-preserving healthcare
system. Sustain Cities Soc 54:101945
58. Zhang Q, Wen Y, Sun X (2019) Blockchain-based privacy-preserving data sharing scheme
for internet of vehicles. IEEE Internet Things J 7(2):1194–1204
59. Fang Y, Kanellopoulos D, Xu Y, Ramachandran M (2019) A blockchain-based approach for
enhancing data privacy in vehicle-to-grid networks. IEEE Trans Intell Transp Syst 21(9):3720–
3732
80 S. M. Sabharwal et al.

60. Dutta R, Chatterjee S, Nigam N (2020) Secure data sharing using blockchain for privacy-
preserving in social IoT. J Netw Comput Appl 149:102474
61. Fernandes CA, Ferreira R (2020) Enhancing privacy for IoT and mobile devices through
blockchain. J Netw Comput Appl 168:102711
62. Liu Y, Huang X (2020) Blockchain-based efficient and privacy-preserving user revocation
scheme for fog computing. IEEE Trans Industr Inf 17(5):3488–3498
63. Rashidi B, Gharibzadeh S (2020) A blockchain-based secure data sharing framework for
smart city applications. Futur Gener Comput Syst 111:843–855
64. Elgamal MA, Elmahdy N (2020) A secure and decentralized data sharing model for cloud-
based healthcare systems using blockchain technology. Int J Inf Manag 50:416–427
65. Zhang Q, Zhang Y, Shi W (2019) A survey of blockchain-based secure sharing of big data.
IEEE Access 7:151485–151497
66. Ghosh A, Banerjee T, Ruj S (2019) Secure data sharing in a consortium blockchain network.
Futur Gener Comput Syst 98:542–555
67. Keshavarz-Haddad A, Mousavi SM (2019) A secure and dynamic data sharing mechanism
for cloud storage using blockchain. J Netw Comput Appl 147:52–62
68. Li Z, Yuan Y (2019) PFS: A privacy-aware data sharing scheme based on blockchain for IoT
environments. IEEE Internet Things J 6(4):6548–6558
69. Ruj S, Mukherjee S (2020) Secure data sharing in consortium blockchains: a survey. J Parallel
Distrib Comput 140:35–53
70. Song C, Lee J (2019) Trust-free collaborative data sharing via blockchain in industrial Internet
of Things. IEEE Trans Industr Inf 16(6):4102–4110
71. Tang Y, Wang R, Xie D (2019) A decentralized data sharing scheme based on consortium
blockchain for IoT. IEEE Internet Things J 6(3):4137–4145
72. Yue X, Wang H, Jin D (2019) FSSDS: a flexible and secure searchable data sharing scheme
over cloud storage. Futur Gener Comput Syst 91:341–351
73. Javaid U, Li Z, Aman MN, Shao D, Yee K, Sikdar B (2022) Blockchain based secure group
data collaboration in cloud with differentially private synthetic data and trusted execution
environment. In: Proceedings of the 2022 IEEE international conference on big data (big
data)
74. Dong L, Zhao J, Chen T, Yu Y, Duan Z, Zhu J (2022) The secure data sharing and interchange
model based on blockchain for single window in trade facilitation. In: Proceedings of the 2022
international conference on blockchain technology and information security (ICBCTIS)
75. Cheng X, Zhang X, Zhang H (2019) Blockchain-based secure data sharing scheme for IoT
with fine-grained access control. J Netw Comput Appl 136:1–12
76. Qu Y, Zhang H, Li L (2020) Secure and privacy-preserving data sharing in industrial Internet
of Things based on consortium blockchain. Futur Gener Comput Syst 103:532–541
77. Liu L, Yu S, Wang L (2019) Secure data sharing scheme based on consortium blockchain for
industrial Internet of Things. J Netw Comput Appl 133:13–21
78. Dinh TTA, Liu D, Zhang M (2020) A blockchain-based framework for secure and efficient
data sharing in industrial IoT. IEEE Trans Industr Inf 16(2):1103–1111
79. Hengameh M, Hosseinzadeh M, Bozorgi M (2019) A secure and efficient data sharing scheme
for IoT using blockchain and smart contracts. Futur Gener Comput Syst 98:489–500
80. Konečný J, McMahan HB, Yu FX, Richtárik P, Suresh AT, Bacon D (2016) Federated learning:
strategies for improving communication efficiency. arXiv:1610.05492
81. McMahan HB, Moore E, Ramage D, Hampson S, Arcas BA (2017) Communication-efficient
learning of deep networks from decentralized data. In: Artificial intelligence and statistics
(AISTATS)
82. Smith V, Chiang K, Sanjabi M, Recht B (2017) Federated multi-task learning. In: Advances
in neural information processing systems (NeurIPS)
83. Kairouz P, McMahan HB, Avent B, Bellet A, Bennis M, Bhagoji AN, Singh V (2019) Advances
and open problems in federated learning. arXiv:1912.04977
84. Hard A, Rao V, Mathews A, Ramachandran B, Beutel A, Smelyanskiy M, Agrawal S (2018)
Federated learning for mobile keyboard prediction. arXiv:1811.03604
AI and Blockchain for Secure Data Analytics 81

85. Yang Q, Liu Y, Chen T, Tong Y (2019) Federated machine learning: Concept and applications.
ACM Trans Intell Syst Technol (TIST) 10(2):1–19
86. Bonawitz K, Eichner H, Grieskamp W, Huba D, Ingerman A, Ivanov V, Velez F (2019)
Towards federated learning at scale: system design. arXiv:1902.01046
87. Smith AJ, Chiang K, Sanjabi M, Recht B (2017) Federated multi-task learning. In: Advances
in neural information processing systems (NeurIPS)
88. Wang Z, Liu Z, Zhu L, Ma Y (2018) Secure federated transfer learning. In: Proceedings of
the 24th ACM SIGKDD international conference on knowledge discovery and data mining
89. McMahan B, Moore E, Ramage D, Hampson S, Agüera y Arcas B (2017) Communication-
efficient learning of deep networks from decentralized data. In: Artificial intelligence and
statistics
90. Castellanos JAF, Coll-Mayor D, Notholt JA (2017) Cryptocurrency as guarantees of origin:
simulating a green certificate market with the Ethereum blockchain. In: Proceedings of the
2017 IEEE international conference on smart energy grid engineering (SEGE)
91. Abognah A, Basir O (2022) Distributed spectrum sharing using blockchain: a hyperledger
fabric implementation. In: Proceedings of the 2022 IEEE 1st global emerging technology
blockchain forum: blockchain and beyond (iGETblockchain)
92. Cheung J (2021) Real estate politik: democracy and the financialization of social networks. J
Soc Comput
93. Zhang H, Xu C, Huang C (2020) Tokenomics in blockchain: a comprehensive survey. IEEE
Access 8:188689–188706
94. Han T, Kim Y (2019) A survey on token economics: design principles and blockchain
applications. Sustainability 11(10):2932
95. Li Z, Ning H (2020) Token economics in blockchain networks: a survey. IEEE Commun Surv
Tutor 22(4):2877–2895
96. Kumar R, Dhiman G (2020) Incentive mechanisms for token-based blockchain networks: a
survey. In: Blockchain and Internet of Things: proceedings of IIoTS 2019. Springer, Singapore,
pp 105–115
97. Zhou M, Lu R (2021) A survey on token economy in blockchain-based crowdfunding. IEEE
Access 9:87175–87192
98. Xu Y, Huang X, Liu L, Zhang J (2019) Blockchain-based decentralized AI: a survey. arXiv:
1908.00700
99. Zhang H, Li Y, Zhang Y (2020) Integrating Blockchain and Artificial Intelligence: challenges
and Opportunities. IEEE Intell Syst 35(4):92–96
100. Kuo TT, Kim HE, Ohno-Machado L (2017) Blockchain distributed ledger technologies for
biomedical and health care applications. J Am Med Inform Assoc 24(6):1211–1220
101. Bouhlel MS, Ben Amor N, Ben Ahmed M (2018) Challenges and opportunities of blockchain-
based artificial intelligence in healthcare. In: 2018 3rd international conference on image,
vision and computing (ICIVC). IEEE, pp 1–5
102. Tsai CW, Lai CF (2019) Applying blockchain in securing Internet of Things. IEEE Internet
Things J 7(5):4502–4510
103. Tsang DH, Gilder A (2021) Blockchain and the general data protection regulation (GDPR).
Comput Law Secur Rev 40:105386
104. De Filippi P, Wright A (2018) Blockchain and the law: the rule of code. Harvard University
Press
105. Choudhury OR, Sarker MH (2021) Legal and regulatory issues in blockchain technology:
a comprehensive review. In: Handbook of blockchain, digital finance, and inclusion, vol 2.
Springer, Cham, pp 129–149
106. Henderson MA, King DL (2018) The potential and limits of blockchain for state taxation.
Natl Tax J 71(4):725–752
Synergizing Artificial Intelligence
and Blockchain

Priyanka Tyagi, Neelam Shrivastava, Sakshi, and Vishal Jain

Abstract This chapter provides a comprehensive examination of the confluence

of Artificial Intelligence (AI) and Blockchain, which are two of the most notable
technical advancements in our day. This explores the distinct methods in which the
powerful data processing skills of AI and the solid security features of Blockchain
might be merged to transform different sectors. The combination of AI and
Blockchain is examined through real-world implementations in healthcare to enhance
patient care and safeguard data, in supply chain management to enhance transparency
and efficiency, in finance to detect fraud and ensure secure transactions, and in
legal systems to automate and secure contract management. The chapter delves into
the obstacles and potential future developments of integrating AI with Blockchain.
It highlights the possibility of AI to enhance the intelligence and autonomy of
Blockchain systems, while highlighting the role of Blockchain in guaranteeing the
trustworthiness and consistency of AI applications. The integration of these methods
has the potential to develop advanced, protected, and streamlined systems, hence
enabling the emergence of inventive solutions in data administration and industrial
processes.

P. Tyagi · V. Jain
Department of Computer Science and Engineering, Sharda University, Greater Noida, Uttar
Pradesh, India
N. Shrivastava
Department of Computer Science and Engineering, Hi-Tech Institute of Engineering and
Technology, Ghaziabad, Uttar Pradesh, India
Sakshi (B)
Department of Computer Science and Applications, Sharda University, Greater Noida, Uttar
Pradesh, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 83
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_4
84 P. Tyagi et al.

1 Introduction to AI and Blockchain

Artificial Intelligence (AI) and Blockchain are two revolutionary technologies that
are transforming industries and reshaping the way we interact with the digital world.
AI encompasses the emulation of human intelligence in machines, enabling them to
learn, reason, and make decisions. On the other hand, Blockchain is a decentralized
and tamper-resistant digital ledger that ensures transparency and security in trans-
actions. While they serve distinct purposes, the convergence of AI and Blockchain
holds the potential to unlock new frontiers of innovation [1]. AI, with its subsets like
Machine Learning and Deep Learning, has shown its prowess in tasks ranging from
image recognition to natural language processing. By analyzing massive datasets
and learning patterns, AI systems can make predictions and decisions that were
once reserved for human experts. Additionally, AI-powered algorithms are driving
automation, efficiency, and personalization across various industries.
Blockchain, known for its secure and transparent nature, is disrupting industries
like finance, supply chain, and healthcare. It eliminates the need for intermediaries,
reduces fraud, and ensures data integrity through its decentralized architecture. Trans-
actions recorded on the Blockchain are immutable and traceable, fostering trust in
digital interactions. The synergy between AI and Blockchain brings forth exciting
possibilities. Blockchain can enhance the security and privacy of AI applications
by safeguarding data and ensuring transparent data usage. Simultaneously, AI can
empower Blockchain networks by optimizing processes, analyzing data, and enabling
predictive insights. In sectors like healthcare, the combination of AI and Blockchain
can facilitate secure sharing of patient data for research while maintaining privacy
[2]. Supply chains can leverage AI’s predictive capabilities to optimize logistics,
while Blockchain ensures the authenticity and traceability of products. The inter-
section of AI and Blockchain also paves the way for decentralized AI marketplaces,
where individuals can access and contribute AI services
without intermediaries. This democratization of AI could fuel innovation and
empower a broader range of stakeholders. As AI and Blockchain continue to evolve,
their convergence holds the promise of creating more robust, transparent, and intelli-
gent systems. This introduction scratches the surface of these complex technologies,
which individually have already reshaped industries, and together have the potential
to reshape our digital landscape even further.

2 Foundations of AI and Blockchain

AI Techniques and Their Relevance in Data Analysis:

Artificial Intelligence (AI) encompasses several techniques that enable machines
to learn from data and perform tasks that mimic human cognitive abilities. Three
prominent techniques within AI—Machine Learning, Deep Learning, and Natural
Synergizing Artificial Intelligence and Blockchain 85

Language Processing (NLP)—are integral to data analysis, as they extract insights

from vast and complex datasets, enabling informed decision-making.
Machine Learning (ML): Machine Learning encompasses the utilization of algo-
rithms enabling computers to glean knowledge from data sans explicit program-
ming. It encompasses supervised learning, unsupervised learning, and reinforcement
learning. Significance in Data Analysis: ML algorithms are instrumental in catego-
rizing data, making projections, and uncovering patterns within datasets. They have
the capacity to discern trends, segment data, and prognosticate results, rendering
them indispensable for tasks such as customer categorization, fraud identification,
and anticipatory upkeep [3].
Deep Learning (DL): Deep Learning constitutes a subset of Machine Learning that
employs artificial neural networks with numerous layers (deep neural networks).
It excels in processing extensive volumes of unstructured data, including images
and textual content. Significance in Data Analysis: DL is especially adept at tasks
such as image recognition, speech understanding, and natural language processing. It
autonomously acquires intricate data features, thereby facilitating precise recognition
of objects, sentiments, or contextual information.
Natural Language Processing (NLP): NLP entails the interaction between
computers and human language. It empowers machines to comprehend, construe,
and generate human language, thereby fostering seamless communication between
humans and computers. Significance in Data Analysis: NLP is pivotal in the scrutiny
of unstructured textual data, encompassing social media posts, customer feedback,
and documents. Sentiment evaluation, identification of named entities, and topic
modeling represent a subset of applications that extract meaningful insights from
textual data, aiding in the comprehension of customer sentiments and trends.

2.1 Relevance of AI Techniques in Data Analysis

1. Data Pattern Recognition: Machine Learning and Deep Learning can identify
complex patterns in large datasets. This helps in making accurate predictions
and uncovering hidden insights, such as customer preferences, market trends,
and potential risks.
2. Image and Video Analysis: Deep Learning excels in image and video analysis
tasks. It enables object detection, facial recognition, and even autonomous driving
by processing visual data (Fig. 1).
3. Text Analysis: NLP techniques allow machines to understand and analyze text
data. Sentiment analysis helps gauge customer opinions, while text summariza-
tion condenses lengthy content for quick insights.
4. Personalization and Recommendation: Machine Learning algorithms power
recommendation systems that suggest products, services, or content based on
user behavior and preferences, enhancing user experiences.
86 P. Tyagi et al.

Fig. 1 Relevance of AI techniques in data analysis

5. Anomaly Detection: Machine Learning models can identify anomalies in data,

aiding in fraud detection, network security, and quality control.
6. Predictive Maintenance: Machine Learning algorithms predict equipment
failures by analyzing sensor data, minimizing downtime and maintenance costs.
7. Medical Diagnosis: Machine Learning and Deep Learning are used to analyze
medical images, assisting doctors in diagnosing diseases like cancer from scans.
8. Language Translation: NLP techniques enable real-time language translation,
breaking down language barriers and facilitating global communication.

2.2 Blockchain: Components and Principles of Decentralized

Consensus

Blockchain is a transformative technology that underpins various digital systems

with its decentralized, secure, and transparent nature. It is a distributed digital ledger
that records transactions in an immutable and chronological manner. Below is an
outline of its components and the principles of decentralized consensus:
Blockchain Components:
1. Blocks: Transactions are organized into blocks, which contain information like
transaction details, timestamps, and a reference to the previous block.
Synergizing Artificial Intelligence and Blockchain 87

Fig. 2 Components of block

chain

2. Chain: Blocks are interconnected in chronological order, creating a chain. The

cryptographic hash of each block is used to connect it securely to the preceding
block, ensuring resistance to tampering.
3. Decentralization: Instead of relying on a central authority, blockchain operates
on a network of nodes (computers) that collaboratively validate transactions and
uphold the integrity of the ledger (Fig. 2).
4. Cryptography: Transactions and blocks are secured using cryptographic tech-
niques. Each block’s hash depends on the data within the block and the previous
block’s hash, making it computationally infeasible to alter a block without
affecting subsequent blocks.

2.3 Principles of Decentralized Consensus

Principles of Decentralized Consensus form the bedrock of blockchain technology,

reshaping the way trust and agreement are achieved in digital systems. At the core of
decentralized consensus lies the principle of eliminating reliance on central authori-
ties by distributing decision-making across a network of nodes. These nodes collab-
orate to validate transactions, ensuring that all participants agree on the state of the
shared ledger without the need for intermediaries. This consensus is achieved through
mechanisms like Proof of Work (PoW) or Proof of Stake (PoS), where participants
compete or are chosen based on their computational effort or stake in the network.
This approach not only enhances security by removing single points of failure but also
fosters transparency through the verification of transactions by multiple parties. By
88 P. Tyagi et al.

leveraging cryptographic techniques, decentralization, and the immutable nature of

the blockchain, decentralized consensus establishes a trustless system where interac-
tions can occur securely and transparently, revolutionizing industries and challenging
traditional models of governance and authority. In essence, blockchain’s decentral-
ized consensus principles revolutionize how trust is established, transactions are
conducted, and data is secured. Its transparent, tamper-resistant, and trustless nature
has applications across finance, supply chain management, healthcare, and beyond,
offering new paradigms for digital interactions [4].
Principles of Decentralized Consensus are at the heart of blockchain technology,
redefining trust and cooperation in digital ecosystems. Instead of relying on a central
authority, these principles establish a network-wide agreement mechanism, ensuring
the integrity of transactions and data. Through consensus mechanisms like Proof
of Work (PoW) or Proof of Stake (PoS), nodes within the decentralized network
collaborate to validate transactions and create new blocks. This decentralized vali-
dation process not only enhances security by eliminating single points of failure but
also fosters transparency, as each participant can independently verify the correct-
ness of transactions. Immutability, another key principle, ensures that once data is
added to the blockchain, it becomes tamper-resistant and practically immutable due
to cryptographic linkage and computational difficulty [5]. These principles empower
blockchain networks to operate trustlessly, allowing participants to interact and
transact without requiring explicit trust in one another, thereby ushering in a new era
of secure, transparent, and autonomous digital collaboration.

3 Challenges in Secure Data Analytics

Challenges in Secure Data Analytics encompass a spectrum of issues that impede the
effective utilization of data for valuable insights while maintaining robust security
measures. These challenges highlight the complexities involved in reconciling data
analysis and privacy in today’s digital landscape. Here are the outlined challenges:
• Data Privacy Concerns and Regulatory Issues:
As data becomes a valuable commodity, concerns over the privacy of personal and
sensitive information have escalated. Striking a balance between data utility and indi-
vidual privacy is crucial. Regulations like the General Data Protection Regulation
(GDPR) and the California Consumer Privacy Act (CCPA) impose strict require-
ments on how data is collected, stored, and used. Meeting these regulatory demands
without compromising analytical potential is a significant challenge [6]. Data privacy
concerns and regulatory issues have become paramount as the digital landscape
evolves, requiring organizations to navigate intricate legal frameworks while safe-
guarding individual privacy. Here are key notes on the challenges and considerations
within this domain (Fig. 3):
Synergizing Artificial Intelligence and Blockchain 89

Fig. 3 Challenges in secure data analytics

1. Proliferation of Personal Data: The exponential growth of data collection,

driven by IoT devices, social media, and digital transactions, has led to vast
repositories of personal information. This includes sensitive data like financial
details, medical records, and biometric information.
2. Privacy Breaches and Identity Theft: High-profile data breaches have exposed
millions of individuals to identity theft, financial fraud, and reputational damage.
The mishandling of personal data can lead to severe consequences for both
individuals and organizations.
3. Global Regulatory Landscape: Various regions have introduced stringent data
protection regulations to address these concerns. Notably, the General Data
Protection Regulation (GDPR) in the EU and the California Consumer Privacy
Act (CCPA) in the US have set a precedent for data privacy rights and transparency
[7].
4. Individual Rights and Consent: Regulations emphasize individual rights,
including the right to access, rectify, and erase personal data. Organizations
must obtain explicit and informed consent from individuals before collecting
and processing their data.
5. Cross-Border Data Transfers: Regulations restrict the transfer of personal data
to countries with inadequate data protection standards. Adequacy decisions, stan-
dard contractual clauses, and binding corporate rules govern cross-border data
flows.
90 P. Tyagi et al.

6. Data Protection Officers (DPOs): Some regulations mandate appointing DPOs

within organizations to oversee data protection efforts, facilitate compliance, and
act as a point of contact for data protection authorities.

3.1 Security Weaknesses in Conventional Data Analytics

Conventional data analytics systems face a range of cybersecurity threats, including

data breaches, unauthorized entry, and malware assaults. As data analytics workflows
encompass multiple phases of data handling and transmission, each juncture presents
a potential vulnerability. Establishing comprehensive end-to-end security for these
workflows is a formidable task, particularly in light of the ever-evolving landscape of
cyber threats. While conventional data analytics methods are valuable for extracting
insights, they are vulnerable to a spectrum of security weaknesses that could compro-
mise data integrity and confidentiality. Recognizing these vulnerabilities is impera-
tive to implement suitable safeguards for data protection and to uphold the confidence
of stakeholders. The following are key points regarding security vulnerabilities in
conventional data analytics:
Data Breaches: Traditional data analytics workflows encompass various stages of
data gathering, storage, and transmission. Weaknesses at any of these stages can result
in data breaches, where unauthorized parties gain access to sensitive information.
Unauthorized Access: Inadequate access controls and feeble authentication mech-
anisms may permit unauthorized individuals to access data and analytics systems,
potentially resulting in data theft or manipulation.
Malware and Cyberattacks: Traditional data analytics systems may be vulnerable
to malware attacks such as viruses, worms, and ransomware, which can disrupt
operations and compromise data integrity.
1. Insider Threats: Malicious actions by employees or individuals within an orga-
nization, known as insider threats, can expose sensitive data to unauthorized
parties (Fig. 4).
2. Injection Attacks: SQL injection, a common vulnerability, occurs when
attackers manipulate input fields to execute malicious SQL commands in
databases, potentially leading to data leakage or unauthorized access.

3.2 Trust and Transparency Gaps in Data Sharing

Collaborative data sharing is pivotal for holistic insights. However, establishing trust
among different entities to share their data is challenging. Concerns about data
misuse, ownership, and the lack of transparency in how data will be handled hinder
seamless data sharing. Overcoming these trust issues while maintaining transparency
and safeguarding the interests of all parties involved is a complex endeavor [4, 4].
Synergizing Artificial Intelligence and Blockchain 91

Fig. 4 Security vulnerabilities in traditional data analytics

Trust and transparency are essential cornerstones of effective data sharing in the
digital age. However, several challenges and gaps in these areas can hinder the
successful exchange of data among organizations and individuals. Understanding
these issues is crucial for building a more trustworthy and transparent data-sharing
ecosystem. Here are key insights into the trust and transparency gaps in data sharing:
1. Data Ownership and Control: The ambiguity surrounding data ownership and
control often leads to mistrust. Organizations and individuals are concerned
about losing control over their data once it’s shared, fearing potential misuse
or unauthorized access.
2. Data Misuse and Exploitation: Data owners worry that shared data might be
exploited for unintended purposes, leading to adverse consequences. Lack of
transparency about how shared data will be utilized exacerbates these concerns.
3. Lack of Accountability: In some cases, there’s a lack of clarity about who is
responsible for the security and appropriate use of shared data. This absence of
accountability undermines trust.
4. Opaque Data Handling: When data recipients don’t provide insights into how
shared data will be processed, concerns about data integrity, anonymization, and
adherence to privacy regulations arise.
5. Data Quality and Accuracy: The accuracy and reliability of shared data can
be questionable, eroding trust. Without transparency in data collection and
validation processes, recipients may doubt the quality of the shared information.
92 P. Tyagi et al.

4 Synergy Between AI and Blockchain

The synergy between AI and Blockchain represents a dynamic convergence of two

transformative technologies, fostering innovation across various domains. Artifi-
cial Intelligence (AI) empowers machines to learn, reason, and make decisions,
while Blockchain provides decentralized, tamper-resistant data storage [8]. Together,
they create a powerful combination with the potential to revolutionize industries.
AI enhances data analysis on the Blockchain, extracting insights from distributed
ledgers. Blockchain ensures the security and transparency of AI data, addressing
concerns about data privacy and integrity [1]. The decentralized nature of Blockchain
can support AI models’ democratization, enabling individuals to contribute, access,
and monetize AI services without intermediaries. This synergy also bolsters trust
in AI decision-making through transparent and auditable records. The marriage
of AI and Blockchain paves the way for secure, intelligent systems that embrace
transparency, privacy, and empowerment, ushering in a new era of technological
advancement [2]. This synergy is driven by the following key aspects.
Enhancing Data Privacy Through Encryption and Decentralized Control:
Blockchain’s decentralized architecture offers a secure and transparent environment
for data storage and sharing. AI’s encryption capabilities further bolster data privacy
by encoding sensitive information. Together, they provide a robust framework where
data remains confidential, and access is controlled through cryptography and decen-
tralized consensus mechanisms. The collaboration between encryption and decen-
tralized control within the realms of data privacy signifies a robust approach to
safeguarding sensitive information in today’s digital landscape [3]. Encryption, a
technique that transforms data into unreadable code, acts as a shield against unautho-
rized access. When coupled with decentralized control, as facilitated by blockchain
technology, a powerful framework emerges to ensure data privacy.
Encryption serves as a formidable barrier against data breaches and unauthorized
viewing. By converting data into an encrypted format, even if unauthorized entities
gain access, the information remains incomprehensible without the decryption key
[12]. This approach thwarts potential hackers and safeguards confidential data. In
tandem, decentralized control introduced by blockchain technology addresses the
vulnerabilities associated with centralized data storage. Instead of relying on a single
point of control, data is distributed across a network of nodes, each holding a copy
of the encrypted information. Changes to the data require consensus among multiple
nodes, ensuring tamper resistance and reducing the risk of unauthorized alterations.
The synergy between encryption and decentralized control empowers individuals
and organizations to retain ownership and control over their data while enabling
secure sharing. This collaborative approach assures data subjects that their infor-
mation remains confidential, accessible only to authorized parties [5]. It reinforces
privacy compliance and engenders a digital environment where data protection is
paramount, thus fostering trust and accountability in the realm of data privacy.
Synergizing Artificial Intelligence and Blockchain 93

Using AI for Anomaly Detection, Pattern Recognition, and Predictive Anal-

ysis: AI’s cognitive abilities, particularly in anomaly detection, pattern recognition,
and predictive analysis, complement Blockchain’s data integrity. AI algorithms can
monitor Blockchain transactions for unusual patterns, potentially indicating fraudu-
lent activities. Moreover, AI-driven predictive analysis can unveil trends and insights
from the Blockchain’s decentralized data, empowering informed decision-making.
Smart Contracts and Self-Executing Agreements to Automate Data-Sharing
Protocols: Blockchain’s smart contracts enable self-executing agreements with
predefined conditions. This automation streamlines data-sharing protocols, ensuring
compliance and reducing the need for intermediaries. AI can enhance these contracts
by autonomously monitoring and triggering actions based on data inputs, optimizing
processes and reducing human intervention.

5 Applications of AI-Blockchain Integration

The fusion of artificial intelligence (AI) and blockchain technology has ushered
in a wave of transformative applications across diverse sectors. This integration
leverages AI’s robust data analysis capabilities and pattern recognition alongside
blockchain’s inherent security and transparency [6, 7]. One prominent application
lies in supply chain management, where AI-driven sensors capture real-time data on
products, securely documented on the blockchain. This synergy ensures end-to-end
traceability, minimizing fraud and enhancing provenance. In the healthcare realm,
blockchain’s secure data storage complements AI’s diagnostic prowess, enabling
the analysis of medical information while preserving patient privacy. Moreover, the
financial landscape benefits from AI-powered market predictions executed as smart
contracts on blockchain, resulting in heightened transparency and automation within
cryptocurrency transactions. The legal industry embraces this synergy by utilizing
AI to simplify the creation
and execution of complex smart contracts, securely recorded on the blockchain.
Finally, the establishment of tamper-proof digital identities through AI-based veri-
fication combined with blockchain immutability holds promise for secure authen-
tication and streamlined Know Your Customer (KYC) processes. These applica-
tions collectively underscore the potential of AI- blockchain integration to reshape
industries by fortifying security, driving efficiency, and ushering in a new era of
transparency.
Supply Chain Management:
AI Sensors and Real-Time Tracking: Artificial intelligence integrated with sensors
provides real-time tracking of products throughout the supply chain. Blockchain
Security, Data collected by AI sensors is securely recorded on a blockchain, ensuring
94 P. Tyagi et al.

the integrity and immutability of the information [18]. Traceability and Authen-
ticity, this integration guarantees the traceability of products from origin to desti-
nation, enhancing transparency in supply chain operations. Fraud Reduction, by
recording each step on the blockchain, fraudulent activities such as counterfeiting
can be detected and minimized.
Healthcare:
Blockchain for Data Security, Blockchain technology securely stores sensitive
medical data, making it tamper-proof and easily auditable. AI-Driven Diagnosis, AI
analyzes medical data, such as diagnostic images and patient records, to provide accu-
rate and timely diagnoses. Patient Privacy, Blockchain’s decentralized and encrypted
nature ensures patient data privacy while allowing authorized parties to access neces-
sary information [10]. Interoperable Data Sharing: Different healthcare providers can
securely share patient data, fostering better collaboration and improving patient care.
Finance and Cryptocurrency:
AI Market Predictions: Artificial intelligence algorithms analyze market trends and
predict future outcomes, guiding investment decisions. Smart Contracts in Finance,
Smart contracts automatically execute financial agreements once predefined condi-
tions are met, minimizing manual intervention and the potential for errors [8].
Blockchain for Cryptocurrency, Blockchain ensures transparency and security in
cryptocurrency transactions, preventing double-spending and unauthorized access.
Automated Transactions, Integration of AI with blockchain facilitates autonomous
execution of transactions based on AI predictions.
Smart Contracts and Legal Industry:
AI-Assisted Contract Creation, AI simplifies the creation of complex contracts by
identifying relevant clauses and legal terminology. Secure Record Keeping, Smart
contracts are recorded on a blockchain, ensuring an immutable and tamper-proof
record of contractual agreements.
Reducing Intermediaries, Smart contracts eliminate the need for intermediaries,
reducing costs and increasing the efficiency of contract execution. Enhanced Contract
Efficiency, The combination of AI and blockchain streamlines contract management
and enforcement, minimizing disputes and delays [9].
Identity Verification:
AI-Driven Identity Verification, AI analyzes biometric and personal data for identity
verification, enhancing accuracy and fraud detection. Blockchain Immutability, Veri-
fied identities are stored on a blockchain, making them tamper-proof and enabling
secure authentication. Applications in KYC and Voting [10]. This technology has
applications in Know Your Customer (KYC) processes and digital voting, ensuring
secure and transparent verification.
Synergizing Artificial Intelligence and Blockchain 95

6 Case Studies

Health Data Analysis Using AI and Blockchain

Background A: healthcare organization aimed to improve patient outcomes through
advanced data analysis while maintaining patient data privacy and security.
Solution:
• Data Collection: AI-powered devices collected patient health data such as vital
signs, activity levels, and medical history.
• Data Analysis: AI algorithms processed the collected data to identify patterns,
correlations, and potential health risks.
• Blockchain Integration: Analyzed data was securely stored on a blockchain,
ensuring tamper-proof records.
• Patient Control: Patients were given control over their data and could grant
permission for research purposes through smart contracts.
• Research Collaboration: Researchers accessed anonymized and authorized data
through the blockchain, accelerating medical research.
Outcome:
• Improved patient care through personalized health insights and early risk
detection.
• Enhanced data security and privacy due to blockchain’s encryption and decen-
tralized nature.
• Efficient research collaboration while respecting patient data ownership.

7 Supply Chain Traceability and Verification

In the complex landscape of supply chains, ensuring the authenticity and traceability
of products has long been a challenge, often resulting in fraudulent activities and oper-
ational inefficiencies. However, a consortium of supply chain stakeholders embraced
a transformative solution by integrating artificial intelligence (AI) and blockchain
technology. By combining these two cutting-edge technologies, they revolutionized
the way products are tracked and verified throughout the entire supply chain [11].
In this case, AI-driven sensors were strategically integrated into products, enabling
real-time data collection as these items traversed the supply chain journey. The
collected data encompassed crucial information such as origin points, transporta-
tion conditions, and storage environments. This data was then securely recorded on
a blockchain, creating an immutable and tamper-proof digital ledger.
The integration’s implications extended far and wide. Stakeholders were afforded
an unprecedented level of visibility into the supply chain, enabling them to validate
the authenticity and integrity of each product at any point in its journey [12]. This
transparency not only thwarted the infiltration of counterfeit goods but also cultivated
96 P. Tyagi et al.

a heightened level of trust among all participants. Additionally, the automated process
of collecting and verifying data streamlined supply chain operations, significantly
reducing the delays and errors that had traditionally plagued the process.
From a consumer perspective, the benefits were evident. The capacity to verify the
origin and authenticity of products through a secure and transparent blockchain-based
system significantly bolstered consumer confidence. Consequently, trust in the supply
chain increased, ensuring that consumers received authentic, high-quality products
[13]. This case study underscores the transformative potential of integrating AI and
blockchain in addressing long-standing challenges within supply chain management.
By offering real-time tracking, irrefutable verification, and heightened transparency,
this approach serves as an exemplar of how technology can reshape industries,
enhancing both operational efficiency and consumer trust.

8 Conclusion

The fusion of Artificial Intelligence (AI) and Blockchain technology has initiated a
new era of data analytics security, effectively addressing enduring challenges related
to data privacy, security, transparency, and trust. As the digital landscape continues
to evolve, this collaboration presents a potent framework that not only unlocks valu-
able insights but also guarantees the integrity and confidentiality of the underlying
data that informs these insights. In this chapter, we embarked on a journey through
the domains of AI, Blockchain, and their seamless integration. We delved into the
numerous advantages that arise from merging these technologies and observed how
their individual strengths lay a solid foundation for secure data analytics. From the
precision of AI algorithms to the immutability of Blockchain, the resulting combi-
nation offers a comprehensive solution that resonates across various industries. The
significance of harnessing the potential of AI and Blockchain for secure data analytics
cannot be overstated. The insights derived from AI-driven analytics hold the poten-
tial to steer strategic decision-making processes, to optimize processes and elevate
customer experiences has been a central goal of the data-driven revolution. However,
the increasing frequency of data breaches and mounting privacy concerns have cast
a cloud over these aspirations. Enter Blockchain, with its decentralized structure and
robust cryptographic protections, which guarantee the immutability and transparency
of data at every stage of its existence. By exploring real-world applications across
various sectors, we have witnessed the concrete benefits of this fusion. Whether it’s
ensuring supply chain authenticity or enabling secure healthcare data sharing, the
synergy between AI and Blockchain is reshaping the way industries engage with data
analytics. The confidence instilled by Blockchain’s trustless nature and AI’s analyt-
ical prowess serves as a cornerstone for innovation and collaboration in a data-driven
world.
Yet, this journey is not without its challenges. The intricate dance between AI and
Blockchain brings about computational complexities, scalability concerns, and regu-
latory considerations. These hurdles are not insurmountable but beckon for continued
Synergizing Artificial Intelligence and Blockchain 97

research and innovation. As we gaze into the horizon, we glimpse the emergence of
Federated Learning and privacy-preserving algorithms, hinting at a future where data
can be analyzed without leaving its secure enclave.
In conclusion, the union of AI and Blockchain holds immense promise for the
future of secure data analytics. It heralds a paradigm shift wherein data-driven
insights need not come at the cost of privacy and security. With each passing day,
advancements in both fields pave the way for a more seamless and trustworthy data
analytics ecosystem. This chapter serves as a testament to this exciting journey,
inviting researchers, practitioners, and enthusiasts to explore, collaborate, and propel
us towards a more secure and insightful digital future.

References

1. Wang Q, Su M (2020) Integrating blockchain technology into the energy sector—from theory
of blockchain to research and application of energy blockchain. Comput Sci Rev 37:100275.6
2. Sengupta U, Kim H (2021) Meeting changing customer requirements in foodand agriculture
through the application of blockchain technology. Frontiers in Blockchain
3. Düdder B, Fomin V, Gürpinar T, Henke M, Iqbal M, Janavičienė V, Matulevičius R, Straub
N, Wu H (2021) Interdisciplinary blockchain education: utilizing blockchain technology from
various perspectives. Frontiers in Blockchain
4. Zhang Z, Song X, Liu L, Yin J, Wang Y, Lan D (2021) Recent advances in blockchain and
artificial intelligence integration: feasibility analysis, research issues, applications, challenges,
and future work. Secur Commun Netw, 15. Article ID 9991535. https://doi.org/10.1155/2021/
9991535
5. Xing B, Marwala T (2018) The synergy of blockchain and artificial intelligence. Available at
SSRN: https://ssrn.com/abstract=3225357, https://doi.org/10.2139/ssrn.3225357
6. Taherdoost H (2022) Blockchain technology and artificial intelligence together: a critical review
on applications. Appl Sci 12:12948. https://doi.org/10.3390/app122412948
7. Chamola V, Goyal A, Sharma P, Hassija V, Binh HTT, Saxena V (2022) Artificial intelligence-
assisted blockchain-based framework for smart and secure EMR management. Neural
Comput Appl:1–11. https://doi.org/10.1007/s00521-022-07087-7. Epub ahead of print. PMID:
35310553; PMCID: PMC8918902
8. Turner SW, Karakus M, Guler E, Uludag S (2023) A promising integration of SDN and
blockchain for IoT networks: a survey. IEEE Access 11:29800–29822.https://doi.org/10.1109/
ACCESS.2023.3260777
9. Lin S-Y, Zhang L, Li J, Ji L-l, Sun Y (2022) A survey of application research based on blockchain
smart contract Wireless Netw 28(2):635–690. https://doi.org/10.1007/s11276-021-02874-x
10. Jie X, Zhou W, Zhang S, Jinhua F (2021) A review of the technology and application of deposit
and traceability based on blockchain. J High Speed Netw 27(4):335–359. https://doi.org/10.
3233/JHS-210671
11. Wazid M, Das AK, Park Y (2021) Blockchain-enabled secure communication mechanism for
IoT-driven personal health records. Trans Emerg Telecommun Technol 33:4.https://doi.org/10.
1002/ett.4421
12. Sharma C, Sharma S, Sakshi (2022) Latent DIRICHLET allocation (LDA) based informa-
tion modelling on BLOCKCHAIN technology: a review of trends and research patterns used
in integration. Multimedia Tools and Applications 81(25):36805–36831. https://link.springer.
com/article/10.1007/s11042-022-13500
13. Sakshi, Kukreja V (2023) Machine learning and non-machine learning methods in mathemat-
ical recognition systems: Two decades’ systematic literature review. Multimedia Tools and
Applications 1–70. https://link.springer.com/article/10.1007/s11042-023-16356-z
Blockchain-Based Smart Contracts:
Technical and Usage Aspects

Gulbir Singh

Abstract Blockchains are decentralized, distributed, and immutable ledgers that

can execute contracts on the blockchain, called smart contracts. Blockchain-based
smart contracts have different challenges, opportunities, and applications in various
fields like cryptocurrency systems, real estate, agriculture, logistics, and fire industry.
The codes that can be executed and which operate on top of a blockchain to facilitate,
implement, and enforce agreements between numerous parties in the absence of a
trusted third party are called smart contracts. This chapter uses a taxonomy to classify
different processes, such as comparing and contrasting the strengths and weaknesses
of various methods and identifying research gaps and unanswered questions. The
chapter also outlines some promising research guidelines for blockchain-based smart
contracts and discusses the benefits and challenges. Researchers and practitioners in
the field of blockchain-based smart contracts may find this chapter useful because it
offers an overview of current and emerging developments in the field.

Keywords Blockchain · Smart contracts · Taxonomy · Future directions

1 Introduction

The purpose of this study is to give a complete assessment of the existing literature on
blockchain-based smart contracts from both a technical and an application standpoint.
We make use of a taxonomy to classify the papers in accordance with several criteria,
such as blockchain platforms, smart contract languages, security issues, scalability
solutions, privacy upgrades, and programmability factors, among others [1–3]. We
evaluate and contrast the positives and negatives of a variety of approaches, as well as
uncover research gaps and outstanding issues. We also examine the possible benefits

G. Singh (B)
Assistant Professor, Department of Computer Science and Engineering, Graphic Era Hill
University, Haldwani Campus, Haldwani, Uttarakhand, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 99
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_5
100 G. Singh

and obstacles of pursuing these interesting future research avenues for blockchain-
based smart contracts and offer some future research directions for blockchain-based
smart contracts.
A blockchain is a distributed, decentralized, and unchangeable ledger that is built
of a chain of record collection that is linked cryptographically. Blockchains are used
in cryptocurrencies like Bitcoin and Ethereum. Transactions and events are terms that
are used to refer to the individual records themselves, while blocks are terms that
are used to refer to groups of data. Each individual who is a part of the blockchain
network has a copy of the distributed ledger stored in the account that they use
to access the network. These transactions are added to the distributed ledger only
after passing through a process of verification and agreement with the other parties
involved in the blockchain. Blockchain technology is distinguished by a number of
basic qualities, the most important of which are immutability, decentralization, and
a cryptographic link.
Immutability: The fact that the records of transactions in the ledger, which continue
to be spread across the nodes, are permanent and cannot be modified is what is meant
by the term “immutability.” The immutability of the data stored on a blockchain is a
property that distinguishes it from centralized database systems and raises to a higher
level the level of data integrity that may be maintained on a ledger.
As long as the cryptographic links are in place, it will be difficult to manipulate
the records via computational means.
Decentralization: As a result of blockchain’s decentralization, the network’s partic-
ipants are given equal authority over the distributed ledger. In contrast to centralized
systems that are managed by a reliable third party, the blockchain has a feature
that ensures redundancy, and this sets it apart from other types of systems. Through
decentralization, the service availability can be guaranteed, the risk of failure can be
reduced, and ultimately, the trust in the service may be improved [4–6].
Cryptographic Link: The block is what establishes the chain of integrity throughout
the remainder of the blockchain. It is a link between each record ordered in chrono-
logical order. A blockchain is a decentralized database that logs transactions. During
the validation of each record’s digital signature, hashing techniques and asym-
metric key cryptography are employed. This ensures the record’s authenticity and
prevents manipulation. Any modification to a block or transaction record constitutes
a violation of integrity [7], resulting in the invalidation of both the record and the
block.
In addition, cryptocurrencies enable seamless peer-to-peer payments and dissem-
inate the financial context while emphasizing that they are the first public blockchain
generation. Moreover, cryptocurrencies enable payments to be made between parties
who are not acquainted with one another. Cryptocurrencies are a type of digital
currency that can only be used on the blockchain, which is an online-only distributed
ledger. Bitcoin [8] is the world’s first significant cryptocurrency that facilitates peer-
to-peer financial transactions without the need for trusted third parties such as inter-
national payment channels. Previously, these types of transactions were only possible
Blockchain-Based Smart Contracts: Technical and Usage Aspects 101

through international payment channels. Prior to this juncture, this type of transaction
could only be completed using foreign payment methods.
A third party is not required for the operation of the system, and the verification of
transactions that are committed to the network is taken care of by specialized nodes
known as miners who employ cryptographic techniques.
In centralized systems, the use of a trustworthy third party, like a bank, is required
so that participants who do not trust one another can connect with one another
and send and receive financial transactions with one another. When you rely on a
trustworthy third party, on the other hand, you run the risk of encountering issues
with both your privacy and your data’s security [9], in addition to experiencing
higher transactional expenses. Blockchain technology was created with the intention
of resolving this issue by allowing parties who do not trust one another to reach an
accord over their transactions and communications without a trusted third party. The
blockchain can be viewed as a decentralized database that documents and maintains
the history of all blockchain network transactions [10, 11]. This is one approach
to conceptualize the blockchain. Bitcoin, the first decentralized digital money, was
constructed on top of a technological advancement known as the blockchain, which
acted as the underlying infrastructure for the currency. In addition to its use in the
financial sector, blockchain technology has progressed to the point that it can now
support a wide range of decentralized applications. A number of these applications
depend on the deployment and execution of smart contracts on the blockchain in
order to function properly.
A smart contract [12] is a computer program that encodes the agreement between
parties that do not trust one another and that is carried out in accordance with
certain pre-defined circumstances. This agreement is carried out in accordance with
some pre-defined conditions. As a component of a blockchain transaction, smart
contracts are gaining in popularity, and one or more blockchain systems may be
used to construct or carry out a smart contract [8]. A smart contract may also be
referred to as an intelligent contract. Miners are a special category of participants
in the blockchain network that are responsible for the deployment of new contracts
as well as the execution of contracts that already exist in the system. Miners are
paid for their efforts based on the amount of computational work that is required to
fulfill the contracts. This determines how much compensation miners earn. Currently,
Ethereum and Hyperledger Fabric are the two platforms that have the largest market
share when it comes to providing support for the deployment and execution of smart
contracts.
The usage of smart contracts enables the regulations of the company to be rewritten
as computer programs. As a result of the diverse requirements imposed by various
industries, a number of distinct platforms for smart contracts have recently come into
existence [13].
Every single one of the several platforms for smart contracts comes with its own
one-of-a-kind collection of features that are geared specifically for the use case in
question. To give just one illustration, the primary concentration of effort that has
been put into the development of Ethereum [14] has been on applications that require
tokenization. Almost every platform includes the fundamental components of a smart
102 G. Singh

contract system, including immutable computer code, a decentralized ledger, and a

consensus layer. These components are necessary for the system to implement smart
contracts. The consensus layer is one of these constituents [15, 16]. Figure 1 depicts
the most prominent smart contract systems that are presently available.
In order to achieve its objectives, this study will first identify and categorize all
previous research conducted on the topic of smart contract technology, and then have
other academicians evaluate the relevant prior research. It is essential to note that we
do not endeavor to cover all of the most recent breakthroughs in technical, financial,

Fig. 1 Different smart contract platforms

Blockchain-Based Smart Contracts: Technical and Usage Aspects 103

or political matters that have been disseminated via other channels, primarily the
Internet [17, 18]. This is something we believe should be clarified. The longitudinal
(year after year) characteristics of academic contributions to smart contracts are of
particular interest to us because they will allow us to not only document and evaluate
the growth in research outputs, but also compare the emphasis placed on various
issues across the board. In addition, the purpose of this study is to identify open issues
in smart contracts that, once resolved, will necessitate further research. We came to
the conclusion that the most efficient approach to proceed would be to use the method
of systematic mapping study described in [1] to construct a classification map and
search for relevant papers in the most important scientific databases. We reached this
conclusion after determining that this method would be the most effective means of
achieving our goal. The generated map can be used to obtain a deeper understanding
of both the issues of interest and the gaps that must be filled in terms of future work
[19, 20].
This study makes a contribution to the existing body of research by offering
a comprehensive, up-to-date, and methodical review of blockchain-based smart
contracts. This overview includes both theoretical and practical elements. Addition-
ally, it provides some observations and suggestions for future work that can assist
in advancing the research and development of smart contracts that are based on
blockchain technology.
The remaining parts of this essay are structured as follows: In Sect. 2, we cover
some of the fundamentals of blockchain technology and smart contracts. The methods
that we used to conduct our literature review are discussed in Sect. 3. In Sect. 4, we
will discuss the findings that emerged from our analysis of the relevant literature
in light of our taxonomy. In the fifth section, we address potential future research
directions for smart contracts that use blockchain technology. The final section of
the paper is Sect. 6.

2 Background

We will provide some background knowledge on blockchain technology and smart

contracts in the next section. We discuss the fundamental ideas, characteristics, and
issues associated with these technologies, as well as how they are related to one
another.

2.1 Blockchain Technology

The unchangeable, decentralized, and distributed ledger that blockchain technology

uses to record transactions that take place between several parties in a peer-to-peer
network is known as the blockchain. A blockchain is a distributed ledger that is
made up of a chain of blocks that are connected by cryptographic hashes [19, 21]. A
104 G. Singh

consensus process used by the nodes in the network verifies the validity of a set of
transactions that are contained within each block. The consensus mechanism ensures
that all nodes agree on the validity and sequence of transactions, preventing double
spending as well as malicious attacks. It also ensures that all nodes agree on the
validity and order of transactions. It is impossible to modify or delete a block once
it has been added to the blockchain since doing so would break the chain of hashes,
which renders the blockchain both tamper-proof and transparent [22, 23].
Blockchain technology has several features that make it suitable for various
applications, such as:
• Decentralization: The blockchain network is not governed or controlled by a
centralized authority or an intermediary because there isn’t one. Each node in the
network is capable of independently verifying transactions and taking part in the
network. This lowers the likelihood of there being a single point of failure, as well
as the chances of censorship and corruption [24, 25].
• Immutability: Transactions that are recorded on a blockchain cannot be undone
and are therefore permanent. Because of this, the data that is recorded on the
blockchain is guaranteed to be accurate and trustworthy, and it also provides
auditability and accountability [26, 27].
• Cryptographic link: Cryptographic methods, like as digital signatures and
encryption, are utilized to ensure the safety of all transactions carried out on
the blockchain. This not only protects the transactions from illegal access and
modification, but also guarantees the transactions’ legitimacy and confidentiality
[28].
• Smart contracts: It is possible to pre-program transactions on the blockchain
so that they carry out automatically when certain circumstances are satisfied.
These exchanges are what are known as “smart contracts,” and they are computer
protocols that facilitate, verify, and enforce agreements between many parties.
However, blockchain technology also faces some challenges that limit its
scalability, efficiency, and usability, such as:
• High resource consumption: The majority of blockchains use a consensus mech-
anism, which requires a large amount of both processing power and energy. This
is done to validate transactions and assure the continuous safety of the blockchain
network. As a consequence of this, the operation and upkeep of a blockchain
network incur enormous expenses and have an adverse impact on the natural
environment [29].
• Low throughput: The amount of transactions that may be handled in a single
second by a blockchain network is constrained by the block size as well as the
block time. This leads to low throughput and excessive latency of transactions,
which affects the performance and user experience of blockchain apps.
• Privacy issues: All nodes in the network are able to view all transactions that
take place on the blockchain publicly. Concerns about users’ ability to maintain
the confidentiality of their identities and data are prompted by this development.
Blockchain-Based Smart Contracts: Technical and Usage Aspects 105

In addition, certain transactions might include private or confidential information

that shouldn’t be seen by the public on the blockchain.
• Interoperability issues: Architectures, protocols, standards, and language imple-
mentations might vary widely among different blockchains. Because of this, it is
difficult for them to communicate and share data with other internal systems as
well as with systems that are external to their organization. This makes it more
difficult to integrate blockchain apps in a variety of contexts and ensure that they
are compatible with one another.

2.2 Smart Contracts

• Smart contracts are executable codes that operate atop a blockchain to facilitate,
implement, and enforce agreements between multiple parties in the absence of a
trusted third party. Other programming languages, such as Solidity for Ethereum,
Java for Hyperledger Fabric, and JavaScript for NXT, can also be used to construct
smart contracts. The ability of smart contracts to communicate with one another
and with other data sources, commonly referred to as oracles, enables the execution
of complex logic and actions. The use of smart contracts rather than regular
contracts has a number of benefits, including the following:
• Automation: Smart contracts can execute automatically when pre-defined condi-
tions are met. This reduces the need for manual intervention and human errors
[30].
• Efficiency: Smart contracts can process transactions faster and cheaper than
traditional contracts. This saves time and money for the parties involved.
• Security: Smart contracts are secured by cryptography and the immutability of
the blockchain. This prevents fraud, manipulation, and breach of contract.
• Transparency: Smart contracts are visible and verifiable by all parties on the
blockchain. This ensures trust and accountability among the parties.

However, smart contracts also have some limitations and challenges that need to
be addressed, such as:
• Design flaws: Smart contracts are prone to bugs, errors, or vulnerabilities in their
code or logic. These flaws can lead to unexpected outcomes or malicious attacks
that can compromise the functionality or security of smart contracts [31].
• Legal issues: In the majority of nations, smart contracts are not recognized by the
law and are not enforceable. The use of smart contracts is not governed by any
specific legal framework or law. In addition, there is no process for the resolution
of disputes or any other kind of remedy for smart contract disagreements [32].
• Usability issues: In order to design, deploy, and use smart contracts, one must
first acquire the necessary technical skills and knowledge. For the creation of
smart contracts and their subsequent interactions, there is a dearth of user-
friendly tools and interfaces. In addition, there is a lack of standardization as
106 G. Singh

Table 1 Summarizes the main application contexts and related works

Platforms Major application contexts Related works
Ethereum • Financial • DAI [21, 35]
• Asset trading • Cryptokitties [36]
Corda • Insurance • Energy block exchange [37]
• MonetaGo [20]
Hyperledger fabric • Supply chain management • Everledger diamond blockchain [38]
• Stock trading • IBM food trust [39]
• Trade finance
Stellar • Remittance • StellarX [40]
• Tempo [41]
• TillBilly [42]

well as interoperability between the various platforms for smart contracts [33, 34]
(Table 1).

3 Methodology

In this section, we describe the methodology of our literature review. We explain how
we searched, selected, categorized, and analyzed the relevant papers on blockchain-
based smart contracts.

3.1 Search Strategy

To conduct a comprehensive and systematic literature search, we followed the

PRISMA guidelines. We used the following keywords and their combinations to
search for papers on blockchain-based smart contracts: blockchain, smart contract,
distributed ledger, decentralized application, DApp, Ethereum, Hyperledger Fabric,
NXT, Solidity, security, scalability, privacy, programmability. We searched the
following databases: Google Scholar, Scopus, Web of Science, IEEE Xplore, and
ACM Digital Library. We also searched the references of the retrieved papers to
identify any additional relevant papers.
We applied the following inclusion and exclusion criteria to filter the papers:
• Inclusion criteria: The paper must be written in English; it must be published in a
journal or conference that is subject to peer review; the study’s primary focus must
be on blockchain-based smart contracts, either from a technical or an application
standpoint; and the paper’s publication date must fall between 2015 and 2023.
• Exclusion criteria: The paper is not written in English; the article is not published
in a journal or conference that is subject to peer review; the paper does not focus
on blockchain-based smart contracts or only briefly mentions them; the paper was
Blockchain-Based Smart Contracts: Technical and Usage Aspects 107

published before 2015 or after April 2023; and the paper was published before or
after those dates.
In April of 2023, we carried out a search of the relevant literature and came up
with a total of 3356 papers. Following the application of the criteria for inclusion
and exclusion, we settled on 539 papers for further investigation.

3.2 Categorization and Analysis

A taxonomy that we built ourselves based on the previous research was utilized
by us in order to classify and evaluate the papers that were chosen. The taxonomy
is comprised of six different dimensions: a blockchain platform, a smart contract
language, a security issue, a solution to scalability, an enhancement to privacy, and
an aspect of programmability. Each dimension is broken down into multiple subcat-
egories that each indicate a particular strategy or method that was utilized by the
papers. As an illustration, some of the platforms that fall under the umbrella of
blockchain technology are Ethereum, Hyperledger Fabric, NXT, and others. Vulner-
abilities can be broken down into its component parts, such as code vulnerability,
transaction vulnerability, network vulnerability, and others [43, 44].
We classified each publication by assigning it to one or more subcategories for
each dimension according to the primary contribution or focus that it presented in
the paper. In addition, we retrieved some descriptive information from each of the
papers, such as the title, authors, year of publication, kind of publication (journal or
conference), type of research (empirical or theoretical), type of technique (qualitative
or quantitative), and the primary findings or implications.
In order to provide a concise summary of the distribution and frequency of the
papers across a variety of dimensions and subcategories, we made use of descriptive
statistics. In addition, we utilized thematic analysis in order to recognize the predom-
inant patterns and ideas that surfaced across the papers. We analyzed the various
approaches and methods described in the papers and compared and contrasted their
positive and negative characteristics. In addition to this, we uncovered the study voids
and unanswered questions that must be answered by subsequent investigations.

4 Results

In this section, we will describe the findings of our literature review in accordance
with the taxonomy that we developed. We describe the distribution of the papers
throughout the various dimensions and subcategories, as well as the frequency of
their appearance. In addition, we analyze the papers to determine the most prominent
recurring themes and trends.
108 G. Singh

4.1 Distribution and Frequency of Papers

The breakdown of the papers into their respective publication years is presented in
Fig. 2. Since 2015, we can see that the number of papers on blockchain-based smart
contracts has dramatically increased, hitting a peak in 2022 with 158 publications.
This indicates that the number of papers is expected to continue to climb. This
suggests that both researchers and practitioners are becoming increasingly interested
in and attentive to this topic.
The breakdown of the papers into their respective types of publishing is presented
in Fig. 3. We can observe that the majority of the papers were presented at conferences
(82%) rather than being published in journals (18%). This shows that blockchain-
based smart contracts are still a developing and active study subject that needs addi-
tional investigations that are more rigorous and mature before they can be published
in journals.
Figure 4 illustrates the breakdown of the publications into several categories of
research. It is clear that the majority of the publications (71%) were empirical, while
only 29% of the studies were theoretical. This indicates that research on smart

180
160
140
120
100 Total Paper
Published
80
Year of publication
60
40
20
0
2015 2016 2017 2018 2019 2020 2021 2022 2023

Fig. 2 Distribution of papers by year of publication

Fig. 3 Distribution of
papers by type of publication

Conferences

Journals
Blockchain-Based Smart Contracts: Technical and Usage Aspects 109

Fig. 4 Distribution of
papers by type of research

Empirical
Theoretical

Fig. 5 Distribution of
papers by type of approach

Quantitative
Qualitative

contracts that are based on blockchains is mostly conducted from a more applied
viewpoint, employing experiments, case studies, surveys, or simulations to evaluate
the performance, usability, or application of blockchain-based smart contracts.
The breakdown of the publications into their respective methods is presented in
Fig. 5. It is clear that the majority of the publications (63%) focused on quantitative
research, while just 37% of the papers explored qualitative topics. This suggests that
the quality, efficacy, or efficiency of blockchain-based smart contracts is evaluated
primarily through the use of numerical data, such as metrics, measurements, or
statistics.
The distribution and frequency of the papers can be seen in Table 2, which is
broken down by dimension and subcategory. We can observe that the majority of the
papers (67%) utilized Ethereum as their blockchain platform of choice, followed by
Hyperledger Fabric (16%), NXT (7%), and others (10%). The majority of the research
articles (58%) made use of the smart contract programming language Solidity. This
was followed by Java (15%), JavaScript (8%), and other languages (19%). The most
common type of security issue that was discussed in the papers was code vulnerability,
which accounted for 42% of the total, followed by transaction vulnerability (24%),
network vulnerability (18%), and other types of vulnerability (16%). Sharding was
the scalability approach that was proposed in the studies the most frequently (35%)
followed by off-chain computation (25%), sidechains (20%), and other scalability
solutions (20%). The technique of enhancing privacy with zero-knowledge proof
was utilized by the majority of the papers (40%), followed by encryption (30%),
ring signature (15%), and other privacy enhancement methods (15%). The majority
110 G. Singh

Table 2 Distribution and frequency of papers by dimension and subcategory

Dimension Subcategory Frequency Percentage (%)
Blockchain platform Ethereum 82 67
Hyperledger fabric 20 16%
NXT 9 7%
Others 12 10%
Smart contract language Solidity 71 58
Java 18 15%
JavaScript 10 8%
Others 23 19%
Security issue Code vulnerability 52 42
Transaction vulnerability 30 24%
Network vulnerability 22 18%
Others 19 16%
Scalability solution Sharding 43 35
Off-chain computation 31 25%
Sidechains 25 20%
Others 24 20%
Privacy enhancement Zero-knowledge proof 49 40
Encryption 37 30%
Ring signature 18 15%
Others 18 15%
Programmability aspect Formal verification 55 45
Testing 31 25%
Debugging 18 15%
Others 19 15%

of the publications (45%) focused on formal verification as the primary method for
evaluating programmability, followed by testing (25%), debugging (15%), and other
methods (15%).

4.2 Themes and Patterns

Based on our thematic analysis of the papers, we identified four main themes and
patterns that emerged from our literature review:
• Theme A: Smart contracts built on blockchain technology make it possible to
create new applications and business models in a variety of industries, including
cryptocurrency systems, supply chain management, agribusiness, real estate, and
Blockchain-Based Smart Contracts: Technical and Usage Aspects 111

energy trading, among others. These applications make use of the capabilities and
advantages offered by smart contracts, such as automation, efficiency, safety, and
transparency. On the other hand, they are also subject to a number of difficulties
and constraints, including regulatory uncertainty, user uptake, interoperability,
and so on.
• Theme B: Blockchain-based smart contracts require high-quality code to ensure
their correct functionality and security. However, writing smart contract code is
not easy due to its complexity, novelty, and irreversibility. Therefore, many papers
propose methods and tools to improve the quality of smart contract code, such as
formal verification, testing, debugging, etc.
• Theme C: Blockchain-based smart contracts suffer from scalability issues due
to their high resource consumption and low throughput. Therefore, many papers
propose solutions to enhance the scalability of smart contracts, such as sharding,
off-chain computation, sidechains, etc. These solutions aim to reduce the load on
the main blockchain or increase its capacity without compromising its security or
decentralization.
• Theme D: Blockchain-based smart contracts raise privacy concerns due to their
public visibility and traceability. Therefore, many papers propose techniques to
enhance the privacy of smart contracts, such as zero-knowledge proof, encryption,
ring signature, etc. These techniques aim to protect the identity or data of smart
contract users or transactions without affecting their functionality or verifiability.

5 Future Research Directions

Following the completion of our literature study, we will now present several possible
future research areas for smart contracts that are based on blockchain technology. We
examine the possible benefits as well as the obstacles that may arise from following
these areas, and we offer some suggestions or guidelines for future work.

5.1 Cross-Chain Smart Contracts

Cross-chain smart contracts are smart contracts that can interact with different
blockchains or other external systems. This is one of the future research directions that
will be pursued in order to enable cross-chain smart contracts. The interoperability
and usefulness of blockchain-based systems, such as decentralized exchanges, cross-
border payments, supply chain management, and other similar applications, can be
improved with the help of cross-chain smart contracts. However, cross-chain smart
contracts also present some technical and security challenges, such as how to ensure
the consistency and atomicity of transactions across different blockchains, how to
verify the validity and authenticity of data from external sources, how to prevent
malicious attacks or frauds, and so on. These are just some of the challenges. As
112 G. Singh

a result, future efforts ought to concentrate on building innovative protocols, struc-

tures, and standards with the purpose of making it easier to design, deploy, and carry
out the execution of cross-chain smart contracts.

5.2 Machine Learning-Based Smart Contracts

Integration of machine learning methods with smart contracts, which are smart
contracts that can learn from data and adapt to changing environments, is another
potential study direction. Smart contracts are smart contracts that can learn from data.
Applications built on blockchain technology, such as prediction markets, insurance
policies, recommendation systems, and so on, can have their intelligence and effi-
ciency improved by the use of smart contracts that are powered by machine learning.
Nevertheless, machine learning-based smart contracts face a number of challenges
and limitations. These include the following: how to ensure the privacy and security
of data used for training and inference; how to deal with uncertainty and noise in
data and models; how to balance the trade-off between complexity and performance
of models; how to handle ethical and legal issues related to the outcomes of machine
learning; and so on. As a result, work that will be done in the future should concen-
trate on establishing unique methodologies, tools, and frameworks that will enable
the construction, deployment, and assessment of smart contracts that are based on
machine learning.

5.3 Human-Centric Smart Contracts

The adoption of a human-centric approach to smart contracts, which are smart

contracts that take into consideration the requirements, preferences, and actions of
human users, is the third future research direction that will be pursued. The usability
and general acceptability of blockchain-based applications, such as e-commerce plat-
forms, social networking platforms, gaming platforms, and so on, can be improved
with the help of human-centric smart contracts. However, human-centric smart
contracts also raise a few challenges and questions, such as how to design user-
friendly interfaces and interactions for smart contracts, how to incorporate human
factors and emotions into the logic of smart contracts, how to align the outcomes
of smart contracts with human values and norms, and how to resolve conflicts and
disputes that arise from the execution of smart contracts, among other things. In
light of this, work that will be done in the future should concentrate on establishing
unique concepts, models, and approaches to support the design, implementation, and
evaluation of human-centric smart contracts.
Blockchain-Based Smart Contracts: Technical and Usage Aspects 113

6 Conclusion

In this chapter, we have provided a comprehensive evaluation of the literature on

blockchain-based smart contracts, both from a technical and an application stand-
point. We have provided a taxonomy of the currently available blockchain-based
smart contract solutions, categorized the included research papers, and discussed the
various blockchain-based smart contract-based studies. In addition, we have identi-
fied and analyzed the major themes and patterns that emerged from our review of the
pertinent literature. On the basis of our findings, we have suggested some potentially
fruitful future research avenues for blockchain-based smart contracts and analyzed
their potential benefits as well as potential drawbacks.
According to the findings of our study, blockchain-based smart contracts are an
emerging and rapidly developing research topic that may be used in a variety of
contexts and present a range of opportunities as well as difficulties. Nevertheless,
blockchain-based smart contracts confront a number of challenges, both technolog-
ical and usage-related, that need to be addressed by researchers and practitioners in
the field. As a result, we have high hopes that our study will be able to operate as
a helpful reference and guide for subsequent research on smart contracts based on
blockchain technology.

Conflict of Interest There are no competing interests involved in this research.

Funding No funding is available for this research.

References

1. Petersen K, Feldt R, Mujtaba S, Mattsson M (2008) Systematic mapping studies in software

engineering. EASE 8:68–77
2. Taherdoost H (2021) Smart contracts in blockchain technology: a critical review. Information
14(2):117
3. Khan SN, Loukil F, Ghedira-Guegan C, Benkhelifa E, Bani-Hani A (2021) Blockchain smart
contracts: applications, challenges, and future trends. Peer-to-Peer Netw Appl 14(5):2901–2925
4. Nakamoto S (2009) Bitcoin: a peer-to-peer electronic cash system. http://www.bitcoin.org/bit
coin.pdf
5. Buterin V et al (2014) A next-generation smart contract and decentralized application platform.
White Paper 3:37
6. Wang S et al (2018) An overview of smart contract: architecture, applications, and future trends.
In: 2018 IEEE intelligent vehicles symposium (IV). IEEE, pp 108–113
7. Wright A, Filippi PD (2015) Decentralized blockchain technology and the rise of Lex
Cryptographia. Available at SSRN 2580664
8. Alharby M, Moorsel AV (2017) Blockchain-based smart contracts: a systematic mapping study.
arXiv:1710.06372
9. Udokwu C et al (2018) An exploration of blockchain enabled smart-contracts application in
the enterprise. Technical report 2.2:36464.97287
10. Seijas PL, Thompson SJ, McAdams D (2016) Scripting smart contracts for distributed ledger
technology. IACR cryptology ePrint, 1156
114 G. Singh

11. Aggarwal S et al (2019) Blockchain for smart communities: applications, challenges and
opportunities. J Netw Comput Appl 144:13–48
12. Buterin V (2018) A next-generation smart contract and de-centralized application platform.
https://github.com/ethereum/wiki/wiki/White-Paper/
13. Polyzos GC, Fotiou N (2022) Blockchain-assisted information distribution for the internet of
things. In: 2017 IEEE International conference on information reuse and integration (IRI), pp
75–78
14. Wust K, Gervais A (2018) Do you need a blockchain? In: 2018 crypto valley conference on
blockchain technology (CVCBT). IEEE, pp 45–54
15. Crawford M (2017) The insurance implications of blockchain. Risk Manag 64:24
16. Guo Y (2018) WISChain: an online insurance system based on blockchain and DengLu1 for
web identity security. In: 2018 1st IEEE international conference on hot information-centric
networking (HotICN), pp 242–243
17. Bogner A, Chanson M, Meeuw A (2016) A decentralised sharing app running a smart contract
on the Ethereum blockchain. In: Proceedings of the 6th international conference on the internet
of things. ACM, pp 177–178
18. Hans R, Zuber H, Rizk A, Steinmetz R (2017) Blockchain and smart contracts: disruptive
technologies for the insurance market. In: 2017 Americas conference on information systems,
pp 01–10
19. Bartoletti M, Pompianu L (2017) An empirical analysis of smart contracts: platforms, appli-
cations, and design patterns. In: International conference on financial cryptography and data
security. Springer, pp 494–509
20. Peters GW, Panayi E (2016) Understanding modern banking ledgers through blockchain tech-
nologies: future of transaction processing and smart contracts on the internet of money. Banking
beyond banks and money. Springer, pp 239–278
21. Sengupta J, Ruj S, Bit SD (2020) A comprehensive survey on attacks, security issues and
blockchain solutions for IoT and IIoT. J Netw Comput Appl 149:102481
22. Sankar LS, Sindhu M, Sethumadhavan M (2017) Survey of consensus protocols on blockchain
applications. In: 2017 4th international conference on advanced computing and communication
systems (ICACCS). IEEE, pp 1–5
23. Singh A et al (2020) Sidechain technologies in blockchain networks: an examination and
state-of-the-art review. J Netw Comput Appl 149:102471
24. Clack CD, Bakshi VA, Braine L (2016) Smart contract templates: essential requirements and
design options. arXiv:1612.04496
25. Chen L et al (2017) Decentralized execution of smart contracts: agent model perspective and its
implications. In: International conference on financial cryptography and data security. Springer,
pp 468–477
26. Sousa J, Bessani A, Vukolic M (2018) A Byzantine fault-tolerant ordering service for the hyper-
ledger fabric blockchain platform. In: 2018 48th annual IEEE/IFIP international conference
on dependable systems and networks (DSN). IEEE, pp 51–58
27. Xu X et al (2017) A taxonomy of blockchain-based systems for architecture design. In: 2017
IEEE international conference on software architecture (ICSA). IEEE, pp 243–252
28. Marino B, Juels A (2016) Setting standards for altering and undoing smart contracts. In:
International symposium on rules and rule markup languages for the semantic web. Springer,
pp 151–166
29. Norta A (2016) Designing a smart-contract application layer for transacting decentralized
autonomous organizations. In: International conference on advances in computing and data
sciences. Springer, pp 595–604
30. Moyano JP, Ross O (2017) KYC optimization using distributed ledger technology. Bus Inf Syst
Eng 59:411–423
31. Hu Y et al (2018) A delay-tolerant payment scheme based on the Ethereum blockchain. CoRR
abs/1801.10295
32. Macrinici D, Cartofeanu C, Gao S (2018) Smart contract applications within blockchain
technology: a systematic mapping study. Telemat Inform 35(8):2337–2354
Blockchain-Based Smart Contracts: Technical and Usage Aspects 115

33. Luu L (2017) Practical decentralized pooled mining. In: 26th fUSENIXg security symposium
(fUSENIXg security 17), pp 1409–1426
34. Feng Q et al (2019) A survey on privacy protection in blockchain system. J Netw Comput Appl
126:45–58
35. Atzei N et al (2018) SoK: unraveling bitcoin smart contracts. In: Bauer L, Küsters R (eds)
Principles of security and trust. POST 2018. Lecture notes in computer science, vol 10804.
Springer, Cham
36. Manzoor Y (2018) A delay-tolerant payment scheme on the Ethereum blockchain. In: 2018
IEEE 19th international symposium on a world of wireless, mobile and multimedia networks
(WoWMoM). IEEE, pp 14–16
37. Guo Y, Liang C (2016) Blockchain application and outlook in the banking industry. Financ
Innov 2:24
38. Rosner MT, Kang A (2015) Understanding and regulating twenty-first century payment
systems: the ripple case study. Mich L Rev 114:649
39. Hopwood D, Bowe S, Hornby T, Wilcox N (2016) Zcash protocol specification, Tech.
rep. 2016–1.10. Zerocoin Electric Coin Company, Tech. rep
40. Rathore S, Kwon BW, Park JH (2022) Blockseciotnet: blockchain-based decentralized security
architecture for IoT network. J Netw Comput Appl 143:167–177
41. Alphand O (2023) IoTChain: a blockchain security architecture for the internet of things. In:
2018 IEEE wireless communications and networking conference (WCNC). IEEE, pp 1–6
42. Nagothu D et al (2023) A microservice-enabled architecture for smart surveillance using
blockchain technology. In: 2018 IEEE international smart cities conference (ISC2). IEEE,
pp 1–4
43. Zheng Z et al (2018) Blockchain challenges and opportunities: a survey. Int J Web Grid Serv
14(2018):352–375
44. He P et al (2017) Survey on blockchain technology and its application prospect. Comput Sci
44(2017):1–7
An Impact of Cyber Security
and Blockchain in Healthcare Industry:
An Implementation Through AI

Monika Dandotiya and Indrajit Ghosal

Abstract The healthcare sector is intrigued by several aspects of AI and blockchain

technology, including the immutability of data stored in a blockchain. Based on
blockchain and AI Blockchain and AI will speed up clinical and biological research
and develop biomedical and healthcare data ledgers. These assessments are based
on fundamental blockchain and artificial intelligence (AI) capabilities, including
decentralized management, data authenticity, immutable audit trails, resilience,
enhanced security, and—most prominently—the restoration of charterers’ rights.
In this chapter, various blockchain-based solutions are examined, including frame-
works and implementations to gauge the effectiveness of such frameworks, such
as Composer, Docker Container, Hyperledger Fabric, and the Wire shark capture
engine. Current limitations in healthcare systems are also discussed. In addition, this
study suggests a Privacy Integration Algorithm among healthcare workers, aiding
in a simulation of settings to implement hyperledger-based electronic healthcare
record (EHR) distribution system that makes use of chain code idea. Round trip time
(RTT), throughput, with other performance parameters in blockchain networks have
also been adjusted for better outcomes. The suggested solution employs blockchain
to increase efficiency and security in comparison to conventional EHR systems.

Keywords Healthcare · Cyber security · Blockchain · AI · EHR · Hyperledger

fabric · Wireshark

M. Dandotiya (B)
Poornima University, Jaipur, Rajasthan, India
e-mail: [email protected]
I. Ghosal
Brainware University, Barasat, Kolkata, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 117
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_6
118 M. Dandotiya and I. Ghosal

1 Introduction

Blockchain technology has gained popularity recently because of the clinical data’s
strong security method [1]. By automating Bitcoin transactions and providing decen-
tralized, secure, and trusted access to a shared ledger of data, transactions, and
records, blockchain technology works to secure data [2]. Through the use of intel-
ligent contracts, it may also manage member interactions without the aid of a
middleman or other reliable third party. Blockchain algorithms are used to assure
security and secret storage of medical data because healthcare researchers are
concerned about the security, storage, and privacy of clinical data [3] (Fig. 1).
Blockchain helps in the analysis and forecasting of health data. Technology helps
in many different ways, according to research, including evaluating medical supply
files, coming across computerized learning with pharmacologic warnings, success-
fully implementing health and treatment areas, and the potential for reassembling.
Medical experts, healthcare professionals, and payers get fast updates, thanks to
the collected medical data [5]. To take into account the aforementioned aspects,
researchers have used a variety of strategies built on blockchain and AI technologies
[6]. These methods enable the healthcare sector to analyze data at extraordinary rates
while maintaining accuracy and data security, and they provide certain professionals
temporary access to their data. The primary motivation for writing this chapter is to
draw attention to the work that academics have done to inform us about the func-
tion and significance of employing AI Blockchain in EHR systems. The healthcare
industry needs to reform badly, from infectious illnesses to cancer to radiography.
There are various ways to use technology to provide more precise, trustworthy,
and efficient therapies. When making a clinical decision, these treatments can be
precise. A computer program with specific instructions called artificial intelligence
is used to carry out tasks that would typically need human intellect. Algorithms are
programmable rules that are programmed [7]. An algorithm may be continuously
improved via machine learning. Huge amounts of data are used in the improvement
process, which is carried out constantly to allow the algorithm to change to increase
the artificial intelligence’s accuracy. AI can comprehend and interpret language,

Fig. 1 Overall financial loss due to cyber-attacks on the healthcare industry (2010–2022) [4]
An Impact of Cyber Security and Blockchain in Healthcare Industry … 119

recognize things, hear noises, and learn patterns to carry out problem-solving tasks.
The reliability and integrity of patient data are not guaranteed by existing medical
data management system. This is a serious shortcoming. Under existing medical data
management system, which is primarily targeted at medical institutions, there is no
assurance of the validity and dependability of patient data. Medical data loss and
hacking are unavoidable risks that must be recognized, and the data gathered is often
vulnerable to data security breaches, violations of personal privacy, and other issues
[8].
Blocks are immutable and auditable since they are cryptographically connected.
Since the same copy of the ledger is replicated across all network nodes, the highest
level of availability and transparency is achieved. Medical data may be verified
via cryptographic linking, which can also provide a tamper-proof duplicate of it.
Public, private, and consortium blockchains are the three kinds of blockchains that
are now accessible. Anyone can join the network and take part in transactions when
using a public blockchain. Private blockchain, in contrast, limits access without
sufficient identification and verification. Public and private blockchain characteristics
are combined in consortium blockchain. The movement of activities in the blockchain
network is shown in Fig. 2.

1 2 3

Initial Transaction Block Created

for Transaction

6 5 4

Results Broadcasted Miner Solves Cryp-

Within miners tography Puzzles

7 8 9

Miner Add Block to

the Chain Transaction Completed

Fig. 2 Blockchain process

120 M. Dandotiya and I. Ghosal

2 Literature Review

Clinical observatories, Doctors, patients, and labs, make up traditional healthcare

systems. The effectiveness of the current healthcare system can be increased by
integrating a variety of contemporary technologies, including AI, cyber security,
blockchain, edge computing, and cloud computing, to provide the automation, intel-
ligence, security, and low cost computational ecosystem that form the foundation of
healthcare. The following are the main technologies that smart healthcare systems
have included.
Costa et al. propose Sec-Health, a blockchain-based protocol that takes into
account all of the key security and supplementary features outlined in existing rules.
By examining Sec-Health under various attack scenarios and detailing how it resolves
the issues with previous solutions, they demonstrate that it is an appropriate solution
[9].
Ramachandran et al. presents a ground-breaking investigation into the security
enabler function of blockchain in Industry 5.0. We were able to determine the most
important drivers, and prospective applications, and give an institutional vision for
Blockchain-based Industry 5.0 across a wide range of industries by using a subjective
survey approach and study objectives. The poll aims to provide industry practitioners,
academia, and researchers with recommendations for fresh even though solutions in
Industry 5.0 verticals. One of the most promising technologies for increasing the
security of Industrial 5.0 settings is blockchain technology, which has the inherent
properties of absoluteness, chronological order, and integrity in industrial systems.
Few books are recommended that provide an overview of Industry 5.0 applications
that use blockchain technology [10].
Gunanidhi et al. suggested (E-PoW) consensus blockchain approach for IoT-based
healthcare monitoring schemes. The primary issue with healthcare analytics systems
is the patient data. Data privacy necessitates high-quality data security measures and
calls for the finest design technique. Many people in the present situation have failed
to ensure protection for data kept on servers that are accessible to other parties [11].
Rajora et al. discussed several mitigating techniques, such as Machine Learning,
Blockchain, Deep Learning, HLE-SO hybrid, etc., and work is suggested for cyber
attacks seen in the healthcare system, as well as the types of techniques used to
prevent attacks in cyber systems. Even though there is no ideal solution, we should
do everything in our ability to reduce cybercrime to guarantee a secure future for
technology. Most security systems require weaknesses before inappropriate conduct
can be investigated and remembered to prevent a repeat. It is futile to take such a
reactive stance while protecting infrastructure, which includes our healthcare systems
[12].
Zhang et al. provide a thorough examination of the security and privacy threats
and requirements, along with technological solution methodologies and strategies,
with a focus on the security and privacy requirements for medical data sharing using
blockchain. They talk about the characteristics needed for electronic medical data
sharing [13].
An Impact of Cyber Security and Blockchain in Healthcare Industry … 121

Omaretal. suggested an approach that combines decentralized storage and

blockchain technology to increase transparency, improve take holder communica-
tion, and shorten the procurement cycle while preventing price differences and errors.
They suggest leveraging smart contracts on the blockchain to automate the GPO
contracting procedure. They provide a general architecture for the HCSC contracting
process that includes intricate algorithms that show the different interconnections
between HCSC stakeholders [14].
Charla et al. used the straightforward and widely accepted concept of cyber secu-
rity as CIA to identify cyber risk in the context of blockchain. Based on the findings, a
cyber-physical risk-based methodology was created, and blockchain implementation
in particular cases was analyzed in order to find and evaluate the breaches following
this framework. Distributed ledger technology, or blockchain, is still being used
in industries like healthcare, energy, manufacturing, financial services, and retail.
Because of its distinct technological features, blockchain proponents contend that
this technology represents a new architectural foundation for establishing trust [15].
Egala et al. Uncompromising system security and privacy are maintained by the
proposed blockchain-based new architecture, which offers a decentralized HER and
smart-contract-based service automation. To discourse on the deficiencies of the
blockchain-based cloud-centric IoMT healthcare system [16].
Im et al. It is suggested to use PHR based on the consortium blockchain, Hyper-
ledger Fabric. In order to access private data, they also evaluate and compare perfor-
mance in terms of ledger size and transaction delay. Personal health information
should not be accessible to anyone because it is sensitive information. It should only
be accessible to authorized users. Blockchain, a new middleware platform paradigm,
can offer data sovereignty, secrecy, and transparency in addition to traceability [17].

3 Research Gaps

We look at how smart technologies are evolving and the security standards that must
be met before they can be used in the healthcare sector. A description of blockchain
technology is given, along with information on its advantages and potential uses
in healthcare systems. Modular IT solutions have started to appear since the 1970s
when healthcare services first began. This era is referred to as Healthcare 1.0. Due
to a lack of resources, healthcare systems at this time were constrained and did not
integrate with digital systems. In a similar vein, networked electronic devices did not
yet include biomedical equipment that might be integrated. At healthcare companies
at the time, paper-based prescriptions and reports were often employed, which raised
expenses and took more time.
122 M. Dandotiya and I. Ghosal

4 Objectives of the Study

The main objectives of the study aim to:

• To provide decentralized records of financial transactions independent of central-
ized institutions or authorities.
• To improve access to patient medical history, device tracking, hospital assets,
• and prescription databases with a whole life cycle of a device inside blockchain
architecture.
• To change the interoperability of healthcare systems.
• To put forward a system architecture with a process for participants’ access control
policies to ensure patient data privacy and security in the HER system.

5 Research Methodology

Suggested electronic health record system is created using blockchain-based infras-

tructure hyperledger fabric with its sandbox hyperledger composer. The permission-
based open-source distributed ledger technology known as hyperledger is (DLT).
Since all users are aware of one another and the blockchain is permissioned and
consortium-controlled, the network can be completely trusted and secured. Partici-
pants in the EHR sharing system fall into four categories: administrators, patients,
physicians, and lab personnel. The exact administration of a blockchain network. The
certifying authority is asked for admin’s enrollment certificate. The administrator has
complete control over the system, including the ability to write, read, update, and
delete users. Whether doctors, patients, or lab workers are legitimate, the adminis-
trator may provide each participant with the appropriate ID to allow access to the
blockchain network. If a participant’s actions are deemed improper, the administrator
may ban them by posting a message on the Hyperledger blockchain network.
Participants in the EHR sharing system fall into four categories: administra-
tors, patients, physicians, and lab personnel. Of Algorithm, the administration of
a blockchain network is executed precisely. The certify in authority is asked for
admin’s enrollment certificate. The administrator has complete control over the
system, including the ability to write, read, update, and delete users. If doctors,
patients, or lab workers are legitimate, the administrator may provide each participant
with the appropriate ID to allow access to the blockchain network. If a participant’s
actions are deemed improper, the administrator may ban them by posting a message
on the Hyperledger blockchain network as shown in Fig. 3. The healthcare service
provider is associated with patients, doctors, caregivers, microbiological laboratory
people, health service administrators, pharma people, insurance people, and other
clinical authorities as participants of the network. In this scenario of a permissioned
Blockchain e-health system, the patient records and all health-related information
are transparent to all the participants of the network. Hence, a privacy-preserving
Hybrid Encryption Technique is designed to protect the privacy of the data. Since
An Impact of Cyber Security and Blockchain in Healthcare Industry … 123

Hyperledger Fabric Distribution

Blockchain Network

Add Records
Laboratory
(Clinical)
Client Application

Update Provider
Grants Provider
Notes

Membership Certified
Service Provi d- Authority
er API
Patient

Fig. 3 Proposed architecture

300

250

200

150 Number of Transaction

Time(mm)
100

50

0
1 2 3 4 5 6

Fig. 4 Effect of block size on transactions

the proposed system is designed with a privacy-preserving Hybrid Encryption tech-

nique, the data in the e-Health system is not transparent to any participant of the
network (Fig. 4).

5.1 Proposed Algorithm

Step 1: Procedure Admin(patientID,ClinicianID,LabID)

Step 2: While True(do)
Step 3: If Clinician ID is valid then
124 M. Dandotiya and I. Ghosal

Step 4: Add the Clinician to the blockchain Framework. Step 5: Add Clinician
(GN, FID)
Step 6: else
Not exist(Clinician ID)
Step 7: end if
Step 8: if(Lab ID is valid)then
Step 9: Add Lab to blockchain Network
Step 10: Add Lab(Blockchain network,Lab ID) Step 11: end for
Step 12: end procedure
The suggested electronic health record system is created using blockchain-
based infrastructure hyperledger fabric and its sandbox hyperledger composer. The
permission-based open-source Distributed Ledger Technology (DLT) is known as
the Hyperledger Project. The Linux Foundation created it to accommodate a variety
of smart contracts and logic for creating several applications on the blockchain
network. A sandbox called Hyperledger Composer is used to execute and test smart
contracts by visualizing the network. Because all users are aware of one another
and the blockchain is permissioned and consortium-controlled, the network can be
completely trusted and secured. The endorsement policy in the fabric will select
which peer nodes (endorsers) will receive the transaction proposal request.
The HER proposal is made up of a variety of parameters, including the patient
identification data provided by the membership service provider, the transaction
payload containing a list of operations to be carried out, the chain code identifier,
a nonce value (counter or random value) that can only be used once by the user
submitting the proposal, and the transaction proposal identifier as provided in the
algorithm below. The method specifies a chain code in which various transaction
actions carried out by the patient are shown as various functions. The proposal phase
or the endorsement phase is what this stage is known as. Evaluation is essential for
conducting targeted tests of system scalability and performance. Pre-processing is
the first step. The Wire shark pcap file may be used to collect all network traffic. This
scans all network traffic and only looks for TCP messages sent via the Hyperledger
Fabric [18]. All communication inside the framework takes place through the gRPC
protocol, which runs on top of TCP. To view the content, if any of the participants wish
to view and access the content, it is important to get access permission. To get access
permission, the proposed system designs a Patient based access control mechanism
for smooth and automatic access with restricted users. To view the data, the proposed
Patient based access control mechanism defines a set of permissions that contain
access rights and transfer rights. Based on these permissions, the participants of the
provider generate a request for viewing the content. In this way, the provider and the
participants are receiving the permissions. The proposed system then automatically
executes the request without intermediates. The content viewing request is permitted
to only those users who are privileged through the access control mechanism [19].
The access permissions ensured in the proposed system are read, write, read-
write, and delete. The permissions are provided based on the privileges given to the
participants. In the same fashion, the resource owner grants access rights and transfer
rights to the requested user.
An Impact of Cyber Security and Blockchain in Healthcare Industry … 125

The steps intricate in Granting Transfer Rights and Access Rights are as follows:
• Step 1: Resource holder of Health Service Provider 1 grants Access Rights to
Health Service Provider 2.
• Step 2: Participants of Health Service Provide 2 generate a request for accessing
data with the public key.
• Step 3: Resource holder of Health Service Provider 1 receives the public key of
requester 1.
• Step 4: Resource Owner of Health Service Provider 1 grants Transfer Rights to
request participant of Health Service Provider 2
• Step 5: Requester 1 of Health Service Provider 2 can access resource.
Local Database. In this subsection, the details of the local database of the
proposed system are presented. In the proposed system, the original data is stored
in the local database. But the hash values of the original data and the blocks gener-
ated during the hashing process are stored in the Blockchain. Utilization of the
Blockchain repository, for all purposes like local storage and hashing process, auto-
matically reduces the performance [20]. Since the Blockchain is positive, it does not
store the original data; instead, it is reasonably the hash values of the original data
to improve the performance of the proposed system. The proposed system utilizes
a local database for rapid access to the original data and hence a local database is
defined for keeping the data secure. This increases the performance of the proposed
access control system.
Blockchain Node. The proposed system utilizes Blockchain as a distributed
repository for storing the data. It reshapes data and historical access control system
records in the permission Blockchain. On the Blockchain network, each provider is
responsible for managing his Blockchain node. A new technology called blockchain
enables dispersed and decentralized architectures to communicate among a wide
network of unreliable parties. Transparency, immutability, cryptography, and oper-
ational resilience are built-in characteristics of blockchain technology that may
enhance the storage capacity of access control systems, making blockchain tech-
nology more suited to decentralized design. Create, read, update, and delete activi-
ties are carried out by conventional database systems [21]. The Blockchain facilitates
the addition of new transactions rather than updating or deleting existing ones; it is
an append-only dataset range system. This append-only paradigm is applied in the
planned Blockchain-based e-Health system. The suggested Blockchain technology
is also a novel way to organize and keep track of the ownership of access rights. In
this way, blockchain technology can be used to build a secure repository.
126 M. Dandotiya and I. Ghosal

6 Patient-Based Access Control Mechanism in E-health

System

In this section, the design of the Patient based Access Control Mechanism is empha-
sized in the e-health systems. In the e-Health system, considering a situation, there
are four health service providers participating to share the data with permissions. The
healthcare service provider is associated with patients, doctors, caregivers, microbi-
ological laboratory people, health service administrators, pharma people, insurance
people, and other clinical authorities as participants of the network. In this scenario
of a permissioned blockchain health system, the patient record and health-related
formation are transparent to all the participants of the network. Hence, a privacy-pre-
serving Hybrid Encryption Technique is designed to protect the privacy of the data.
Since the proposed system is designed with a privacy-preserving Hybrid Encryption
Technique, the data in the e-Health system is not transparent to any participant of
the network. It is further noted that the health records generated during the consul-
tation process. The access rights could be defined as subscriptions for each health
service and the resource owner of a Hospital defines how many times the clinical
authority of another hospital can participate in the Blockchain network within a spec-
ified time limit. One of the important features of this proposed system is that any
provider can transfer access rights to its internal departments such as microbiological
groups, surgery groups, pharmaceutical groups, doctor groups, and patient groups.
This process of transferring can continue from one provider to another and within
the same provider departments.

7 Results

To be more specific, evaluation is essential for conducting targeted tests of systems

capability and performance. Pre-processing is the first step. The Wire shark pcap file
may be used to collect all network traffic. This examines all network traffic and only
filters out TCP messages sent across the hyperledger Fabric. When the transaction
rate rises during different rounds, the latency time also rises. More businesses and
peers exhibit increasing latency. One of the most popular and extensively studied
applications of blockchain technology in healthcare is electronic medical records
(EMR). The majority of the studies that were chosen for the literature review also
dealt with EMR. Multiple healthcare professionals can safely track patients using an
EMR system that has been properly established. EMR also contributes to accurate
and effective healthcare delivery. Data integrity and security are required by the
nature of EMR systems, which call for the exchange of patient data among numerous
healthcare providers.
The anaconda navigator-based spyder IDE is used for evaluation. It makes use of
the statistical data sketching tool, matplotlib. Moreover, it imports pandas3, which
provides data analysis and manipulation. Patient visits a doctor to get his/her ailment
An Impact of Cyber Security and Blockchain in Healthcare Industry … 127

checked. After a thorough investigation, the doctor provides the prescription which
then is added to the blockchain via a web application.
“Docker” is used for setup and initialization to interact with hyperledger fabric and
composer. Developers and/or system administrators may utilize Docker, an operating
system-level container. It helps develop, deploy, and operate hyperledger-based busi-
container. It enables the developer to compile all requirements and features into a
single container. The hyperledger fabric and composer network may operate within
the container by utilizing Docker. Medical devices can charge for patient data, confirm
that the intended patient is receiving the treatment, and exchange procedural data with
patients and regulators in an anonymous manner. The use of blockchain technology
in the healthcare industry is exciting. Recently notable advancements in medical
science and high-quality medical treatments. It is a commonly used, transparent
distributed digital ledger for transaction monitoring in numerous computers.
The above figure shows the number of clinical trials for different ORGPEERs.
It is acknowledged for having a considerable impact on several sectors and indus-
tries. The existence of this technology addresses problems that cannot be solved by
current methods. Confidence, protection, confidentiality, and data interchange among
many systems are necessary. Blockchain presents the opportunity to approach it in
healthcare in innovative ways.
It introduces the suggested method for sharing electronic health records based on
the blockchain network. As a result, an architecture for an HER sharing system based
on blockchain is suggested. The deployment to various techniques and configurations
for block transactions in the network is depicted in Fig. 5. We employed a network
model of three organizations, each with three peers, for our simulation phase. With
thousand transactions per round at 200, 150, 100, 50, and 250 transactions per second,
the experiment is run with basic writing transactions at varying rates. The experiment
is carried out for the following organizations: 1 Org 1 Peer, 2 Org 2 Peer, and 3 Org 3
Peer. Each round of five calculations includes 1000 transactions at various transaction
rates per second, with the total number of transactions in each round totaling 5000
(Fig. 6).
The number of transactions executed per second is measured in transaction per
second (TPS) as shown in Fig. 4.

Fig. 5 DDoS injection

128 M. Dandotiya and I. Ghosal

Fig. 6 The components that are for med along with spinning up the network

7.1 CPU Utilization

Another important parameter to look at how much CPU power the system uses. It
is the most important operating system number to keep an eye on during the tuning
process. Almost all operating systems show how much time the CPU is being used
by people and by the system itself. These extra statistics make it easier to figure out
what is being done on the CPU.
We employed a network model of three organizations with three peers in the
simulation phase mentioned above. The experiment measured throughput and latency
as the number of organizations, block size, and block time increased.
It has been observed that the latency increases with the scaling up of the network
with the addition of new organizations and peers. On the other hand, latency for read
or query operations is lower compared to write operations.
Block time increase reduces latency, which boosts the efficiency of the network
model and increases throughput. The experiment demonstrates that when block time
for write operations is increased from 250 ms to 2 s, throughput increases by up to
4 × in 250tps.
It is also observed that block size 20 has 50% lower latency than block size
40, increasing network performance. Since the network model’s CPU consump-
tion has remained stable over time, therefore, it is feasible to infer that a decrease
in block size and an increase in block time will lead to a significant decrease in
network latency, enhancing network performance, which may be used as a tool for
performance enhancement.
It may result in all services being consumed on the targeted network. Using a
themed-conserver overloads it, cluttering the network and causing the server to fail.
The malicious code then injected a greater volume of DDoS packets onto the server
sport as depicted.
An Impact of Cyber Security and Blockchain in Healthcare Industry … 129

7.2 Network Latency

Network latency is the amount of time it takes for data to be sent from one node
to another. The lower the latency, the faster the connections between nodes in the
network. Network latency is measured in milliseconds, and a connection is faster
when the value is closer to zero. According to Eq. 1, latency may be determined.

Data Size
Latency = (1)
Band width
The attack codes are injected into the propagated chaos-based medicine to measure
latency. Medical data is shared on an SHA and DES encrypt-based blockchain in
order to ensure the validity of the data. The chaotic medichain network had a lower
latency than a DES or SHA encrypted blockchain for 25, 50, 75, and 100% of the
assaults. SHA and DES have a 14 and 18% greater latency than the suggested chaotic
medichain at all attack levels. Connectivity and efficiency are improved when latency
is reduced. Chaos medichain’s decreased latency confirms its effectiveness.

7.3 Computational Time

The amount of time required by the specified architecture to process the data supplied
by the user is referred to as computation all time. The propagated chaos-based
medichain is injected with assaults in order to examine the computational time. SHA
and DES encrypted blockchains are subjected to the same set of attacks to ensure
the validity of the medical data. It depicts the research findings as a graph. Even
after the assault, the suggested architecture still took 0.7 s to analyze the medical
picture, but SHA and DES both took 0.8 s. It takes between 0.5 and 0.7 s to process
a high-definition image. Consequently, the suggested design, which took 0.7 s to
implement, demonstrates the network’s ability to handle data more quickly. Various
malicious attacks do not affect the behavior of the decentralized network-based chaos
in the proposed architecture medichain, as seen by the extensive network latency and
computing time experiments. It prevents an assault on the medichain and protects
sensitive data from being accessed. In order to protect medical data, it is essential
to do architectural quality tests. The server was injected with malicious code from
a variety of attacks to test the proposed architecture. The results of the experiments
show that the assaults sent to the chaotic medichain network had a lower latency of
25%, 50%, 75%, and 100% than the DES, SHA encrypted blockchain. The proposal
demonstrates a 14% lower latency than SHA and an 18% lower latency than DES
at all attack levels. The proposed approach was shown to be more secure than SHA
and DES encryption-based blockchains in terms of network latency and computing
time. As a result, there is a confidence that the chaotic medichain network presented
in the experiment can handle medically sensitive data. There are two types of EHRs
related to the patient i.e., on mandatory and other credentials which don’t hold much
130 M. Dandotiya and I. Ghosal

sensitivity to the privacy or security of the patient. In the blockchain network, it is

in charge of data storage and communication. The chained block of information,
or set of records, is what creates the EHR Smart Contract. Every transaction that
occurs throughout the whole blockchain network must be verified by the blockchain
management module. The processing module makes sure that the transaction under
examination has adhered to all necessary procedures and has received authentication
from the Summary Contract. The authenticated data or records are transferred to the
databases once all the prerequisites have been met. A framework has been created to
create a Blockchain based distributed platform formed medical records that is safe
and secure. With this strategy, it is easier to keep an eye on the lack of candour among
medical practitioners, and only authorized users are given access to the EHR system.
We concentrated on enhancing the security of health records data, which in turn opens
up new vistas for future searches to satisfy other security needs. For the deployment of
essential applications for any medical purpose, other security models that are more
generalized and address the need for health data security as well as detecting the
untrusted hosts will be useful in the future. In order to construct a predictive model
using conventional AI approaches, data storage, and training are centralized, which
increases computational complexity and compromises individual privacy. A feder-
ated blockchain-enabled system is proposed in this study as a solution to the issue,
to improve the perception of numerous heterogeneous CT scans. It enables secure
and private data exchange between hospitals without compromising patient privacy.
Blockchain-based collaborative healthcare training also takes advantage of feder-
ated learning. This deep learning method has been thoroughly tested and compared
to others in the field. A comparison was made between the proposed algorithm and
that of previous federated deep learning modules powered by blockchain.
Participants could sign up via a client application or SDK, then obtain an enroll-
ment certificate from a Membership Service Provider to send to the certificate
authority. An MSP allows peers to sign off on endorsements and validate incoming
transactions. For development purposes, auth.uidai.gov.in was used to connect to the
host, and uidai provided java helper class was used to develop a client application to
connect Aadhar API. Once verification is successful, Names, dates of birth, gender,
and other identifying information can be retrieved from the Aadhar database. Patients
can utilize the client application to update information such as their blood type, aller-
gies, prescriptions, insurance information, and soon. In addition, the patient could
have the capability to fetch data from the local hospital’s EHR.
The anaconda navigator-based spyder IDE is used for evaluation. It makes use
of the statistical data sketching tool, matlablib. Additionally, it imports pandas3,
which provides data analysis and transformation. The evaluation phase involves the
creation of graphs using the Python3 programming language. For better utilization,
all network IP addresses have been substituted with node names from the Hyperledger
caliper and peer organizational of the evaluation-related transactions, including trans-
action send rate, throughput, latency, organizations, peers, maximum CPU consump-
tion, and memory usage, are taken from the caliper report file of HTML and processed
for transformation. The data is then displayed using different angles in matlablib.
An Impact of Cyber Security and Blockchain in Healthcare Industry … 131

Table 1 Varied block time

Parameter Configuration
measurement
Round 10
Network size 1000 per round
Varied factor Block time
Rate 150–250 tps

The anaconda navigator-based spyder IDE is used for evaluation. It makes use
of the statistical data sketching tool, matlablib. Additionally, it imports pandas3,
which provides data analysis and transformation. The evaluation phase involves the
creation of graphs using the Python3 programming language. The Wireshark tool,
which reads all destination port, sending times, source port, and TCP packets from
the pcap file, also considers network data. For better visualization, all network IP
addresses have been substituted with node names from the Hyperledger caliper and
peer organization. All of the evaluation-related transactions, including transaction
send rate, throughput, latency, organizations, peers, maximum CPU consumption,
and memory used, are taken from the caliper report file of HTML and processed
for transformation. The data is then displayed using different angles in matlablib. A
blockchain network benchmarking tool is called Hyperledger Caliper. It is compat-
ible with several Hyperledger frameworks, including Fabric, Composer, Sawtooth,
Iroha, etc. In this study, the caliper device is used to check and test the system’s
performance as well as its many parameters, such as latency, throughput, CPU and
memory utilization, disc write/read, network I/O, and others, as well as metrics for
the system evaluation (Table 1).
The minimal transaction rates per second used in this case are 25, 50, 75,100,
and 125. With this setup, block size 5 has produced better results with a roughly
1.75 × reduction in latency. Similar to this, an early1.75 × increase in transaction
throughput is seen. Through investigation, it has been found that smaller block sizes
and fewer Transactions Per Second (TPS) lead to better system performance. The
blockchain system performs better when the Transaction Rate (TPS) is higher on a
larger block size. Using Wireshark tcpdump, the network traffic and associated statis-
tics are recorded as the caliper is run on the EHR system. In this, Wireshark is used to
capture the packets during execution and save them to a pcap file. Information on one
patient can be kept in a block linked in a chain. Its benefits include reduced costs,
enhanced transparency, accurate tracking, a permanent ledger, a lack of intrinsic
value, and the fact that it has no physical form and is not governed by a centralized
authority. Because of advantages like real-time updates of shared data, distributed
access with security, distributed encryption for data integrity, patient information
protection, decreased transaction costs, and system performance, this technology is
used in medical applications. Blockchain is a decentralized database with chrono-
logical connections between its data blocks. In the healthcare sector, a wide range
of parties, including medical professionals, hospitals, insurer agencies, etc., need
to handle personal EHRs blockchain cooperatively. Electronic Record Systems are
centralized by design and are proprietary. This indicates that the programming base,
132 M. Dandotiya and I. Ghosal

dataset, and system outputs are all controlled by a single supplier, who also provides
the monitoring tools. It is challenging for centralized systems to win the confidence
of patients, physicians, and hospital administration. This problem is resolved by
open-source, independently verifiable systems. EHR is laying the foundation for
blockchain-based healthcare. It includes the specialty of employing blockchain tech-
nology. As it includes an updated patient medical report, blockchain technology
makes it simple to track public health and identify dangers and patterns in the spread
of any illnesses. This encourages patients around the world to receive adequate
care. A single entity does not own it because it is decentralized and highly secure.
After all, the data is kept using cryptography. The records are kept and preserved
under the organization in the current system. To prevent the patient from accessing
these documents for future comparisons. All records will be lost if the specific
server (database) crashes. The proposed system has been created to address these
issues. Electronic Health Records (EHRs) offer a valuable record-keeping service
that encourages the electronic accessibility of paper-based patient medical records
on the Internet. Patients currently disperse their electronic health records (EHRs)
across many locations as life events occur, which causes the EHRs to shift from
one service provider database to another. Therefore, while the service provider often
retains primary management, the patient may loose custody of the current health-
care data. Patients often find it difficult to freely share these data with researchers
because patient access permissions to EHRs are severely restricted. Compatibility
issues between hospitals and diverse providers. The patient should be able to access
his electronic health records (EHRs) for autonomous management and exchange. The
application of transferring the patient’s electronic health record without a central-
ized authority is the main topic of this chapter. Cryptography hash functions can
be used to ensure security in this situation. The authorized party stores the patient
data in blocks that can be connected in a chain using the private key. Bytes from the
message can be joined with it to generate a digital signature that many nodes accept.
By authenticating the public key image, anyone with a requirement for patient infor-
mation can access the data. The current multi-signature algorithm can access up to
256 bytes and solely uses data as keys. Through the use of images that may be trans-
formed into arrays of bytes, it was suggested in this work to increase the key size
up to 512 bytes in order to address these problems. In terms of an array of bytes,
time complexity, accuracy, throughput, and output hash bytes with multiple node
accessibility, the proposed technique can then be compared to cryptography hash
functions such as SHA224, SHA3-224, SHA256, SHA3-256, SHA384, SHA3-384,
SHA512, SHA3-512, and MD-5.

References

1. Islam SMR, Kwak D, Kabir MH, Hossain M, Kwak K-S (2015) The Internet of Things for
health care: a comprehensive survey. IEEE Access 3:678–708
An Impact of Cyber Security and Blockchain in Healthcare Industry … 133

2. Wang L et al (2010) A wireless biomedical signal interface system-on-chip for body sensor
net- works. IEEE Trans Biomed Circuits Syst 4(2):112–117
3. Marrington A, Kerr D, Gammack J (2016) Management of security issues in wearble
technology. IGI Global, Hershey, PA, USA
4. Khanuja SS, Garg S, Singh IP (2009) Method and apparatus for remotely monitoring the
condition of a patient. U.S. Patent 12 259 905
5. Chan M, Estève D, Fourniols JY, Escriba C, Campo E (2012) Smart wearable systems: current
status and future challenges. ArtifIntell Med 56(3):137–156
6. Clarke R Introduction to data veil lance and information privacy, and definitions of terms.
http://www.rogerclarke.com/DV/Privacy.html. Accessed 10 Oct 2017
7. Vora J et al (2018) Blind signatures based secured e-healthcare system. 2018 International
conference on computer, information, and tele-communication systems (CITS). IEEE
8. Bodkhe U et al (2020) Block chain for industry 4.0: a comprehensive review. IEEE Access
8:79764–79800
9. Costa LD, Pinheiro B, Cordeiro W, Araújo R, Abelém A (2023) SecHealth: a blockchain based
protocol for securing health records. IEEE Access 11:16605–16620
10. Ramachandran KK, Nagarjuna B, Akram SV, Bharani J, Raju AM, Ponusamy R (2023) Innova-
tive Cyber Security Solutions Built on Blockchain Technology for Industrial 5.0 Applications.
2023 International Conference on Artificial Intelligence and Smart Communication (AISC),
Greater Noida, India, 643–650
11. Guanidhi GS, Krishnaveni R (2022) Improved Security Blockchain for IoTbased Healthcare
monitoring system. Second international conference on artificial intelligence and smart energy
(ICAIS). Coimbatore, India, pp 1244–1247
12. Rajora R, Kumar A, Malhotra S, Sharma A (2022) Data security breaches and mitigating
methods in the healthcare system: a review. In: International conference on computational
modelling, simulation and optimization (ICCMSO). Pathum Thani, Thailand, pp 325–330
(2022)
13. Zhang R, Xue R, Liu L (2022) Security and privacy for healthcare Block-chains. IEEE Trans
Serv Comput 15(6):3668–3686
14. Omar A, Jayaraman R, Debe MS, Salah K, Yaqoob I, Omar M (2021) Automating pro cere-
ment contracts in the health care supply chain using Blockchain smart contracts. IEEE Access
9:37397–37409
15. Charla GB, Karen J, Miller H, Chun M (2021) The human-side of emerging technologies and
cyber risk: a case analysis of block chain across different verticals. 2021 IEEE technology and
engineering management conference-Europe (TEMSCON-EUR). Dubrovnik, Croatia, pp 1–6
16. Egala BS et al (2021) Fortified-chain: a Blockchain based framework for security and privacy
assured internet of medical things with effective access control. IEEE Internet Things J
8:11717–11731
17. Im H, Kim KH, Kim JH (2020) Privacy and ledger size analysis for healthcare blockchain. In:
International conference on information networking (ICOIN). Barcelona, Spain, pp 825–829
18. Tekeste T, Saleh H, Mohammad B, Ismail M (2019) IoT for healthcare: ultra low power ECG
processing system for IoT devices. International Publishing, Cham, Springer, pp 7–12
19. Zhang Y, Qiu M, Tsai C, Hassan MM, Alamri A (2017) Health-CPS: Healthcare cyber-physical
system assisted by cloud and big data. IEEE Syst J 11(1):88–95
20. Xu X, Zhang X, Gao H, Xue Y, Qi L, Dou W (2020) BeCome: Blockchain enabled computation
offloading for IoT in mobile edge computing. IEEE Trans Industr Inf 16(6):4187–4195
21. Biswas S, Sharif K, Li F, Mohanty SP (2020) Block chain for E-healthcare systems: Easier
said than done. IEEE Comput 53(7):57–67
Deep Learning and Blockchain
Applications in Healthcare Sector Using
Imaging Data

Monika Sethi, Jatin Arora, Vidhu Baggan, Jyoti Verma, and Manish Snehi

Abstract The healthcare industry has witnessed the emergence of deep learning
(DL) along with blockchain technologies as potent instruments with considerable
potential for transforming the sector, in the domain of imaging analysis of informa-
tion. The healthcare industry is highly significant to society as it is responsible for
preserving and enhancing human health. It includes preventive medicine, diagnostics,
rehabilitation, therapy, and palliative care offerings. In recent years, there have been
notable advancements and improvements in the medical field regarding using images
by implementing DL and blockchain applications. Integrating DL and blockchain
technology in healthcare imaging has presented novel prospects for cutting-edge
research and advancement. Scholars are investigating using DL models to examine
voluminous imaging datasets, thereby revealing valuable insights and patterns that
can facilitate progress in healthcare knowledge and treatment methodologies. The
adoption of blockchain technology in clinical trials contributes to promoting trans-
parency and consistency. This is owing to the inherent advantages of blockchain,
which enable the creation of a transparent and visible system. As a result, the integrity
of data collected for research is being protected and trust in the outcomes of clin-
ical trials is being fostered. Integrating DL and the blockchain system presents an
intriguing chance to transform the field of telemedicine by facilitating the safe and
confidential transfer and retention of medical images, thereby enabling remote diag-
nosis and advice. This chapter aims to showcase the applications of DL along with the
blockchain in healthcare using an imaging dataset. Using a colossal collection of data,
combined with deep learning and blockchain techniques, it can be trained to exhibit
the desired behaviour. Applications of DL and blockchain technology with imaging

M. Sethi (B) · J. Arora · V. Baggan

Chitkara University Institute of Engineering & Technology, Chitkara University, Punjab, India
e-mail: [email protected]
J. Arora
e-mail: [email protected]
V. Baggan
e-mail: [email protected]
J. Verma · M. Snehi
Punjabi University, Patiala, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 135
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_7
136 M. Sethi et al.

dataset for various diseases such as cancer, diabetic retinopathy, Alzheimer’s etc. can
aid medical professionals in the early investigation and classification of diseases so
that stricken are given effective therapies.

Keywords Deep learning · Alzheimer’s disease · Diabetic retinopathy · Cancer

1 Introduction

The healthcare industry is pursuing novel technologies to augment patient care,

enhance diagnostic precision, and advance healthcare. Over the past few years, DL
and blockchain-based technologies have surfaced as potent instruments with substan-
tial potential in diverse sectors, such as healthcare. DL algorithms, a constituent part
of artificial intelligence, have shown exceptional abilities in examining intricate data,
such as medical imaging. However, blockchain technology, renowned for its decen-
tralized and impregnable characteristics, has attracted considerable interest due to
its capacity to tackle data security and confidentiality predicaments [1]. With the
availability of different kind of datasets such as biomedical imaging, genomics data,
electronics health records, a novel paradigm in healthcare system is rapidly devel-
oping. In particular, highly precise therapy therefore in regards aims to “ensure that
the effective medication is offered to the sufferer in the proper time” by considering a
variety of individual information, such as variation in genetic features, surroundings,
electronic health records (EHRs), and behavior [2]. The deployment of bio-medical
data collection devices, the enormous improvement in image acquisition devices,
and the rise in throughput have produced a continuous quantity of data. The afore-
mentioned data is extremely varied, with large dimensions (MRI, PET, DTI, CT
Scans etc.), and acquired from multiple (often incompatible) sources of information.
This data being opted as consequence, particularly images, becomes intriguing for
medical analysis. Medical experts put in their to keep up with the sector of medical
scans. Automating the diagnostic process with machine learning (ML) approaches
is a viable option, but conventional ML techniques are insufficient to handle such
challenging situations [3] the development of effective healthcare strategies is greatly
focused on data framework. ML is particularly constrained by the underlying issue
of evaluating associations among all the many pieces of knowledge in such large
sets of data [4]. As in recent years, DL a part of ML has experienced a remarkable
boom; the main driving forces behind this progress are advancements in compu-
tational power and the ability to carry large and innovative databases with ease.
The domain has already witnessed significant improvement in the potential of tech-
nology to interpret and operate with data, such as speech, languages, and imaging
data [5]. DL techniques enable a machine to acquire the representations needed for
recognition or classification from vast quantities of unstructured data. DL techniques
involve various models of information using subsequent modifications that strengthen
Deep Learning and Blockchain Applications in Healthcare Sector Using … 137

and eliminate insignificant divergences while amplifying elements of the input data
which are essential to classification. It can happen under or without supervision i.e.
supervised or unsupervised [6].

2 Applications of Deep Learning and Blockchain

in Healthcare

A. Deep Learning and Blockchain in Alzheimer’s Disease Classification

The utilization of blockchain technology has the potential to significantly contribute

to the classification of Alzheimer’s disease through the provision of safe and trans-
parent administration of patient data [7]. The utilization of blockchain technology
for storing medical records and diagnostic details can provide a secure means for
healthcare professionals to access and analyze data, thereby enabling precise catego-
rization and diagnosis of Alzheimer’s disease (AD), one of the most prevalent kinds
of dementia, is a fatal brain illness that primarily harms old people [8]. Dementia
syndrome is mainly characterized by impairment in cognition and other intellectual
abilities when compared to their past skills [9]. According to the survey reported
in 2022, after COVID-19, AD ranks sixth among top 10 fatal factors in the United
States in 2019 and the seventh-leading reason of fatalities in years 2020 and 2021.
Amongst Americans aged over the age of 65, AD continues to be the fifth most
common cause of mortality rates. Mortality from heart disease, HIV, and stroke all
declined during 2000 and 2019, while recorded fatalities from AD grew by a rate of
145%. Approximately sixteen billion hours of attention were given to persons with
AD or other kinds of dementia in 2021 by over ten million close relatives along with
other caregivers. The medical assessment of individuals and connection to family
members are all necessary for the identification of AD. The post-mortem investiga-
tion, which is not clinically vital, used to establish “a firmly grounded fact” diagnosis
of AD for any scenario. Without access to real-world data, patients require another
basis for confirming AD. Such criteria may help us better understand AD and make
it simpler for live patients to determine it. AD has no known cure, but certain specific
medications can slow down its progression [10]. Hence, identification of patients
at early stage who are more likely to progress to AD is as vital as ever. This was
emphasized in a noteworthy conversation delivered by the German neuropathology’s
and the psychiatrist Dr. Alois Alzheimer, who additionally explained the particular
kind of dementia. Eventually, the disease was acknowledged and identified with the
name of the person who first discovered it, Alzheimer’s disease [11]. He revealed
a Frankfurter lady, 51, that had passed away due to acute mental health issues.
When Dr. Alzheimer studied her brain, he found amyloidal plaques and a mass of
fibres which were likely caused by abnormal proteins aggregation in and around
the neurons cells. Once healthy neurons in the cerebrums abruptly stop functioning
properly, cells begin to lose their connections to other brain cells and eventually
die [12, 13]. Computerized tomography (CT) scans, sMRI, functional MRI, and
138 M. Sethi et al.

diffusion tensor imaging (DTI) are pinch of the diverse data collected from medical
imaging modalities, which are currently exploited for diagnosing AD in different
contexts incorporating distinct ML approaches [14]. The primary obstacle in inter-
preting these images is the vast amount of data they contain relative to the limited
number of samples available, which results in high dimensionality. ML techniques
could be used to deal with the aforementioned problems (limited number of samples
as well as complex dimensionality) [15]. DL techniques contribute to a more accurate
and independent strategy since they demand minimal prior understanding of addi-
tional complicated methods, which involves feature extraction and selection [16].
A wide range of methods of classification were applied across the past decade in
order to classify AD based on one modality or an ensemble of modality. Support
vector machines (SVM) and its enhanced architectures, which constitute traditional
ML prediction models, have been extensively used for this area. The authors in [17]
in order to predict the progression from the intermediate stage of Mild Cognitive
Impairment (MCI) to Alzheimer’s disease, a DL strategy was used that included a
multi-modal neural network with a recurrent component, specifically a Recurrent
Neural Network (RNN). Researchers devised an integrated paradigm that incorpo-
rates longitudinal cerebrospinal fluid (CSF) and cognitively functioning indicators
gathered from the Alzheimer’s disease Neuroimaging Initiative cohort (ADNI) in
addition to cross-sectional brain scans indicators at baseline. The proposed design
included data from several domains collected all through period. The results of study
demonstrated that (i) when using only one modality of data separately, prediction
model for MCI conversion to AD yielded up to 75% accuracy and (ii) and while inte-
grating periodic multi-domain information; prediction system performed the best,
achieving an accuracy rate of 81%. Thus, a multi-modal DL strategy could be imple-
mented to assess clinical studies or to determine those at risk of experiencing AD who
would get the best through these. An in-depth view of AD progression assessment
could be provided with the combined analysis of several data modalities. To group
participants into AD, MCI, and controls (CN), researchers thus employed DL to
thoroughly evaluate brain imaging along with genetic and diagnostic test results. To
derive features from medical and genetics facts, the authors build demonising auto-
encoders; for images, employed three-dimensional convolutional neural networks
(CNNs). Furthermore, they designed a cutting-edge data interpretation approach to
locate the best features that the deep models learnt through clustered and perturbation
evaluation. Researchers proved that deep models outperformed shallower designs,
such as SVM, random forests, decision trees and k-nearest neighbours (KNN)’,
employing ADNI dataset [18]. Various ways and techniques utilized to classify AD
into different categories are shown in Table 1.

B. Deep Learning and Blockchain in Cancer

DL algorithms can analyse copious amounts of cancer data, encompassing genome

sequencing and medical imaging, to enhance the prompt identification, diagnosis,
and individualized therapy of the disease. The utilization of blockchain technology
guarantees a secure and transparent exchange of cancer-related information, which in
Deep Learning and Blockchain Applications in Healthcare Sector Using … 139

Table 1 Deep learning methods and approaches used for AD classification

Ref. Dataset Single or Methods and Summary
year multiple approaches
2014 ADNI Multiple Stacked With the goal to get over the barrier
[19] auto-encoders associated with the diagnosis of AD and its
early stage MCI, researchers devised a
DL framework in this research which
comprises layered auto-encoders with a
softmax at the output level
2016 ADNI Single Ensembled The study evaluates the development of
[20] (patches of deep belief DL-based classification methods specific to
grey matter networks brain areas determined by AAL. The AAL
neuro (EDBN) atlas’ areas were utilised to separate grey
images) matter images from each brain region into 3D
patches, and such patches were utilized for
training multiple DBNs. The overall
conclusion for classification is then generated
by a method of voting. Two DL-based
topologies and four distinct voting
computations were built and compared with
the objective to generate an efficient
classification framework where
discriminatory attributes are extracted
unsupervised
2016 ADNI Single CNN Identifying among the most effective
[21] distinctive characteristics remains the hardest
part of categorizing healthcare information
for neurological conditions like AD. Authors
efficiently discriminated functional MRI data
from individuals with AD sufferers from HC
(Individuals who are normal) adopting the
CNN with the prominent framework LeNet-5,
with an accuracy of 97% on test data. The
results of this study demonstrated which the
most effective technique for separating
clinical information with healthy data in
fMRI is the shift and scaling variance features
generated by CNN following DL classifier
(continued)
140 M. Sethi et al.

Table 1 (continued)
Ref. Dataset Single or Methods and Summary
year multiple approaches
2017 ADNI Single Adapted auto In this research on the authors employed DL
[22] encoder strategy on clinically relevant data along with
neuro images to provide an early
AD diagnoses. Gender, age, and the
individual’s ApoE genotype were all
incorporated in the clinically
relevant information. By applying data from
resting-state functional-MRI, a neural
network is formed by estimating the
operational links among various regions of
the brain. To distinguish between typical
ageing and moderate cognitive impairment,
an early stage of AD, an adapted auto encoder
network was developed. The proposed
approach precisely disclosed discriminatory
neural network aspects and presented an
effective AD classifier
2018 ADNI Single AlexNet Researchers, however, found it challenging to
[23] identify various stages due to elderly
individuals’ and distinct stages’ brain features
are identical. The CNN model AlexNet was
used in this research to classify diverse
AD stages applying fMRI scans. With a
DL technique, they were able to classify
AD into five distinct stages. The method
discriminated between AD, Early- MCI,
Late-MCI, significant memory concern
(SMC), and healthy controls
2019 ADNI Multiple FSBi-LSTM In this study, a conceptual paradigm was
[24] and 3D-CNN created by the authors. The design
they proposed specifically made use of
advantages of FSBi-LSTM and 3D-CNN. In
order to derive depth representation of
features utilizing MRI and PET, they first
developed 3D-CNN architecture. Then, in
order to further enhance its efficacy,
researchers applied FSBi-LSTM on the
disguised spatial information obtained from
deep mapping of features
(continued)
Deep Learning and Blockchain Applications in Healthcare Sector Using … 141

Table 1 (continued)
Ref. Dataset Single or Methods and Summary
year multiple approaches
2020 ADNI Clinical + CNN The researchers developed an interpretable
[25] single DL algorithm which identifies multiple signs
modality associated with AD using heterogeneous
sources of data such as sex, age group, and
MMSE score as well as MRI neuro images.
The framework connected a fully
convolutional network, generating precise,
comprehensible representations of each
individual’s accurate diagnosis, to a
multiple-layer perceptron, that
developed precise maps of disease probability
from localised brain structures
2021 ADNI Multiple Stacked To derive characteristics from clinic and
[18] de-noising 3D genetics samples, researchers
CNNs used stacked de-noising auto-encoders; for
imaging data, they employed 3D CNNs.
Furthermore, they proposed an innovative
data interpretation strategy for finding the
most promising features associated with the
deep models acquired using grouping and
perturbations evaluation
2022 ADNI Single CNN This work effectively established and
[26] analysed a CAD mechanism for
distinguishing AD sufferers from
healthy controlled individuals employing
features from PET
neuro scans. CNN was incorporated in the
building process of the recommended CAD
tool. The decomposition of the PET
neuro scans into multiple 2D slices enabled
an analysis of the features. The individual
slices were then placed at short intervals
avoiding overlapping
2022 ADNI Single Pre-trained Using the analysis of brain MRIs, the initial
[27] models levels of AD were categorised this
investigation as healthy cognitive, mild
cognitive impairment, and AD. Multiple
neuroimages from the ADNI database were
recognised using several models based upon
the CNN paradigm. The efficacy of 29 distinct
pre-trained models on brain images has been
examined and the research was laid out in an
adequately complete comparative framework
(continued)
142 M. Sethi et al.

Table 1 (continued)
Ref. Dataset Single or Methods and Summary
year multiple approaches
2022 Mayo Single YOLOv3 To identify five different kinds of tau lesion,
[28] clinic including neural inclusions, motor plaques,
brain astrocytic deposits, and helical structures,
bank researchers developed the YOLOv3
recognition system. During training, the team
utilised 2522 digital sliding scans of
CP13-immunostained motor cerebral slices
comprising 10 cases. The dataset used for
training was augmented to render it larger in
size. The numerical loads for every tau stroke
in the cerebellum, motor cortex, caudal core,
and the superior temporal gyrus, derived by
the object’s recognition method, were
subsequently leveraged to generate random
forest models
2023 OASIS Single Shallow CNN This study proposed a DL based network for
[29] reliable AD classification and diagnosis. The
proposed analytic pipeline involved a shallow
CNN framework on MRI neuro imaging
dataset

turn promotes cooperation among researchers and facilitates the provision of patient-
centred care. Integrating DL alongside blockchain technology holds promise for the
progression of cancer research, improvement of treatment efficacy, and condition
of patient empowerment in cancer management [30]. Globally, cancer is respon-
sible for more deaths than any other cause. The obstacles of fighting cancer are
being faced by both academics and clinicians. As per the American Cancer Society’s
2019 Novel Cancer Released Research Report, there were approx. 18 thousands
casualties from brain tumours in the year 2019, 142,670 being credited to lung
cancer, 620 with skin cancer, 42,260 from breast cancer, and 31,620 stemming from
prostate cancer [31]. One out of every ten adults in the US has been diagnosed
of cancer disease. Kerala has India’s largest cancer prevalence rate [32]. One in six
casualties worldwide is caused by cancer disease [33]. Even while new methods
could enhance cancer treatment and raise the survival rate, the purpose of cancer
prognosis is to predict the evolution of the disease, provide life estimates, and assist
with patient care [34]. Improved survival prediction based on the medical criteria and
genetic profiles of individuals is a significant objective in cancer prognosis. Mathe-
matical methods, such as log-ranks test [35], Cox proportionality hazards model [36],
and Kaplan Meier estimator [37], have become cutting-edge analytical techniques in
the field of cancer prognosis for longevity analysis. Healthcare data, such as detec-
tion of cancer, cancer categories, tumour levels, genetic background, etc., constitute
to the significant input of data to support these approaches in cancer forecasting for
survival forecasting. Different forms of data are currently available for comparing
Deep Learning and Blockchain Applications in Healthcare Sector Using … 143

the current level of sickness to earlier decades. This data is multi-omics data from
high throughput and multi-dimensional samples of patients [38]. The large amount
of multi-omic input makes it hard for making predictions which believe in utilising
statistical approaches. To deal with such challenges, several techniques including
ML have been used or implemented. Principal component analysis (PCA), clus-
tering, and auto encoder were all developed effectively for classifying various types
of cancer [39]. Furthermore, cancer prognosis prediction has effectively employed
techniques such as semi-supervised learning, SVM, Bayesian networks, and decision
trees. The area of DL among other areas has benefited from significant growth in
computational capacity and advancements in technology in the past decade. Table 2
displays several DL methodologies employed to classify cancer into a number of
types.
C. Deep Learning and Blockchain in Diabetic Retinopathy
Diabetic retinopathy (DR), a rampant eye infliction, is a consequence of diabetes
mellitus. It is characterized by abnormalities on the retina (back of the eye) which
may hamper vision of an individual. If unacknowledged, it might result in blindness
[51]. The Non-Proliferative DR (NPDR) and the Proliferative DR (PDR) are the two
main categories of DR [52]. Early-stage DRs are generally referred to as NPDRs and
can successfully be subdivided into three different categories: Mild, moderated, and
extreme/severe stages [53]. Single aneurysm, which is a tiny red dot at the ends of
blood arteries, exists in the mild phase. In the moderated stage, the aneurysm burst
into beneath the retina to cause a retinal haemorrhage and the formation of an intricate
structure that resemble a flame. In the most advanced stage, each of the four quad-
rants of the retina can exhibit over twenty intra-retinal haemorrhages, while there is
apparent vascular bleeding besides apparent intra-retinal micro-vascular irregulari-
ties. The most extreme forms of DR, referred to as PDR, causes neovascularization,
or the spontaneous development of new vessels of blood in the shape of opera-
tional micro-vascular systems across the innermost layer of the retinal area. Although
various treatments are available to manage the condition, there is no known cure for
DR at present. Detection and therapy of DR at initial stage may considerably mini-
mize the chances of losing one’s vision. Around 2025, there will be estimated 592
million DR afflicted worldwide, compared with 382 million presently. According to
a survey completed by researchers in the Pakistani province of Khyber Pakhtunkhwa
(KPK), 5.6% of diabetic individuals who developed DR were blinded. Whenever the
mild NPDR is not controlled in the initial phases, it eventually matures into PDR
[54]. According to reports, 26% of the identified individuals were having a diagnosis
of PDR, reaching up 24% of the entire population of DR individuals identified in
Sindh, Pakistan. Individuals with the DR remain undiagnosed in the initial stages,
whereas later stages induce floaters, vision problems, and gradual degradation of the
ability to see. Thus, it is challenging but essential to recognise DR in its initial phases
with the objective to prevent the serious implications of its later stages. Coloured
fundus images are used after diagnosing DR. The manual assessment is costly as
well as time-consuming since it needs to be conducted by respective highly skilled
experts. Consequently, it has become vital to adopt computer vision methods that
Table 2 Deep Learning Methods and Approaches used for Cancer Diagnosis
144

Ref. year Dataset Modality Methods and approaches Summary

2016 [40] Histo.path BreakHis CNN In this study, researchers obtained
BreaKHis, a readily available database
accessed at http://web.inf.ufpr.br/vri/bre
ast-cancer-database, to perform initial
experiments employing the DL approach
for classifying breast cancer histological
images. Authors developed an algorithm
that involved extracting image patches for
CNN training and integrating those
patches towards final classification. By
applying the higher definition
histopathology scans via BreaKHis as feed
to the present CNN, this strategy attempts
to prevent model changes that could result
in a more complicated and highly
computational framework
2017 [41] Histo.path BreakHis Structured deep neural network In this study, the researcher’s group
employed a novel DL framework for
classification a breast cancer images in to
multiple categories. It is evident from the
structured DNN framework’s outstanding
performance with an average
of 93% accuracy on an extensive data
that the approach is effective at offering a
tool for breast cancer multi-classification
in healthcare scenarios
(continued)
M. Sethi et al.
Table 2 (continued)
Ref. year Dataset Modality Methods and approaches Summary
2018 [42] Histo.path ICIAR 2018 Deep CNNs with gradients boosted tress In the present
research, researchers adopted a
mathematical strategy for classifying
breast cancer histology
imaging dataset utilizing deep CNNS.
The strategy made use of gradients boosted
trees classifiers and multiple deep CNNs
2019 [43] Histo.path BreakHis CNN with small SE-ResNet In this research, researchers developed an
innovative sort of CNN which is made up
of a Conv Layers, a tiny SE-ResNet
element, and lastly fully connected layer.
They provided a lightweight SE-ResNet
module that outperforms an amalgam of
squeeze-and-excite units and residue
sections alongside having fewer
parameters in order to attain a comparable
level of effectiveness. Also, they
suggested an entirely novel scheduling for
learning rates (LRs) that achieved
exceptional results without requiring
extensively tuning LRs. They implemented
an algorithm to automatically classify
Deep Learning and Blockchain Applications in Healthcare Sector Using …

breast cancer histological pictures into

eight subgroups based on whether the
disease was benign or malignant
(continued)
145
Table 2 (continued)
146

Ref. year Dataset Modality Methods and approaches Summary

2019 [44] Colsanitas Colombia Whole slides images Ensemble DL (VGG 16 and VGG 19) This study employed the combined DL
University’s pathology approach to differentiate the histological
department pictures of breast cancer into two distinct
classes: carcinoma and non-carcinoma.
Using the transfer learning pre-trained
frameworks named VGG 16 and 19, the
authors trained four
different models. Across all of the various
designs, including the fully-trained and
fine-tuned VGG16 and VGG 19, they first
performed fivefold cross-validation
methods. Subsequently adopting an
ensemble technique that involved
averaged forecasted probabilities,
researchers discovered that the combined
performance of the fine-tuned VGG16
and 19 algorithms was comparable in
grouping, primarily with regard to the
carcinoma category. The average
effectiveness of the VGG16 and
VGG19 classifiers was 95%, with a
sensitivity of 98% for the malignancy
cohort
(continued)
M. Sethi et al.
Table 2 (continued)
Ref. year Dataset Modality Methods and approaches Summary
2020 [45] – – Hybrid convolution and recurrent neural For classification of breast cancer
network utilizing histopathology images,
authors employed a novel hybrid
convolutional and recurrent deep
NN. The approach integrated the strengths
of CNN and recurrent NN on the deeper
hierarchical representation of features
from histopathology patches of an image.
Both the short and long-term spatially
correlations across regions were retained.
The experiment findings
demonstrated with an overall performance
of 91% for the multi class
(4-class) classification
2021 [46] – – Auxiliary classifier generative adversarial The
network (ACGAN) researcher’s group adopted an ACGAN,
which generated realistic images alongside
their class labels, to expand the quantity of
a confined dataset. By classifying breast
cancer histopathology image data using
deep CNN classifiers trained on our
expanded dataset, they assessed the
Deep Learning and Blockchain Applications in Healthcare Sector Using …

performance of ACGAN augmentation

technique. They deployed a method known
as transfer learning for building classifier,
extracting the convolutional features from
a pertained network and input these into
multiple extreme gradient booster
(XGBoost) classifiers
147

(continued)
Table 2 (continued)
148

Ref. year Dataset Modality Methods and approaches Summary

2022 [47] Histo.path BreakHis AHoNet This research designed a distinctive
AHoNet by concurrent integrating
attention strategy and high-order
probabilistic estimation in remnant
convolutional network in order to identify
more discriminating hidden features
from breast cancer pathologic imaging
dataset. AHoNet at first applied a robust
channel attention component
without dimensionality reductions and
locally cross-channel contact to
identify localized significant deep features
of pathological images of breast cancer.
Later that, using matrix
power standardization, the second-order
covariance statistics were progressively
computed, offered breast cancer
pathological imaging an increasingly
reliable overall feature depiction
2022 [48] Histo.path ICIAR 2018 Challenge Ensemble VGG-16 and DenseNet-201 Towards reliable breast histopathology
datasets imaging grouping, the proposed DEEP_
Pachi integrated both local and global
features. The ensemble VGG-16
and DenseNet-201 frameworks constituted
the basis of the mentioned model. Whilst
DEEP_Pachi gathered spatial or local
features from patches, the
ensembled model extracted global generic
features from the whole image
(continued)
M. Sethi et al.
Table 2 (continued)
Ref. year Dataset Modality Methods and approaches Summary
2022 [49] Histo.path BreakHis Efficient-Net framework with SVM For the diagnosis of breast cancer on
classifier histo-pathological imaging dataset, the
study proposed a DL along with transfer
learning model. The researchers applied
transfer learning technique on only a few
training histopathology samples to classify
into classes without compromising
efficiency. With the goal to derive features,
patches/regions were initially extracted
from Entire Slide Images and input to the
CNN. The discriminative patches
were chosen on the basis of these features
before being input to an Efficient-Net
framework which was already trained on
the ImageNet data set. The Efficient-Net
architecture’s features were also exploited
for training SVM classifier as well
2023 [50] Histo.path BreakHis Arithmetic optimized DL technique for In the present study, AOADL-HBCC is
histological breast cancer proposed by the authors. The
(AOADL-HBCC) AOADL-HBCC technique included a
mechanism that improves contrast and
eliminating background noise based on a
Deep Learning and Blockchain Applications in Healthcare Sector Using …

median filter (MF). Furthermore, the

AOADL-HBCC approach integrated an
AOA alongside a Squeeze Net framework
to extricate features. In order to classify
breast cancer, a deep belief network (DBN)
classifier that includes a hyper parameter
optimizer (Adamax) was adopted
149
150 M. Sethi et al.

automatically interpret the fundus scans to aid radiologists and specialists. Hands-on
feature engineering [55] and end-to-end learning that is DL [56] represents two cate-
gories of computer vision-based approaches. The conventional approaches adopted
by the hands-on engineering procedures to extract features are time consuming and
error prone. The potential applications of DL along with blockchain technology in
the setting of diabetic retinopathy, a common cause of blindness among those who
have diabetes, have demonstrated promising outcomes. The application of DL algo-
rithms enables the evaluation of retinal images to identify symptoms of diabetic
retinopathy, thereby facilitating timely treatment and diagnosis. The implementa-
tion of blockchain technology has the potential to safeguard the confidentiality and
integrity of medical records, facilitating the efficient and protected exchange of retinal
images between medical professionals. This can lead to improved precision in diag-
nosing and shrivelling diabetic retinopathy. The amalgamation of DL with blockchain
computing holds promise for enhancing the management of diabetic retinopathy
and averting vision loss among individuals with diabetes. Table 3 lists various DL
techniques that were used to classify DR fundus images into different categories.

3 Conclusion

Integrating DL with blockchain technology has exhibited noteworthy potential in

healthcare imaging. The utilization of DL algorithms in the analysis of medical
images has demonstrated exceptional proficiency in facilitating precise diagnosis,
computer-assisted detection, and individualized treatment planning. In contrast,
blockchain technology ensures the security, privacy, compatibility, and transparency
of data, thereby offering a solution to the significant challenges that arise in managing
healthcare data. The impact of these technologies on healthcare imaging is signif-
icant, resulting in better patient outcomes, heightened diagnostic precision, effi-
cient data sharing, and amplified healthcare advancement. Moreover, DL algorithms
are effective in enhancing customary ML, enabling computers to acquire data in
order to recognize strategies for advancing outcomes. Approaches ranging from
natural language processing to computer vision, have had previously exploited these
methods. Furthermore, considering the emphasis on representation learning, deep
networks offer the ability to fuse assorted sets of data throughout distinctive types.
Hierarchical learning design enables higher generality. Such in-depth representa-
tions can subsequently be applied for assisting medical procedures within areas, like
health evaluation, individualised medicines suggestions, guidance on treatment, clin-
ical study employment, and statistical data gathering and assessment. This chapter,
focused on various DL approaches, can be applied in the healthcare industry. In
light of this, we believe that DL has the power to usher in a new era of forecasting
systems for healthcare, which may prove to be a game-changer for the field (i)
scale to incorporate millions to a billion of medical details and (ii) employ only
one, distributed individual’s illustration to successfully assist medical professionals
in everyday tasks—instead of multiple mechanisms through distinct patients data.
Deep Learning and Blockchain Applications in Healthcare Sector Using … 151

Table 3 Deep learning methods and approaches used for diabetic retinopathy image classification
Ref. Dataset Modality Methods and Summary
year approaches
2016 Kaggle dataset Fundus CNN with In this study, researchers suggested a
[57] (https://www. retinal augmentation CNN as a reliable methodology for
kaggle.com) images identifying DR from digitised retinal
ocular illustrations (images) and
grading its level of severity.
Researchers designed CNN network
architecture along with augmented
data that can recognise the complex
elements needed for the
classification task, such as
micro-aneurysms, fluid, and
haemorrhaging on retinas, and
thereafter delivered an automatic
diagnosis without requiring any
human intervention
2017 FINDeRS Fundus Deep-DR-Net In this study, a cost-effective
[58] dataset retinal embedded framework driven by
images DL was developed that allows
professionals estimate the extent of
DR using retinal fundus images.
Then, a simple embedded
board-compatible DL model called
Deep-DR-Net was recommended
for employing in such a scenario. A
cascaded codec-classifier
framework was established up
utilising residual approach at the
core of Deep-DR-Net that ensured
an optimal modelled size
2018 Ophthalmology Fundus Deep CNN The authors created an archive of
[59] Department, retinal DR fundus images featuring the
Health images relevant management approach
Management labelled in order to automate the
Center, and diagnosis of DR and
Endocrinology offered appropriate
& Metabolism recommendations to DR individuals.
Department Researchers built deep CNN models
on this dataset to classify the level of
severity of DR fundus images
(continued)
152 M. Sethi et al.

Table 3 (continued)
Ref. Dataset Modality Methods and Summary
year approaches
2019 Beijing Tongren Fundus Integrated DL This study employs numerous
[60] Eye Center retinal models appropriately trained DL models to
images illustrate an autonomous
image-level DR recognition
approach. Additionally, with the aim
to lessen the biases of every
individual framework, a number of
DL models were combined via the
Adaboost method. The weighted
CAMs that could reveal the probable
location of damages have been
included in this study in order to
clarify the DR findings. To expand
the number of fundus images, eight
image manipulation techniques
were also incorporated during
pre-processing
2019 Kaggle dataset Fundus Ensemble of five In the present research, the
[61] images deep contributors established a group of
CNN framework five deep CNN architectures
namely Inceptionv3, DenseNet (121
and 169), Resnet50, and Xception to
capture the meaningful features
while improving classification
accuracy across various phases of
DR employing the freely accessible
Kaggle database of retinal images
2020 Messidor-1 with Fundus Hybrid CNN The paper discussed the challenge
[62] the APTOS images transfer learning for automatically identifying
2019 vision framework DR and presented a novel DL hybrid
diagnosis model to fix the problem. In order to
develop the hybrid framework, they
used an additional blocks of CNN
layering over top of the previously
trained Inception-Res
network utilizing the technique of
transfer learning. Considering the
Messidor-1 database for DR with
the APTOS 2019 vision diagnosis
(Kaggle dataset),
researchers analysed the accuracy of
the suggested model. Compared
with other results reported, their
approach performed better. Using
the APTOS and Messidor-1 and
datasets, the developers obtained
test accuracy of 82%
and 72% respectively
(continued)
Deep Learning and Blockchain Applications in Healthcare Sector Using … 153

Table 3 (continued)
Ref. Dataset Modality Methods and Summary
year approaches
2020 Kaggle dataset Fundus VGG-16 and This study introduced a
[63] (EyePACS images VGG-19 computer-aided classification
dataset) approach that used CNN, and two
transfer learning VGG-16, and
VGG-19 DL models to evaluate
coloured fundus images involving
various illumination and angles
views and yielded a level of
seriousness level for DR
2021 Kaggle dataset Fundus – In order accelerate the training
[64] images process and convergence,
researchers focused on identifying
the DR’s various stages based on the
minimal learnable parameters
needed. The VGG-Network in
Network (NiN) model, a
significantly nonlinear
measure-invariant DL model,
was built by stacking the VGG16,
the spatially pyramidal layer for
pooling (SPP), and the NiN. In
addition to the advantages of the
SPP layer, the recommended
VGG-NiN model processed a DR
image at any dimension.
Furthermore, the stacking of NiN
offered the model significant
nonlinearity and enhanced
classification results
2022 DRIVE and Fundus CNN The goal of this research was to
[65] messidor images establish a computer-based
datasets classification method for a set of
retinal images used to recognize DR.
CNN DL technique has been used to
generate a multiclass classification
framework which could
automatically diagnose and classify
sickness levels
(continued)
154 M. Sethi et al.

Table 3 (continued)
Ref. Dataset Modality Methods and Summary
year approaches
2022 – Fundus Hybrid CNN-SVD In this study, the authors proposed
[66] images an innovative a two-phase method
for automatic DR categorization.
Pre-processing and data
augmentation techniques
were utilised to improve the quality
of the images and quantity because
the asymmetric Optical Disc (OD)
and blood vessels recognition
technique has a small proportion of
positive instances. In the initial
phase, the segmentation of the OD
and blood vessel was carried out
employing two different
U-Net algorithms. The next step
involved building of the symmetrical
hybrid CNN-SVD model, which
observes DR by identifying retinal
indicators like micro
aneurysms, haemorrhages), and
exudates, following initial
processing to obtain and find the
strongest discriminant features
through Inception-V3
2022 Kaggle dataset Fundus ResNet and Alex In this investigation,
[67] (EyePACS t images Net Multi-Resolution Analysis (MRA)
and CNN architecture were utilised
together for the input image features
enhancement with no adding
additional convolution filters. This
study proposed a new HW step
activated function having distinct
properties associated with the
wavelet channel-bands
2023 DIARETDB1 Fundus Deep ensembled For the DR recognition and DR, a
[68] and APTOS images DenseNet101 and fully automated ensemble DL model
2019 ResNeXt was presented in this research. For
the purpose of detecting diabetic
retinopathy, two DL algorithms,
namely improved DenseNet101 and
ResNeXt, are integrated
Deep Learning and Blockchain Applications in Healthcare Sector Using … 155

In this chapter, we showed how DL techniques can be applied in the healthcare

industry using an imaging dataset. DL approaches can train it using comprehensive
data. Applications of DL technology with imaging datasets for a range of diseases,
including cancer, diabetic retinopathy, Alzheimer’s disease, etc., can facilitate early
investigation and classification of diseases so that patients can receive effective
therapies.

References

1. Singh B, Tatiya M, Shrivastava A, Verma D, Pratap Srivastava A, Rana A (2022) Detection

of Alzheimer’s disease using deep learning, blockchain, and IoT cognitive data. In: 2022 2nd
international conference on technological advancements in computational sciences (ICTACS),
Oct 2022, pp 863–869.https://doi.org/10.1109/ICTACS56270.2022.9988058
2. Lyman GH, Moses HL (2016) Biomarker tests for molecularly targeted therapies—the key
to unlocking precision medicine. N Engl J Med 375(1):4–6. https://doi.org/10.1056/NEJMp1
604033
3. Razzak MI, Naz S, Zaib A (2018) Deep Learning for medical image processing: overview,
challenges and the future. In: Lecture notes in computational vision and biomechanics, vol 26,
pp 323–350
4. Miotto R, Wang F, Wang S, Jiang X, Dudley JT (2017) Deep learning for healthcare: review,
opportunities and challenges. Brief Bioinform 19(6):1236–1246. https://doi.org/10.1093/bib/
bbx044
5. Esteva A et al (2019) A guide to deep learning in healthcare. Nat Med 25(1):24–29. https://
doi.org/10.1038/s41591-018-0316-z
6. Panch T, Szolovits P, Atun R (2018) Artificial intelligence, machine learning and health systems.
J Glob Health 8(2):1–8. https://doi.org/10.7189/jogh.08.020303
7. Pilozzi A, Huang X (2020) Overcoming Alzheimer’s disease stigma by leveraging artificial
intelligence and blockchain technologies. Brain Sci 10(3):183. https://doi.org/10.3390/brains
ci10030183
8. Bergstra J, Bengio Y (2012) Random search for hyper-parameter optimization. J Mach Learn
Res 13:281–305
9. Chang CC, Lin CJ (2011) LIBSVM: A Library for support vector machines. ACM Trans Intell
Syst Technol 2(3). https://doi.org/10.1145/1961189.1961199
10. Sethi M, Ahuja S, Rani S, Bawa P, Zaguia A (2021) Classification of Alzheimer’s disease
using gaussian-based Bayesian parameter optimization for deep convolutional LSTM network.
Comput Math Methods Med 2021. https://doi.org/10.1155/2021/4186666
11. Yiğit A, Işik Z (2020) Applying deep learning models to structural MRI for stage prediction
of Alzheimer’s disease. Turkish J Electr Eng Comput Sci 28(1):196–210. https://doi.org/10.
3906/elk-1904-172
12. Brückner G, Hausen D, Härtig W, Drlicek M, Arendt T, Brauer K (1999) Cortical areas abundant
in extracellular matrix chondroitin sulphate proteoglycans are less affected by cytoskeletal
changes in Alzheimer’s disease. Neuroscience 92(3):791–805. https://doi.org/10.1016/S0306-
4522(99)00071-8
13. Wang SH, Phillips P, Sui Y, Liu B, Yang M, Cheng H (2018) Classification of Alzheimer’s
disease based on eight-layer convolutional neural network with leaky rectified linear unit and
max pooling. J Med Syst 42(5):85. https://doi.org/10.1007/s10916-018-0932-7
14. Azari NP et al (1993) Early detection of Alzheimer’s disease: a statistical approach using
positron emission tomographic data. J Cereb Blood Flow Metab 13(3):438–447. https://doi.
org/10.1038/jcbfm.1993.58
156 M. Sethi et al.

15. Razavi F, Tarokh MJ, Alborzi M (2019) An intelligent Alzheimer’s disease diagnosis method
using unsupervised feature learning. J Big Data 6(1). https://doi.org/10.1186/s40537-019-
0190-7
16. Lecun Y, Bengio Y, Hinton G (2015) Deep learning. Nature 521(7553):436–444. https://doi.
org/10.1038/nature14539
17. Lee G et al (2019) Predicting Alzheimer’s disease progression using multi-modal deep learning
approach. Sci Rep 9(1):1–12. https://doi.org/10.1038/s41598-018-37769-z
18. Venugopalan J, Tong L, Hassanzadeh HR, Wang MD (2021) Multimodal deep learning models
for early detection of Alzheimer’s disease stage. Sci Rep 11(1):1–13. https://doi.org/10.1038/
s41598-020-74399-w
19. Liu S, Liu S, Cai W, Pujol S, Kikinis R, Feng D (2014) Early diagnosis of Alzheimer’s disease
with deep learning. In: 2014 IEEE 11th int. symp. biomed. imaging, ISBI 2014, pp 1015–1018.
https://doi.org/10.1109/isbi.2014.6868045
20. Ortiz A, Munilla J, Górriz JM, Ramírez J (2016) Ensembles of deep learning architectures for
the early diagnosis of the Alzheimer’s disease. Int J Neural Syst 26(7):1–23. https://doi.org/
10.1142/S0129065716500258
21. Sarraf S, Tofighi G (2017) Deep learning-based pipeline to recognize Alzheimer’s disease using
fMRI data. In: FTC 2016 - proc. futur. technol. conf., no December, pp 816–820. https://doi.
org/10.1109/FTC.2016.7821697.
22. Pan D, Huang Y, Zeng A, Jia L, Song X (2019) Early Diagnosis of Alzheimer’s disease based on
deep learning and GWAS. Commun Comput Inf Sci 1072(1):52–68. https://doi.org/10.1007/
978-981-15-1398-5_4
23. Kazemi Y, Houghten S (2018) A deep learning pipeline to classify different stages of
Alzheimer’s disease from fMRI data. In: 2018 IEEE conf. comput. intell. bioinforma. comput.
biol. CIBCB 2018, no Mci, pp 1–8. https://doi.org/10.1109/CIBCB.2018.8404980
24. Feng C et al (2019) Deep learning framework for Alzheimer’s disease diagnosis via 3D-CNN
and FSBi-LSTM. IEEE Access 7:63605–63618. https://doi.org/10.1109/ACCESS.2019.291
3847
25. Qiu S et al (2020) Development and validation of an interpretable deep learning framework
for Alzheimer’s disease classification. Brain 143(6):1920–1933. https://doi.org/10.1093/brain/
awaa137
26. Ahila A, Poongodi M, Hamdi M, Bourouis S, Rastislav K, Mohmed F (2022) Evaluation of
neuro images for the diagnosis of Alzheimer’s disease using deep learning neural network.
Front Public Heal 10, no February, pp 834032. https://doi.org/10.3389/fpubh.2022.834032
27. Savaş S (2022) Detecting the Stages of Alzheimer’s disease with pre-trained deep learning
architectures. Arab J Sci Eng 47(2):2201–2218. https://doi.org/10.1007/s13369-021-06131-3
28. Koga S, Ikeda A, Dickson DW (2022) Deep learning-based model for diagnosing Alzheimer’s
disease and tauopathies. Neuropathol Appl Neurobiol 48(1):1–12. https://doi.org/10.1111/nan.
12759
29. EL-Geneedy M, Moustafa HED, Khalifa F, Khater H, AbdElhalim E (2023) An MRI-based deep
learning approach for accurate detection of Alzheimer’s disease. Alexandria Eng J 63:211–221.
https://doi.org/10.1016/j.aej.2022.07.062
30. Kumar R et al (2021) An Integration of blockchain and AI for secure data sharing and detection
of CT images for the hospitals. Comput Med Imaging Graph 87:101812. https://doi.org/10.
1016/j.compmedimag.2020.101812
31. Munir K, Elahi H, Ayub A, Frezza F, Rizzi A (2019) Cancer diagnosis using deep learning: a
bibliographic review. Cancers (Basel) 11(9):1–36. https://doi.org/10.3390/cancers11091235
32. Kaushal C, Singla A (2020) Automated segmentation technique with self-driven post-
processing for histopathological breast cancer images. CAAI Trans Intell Technol 5(4):294–
300. https://doi.org/10.1049/trit.2019.0077
33. Zhu W, Xie L, Han J, Guo X (2020) The application of deep learning in cancer prognosis
prediction. Cancers (Basel) 12(3):1–19. https://doi.org/10.3390/cancers12030603
34. Kaushal C, Kaushal K, Singla A (2021) Firefly optimization-based segmentation technique to
analyse medical images of breast cancer. Int J Comput Math 98(7):1293–1308. https://doi.org/
10.1080/00207160.2020.1817411
Deep Learning and Blockchain Applications in Healthcare Sector Using … 157

35. Society RS (2013) Asymptotically efficient rank invariant test procedures author (s): Richard
Peto and Julian Peto Reviewed work (s): Source : Journal of the Royal Statistical Society .
Series A (General), 135(2) (1972), pp Published by : Wiley for th, vol 135, no 2, pp 185–207
36. Ahmed FE, Vos PW, Holbert D (2007) Modeling survival in colon cancer: a methodological
review. Mol Cancer 6:1–12. https://doi.org/10.1186/1476-4598-6-15
37. Kaplan EL, Meier P (1958) Nonparametric estimation from incomplete observations. J Am
Stat Assoc, vol 53, no March 2013, pp 457–81
38. Goossens N, Nakagawa S, Sun X, Hoshida Y (2015) Cancer biomarker discovery and
validation. Transl Cancer Res 4(3):256–269. https://doi.org/10.3978/j.issn.2218-676X.2015.
06.04
39. Tan M et al (2014) Lysine glutarylation is a protein posttranslational modification regulated by
SIRT5. Cell Metab 19(4):605–617. https://doi.org/10.1016/j.cmet.2014.03.014
40. Jiang Y, Chen L, Zhang H, Xiao X (2019) Breast cancer histopathological image classification
using convolutional neural networks with small SE-ResNet module. PLoS One 14(3). https://
doi.org/10.1371/journal.pone.0214587
41. Han Z, Wei B, Zheng Y, Yin Y, Li K, Li S (2017) Breast cancer multi-classification from
histopathological images with structured deep learning model. Sci Rep 7(1):1–10. https://doi.
org/10.1038/s41598-017-04075-z
42. Kumar ES, Bindu CS, Madhu S (2020) Deep convolutional neural network-based analysis for
breast cancer histology images, vol 1. Springer International Publishing
43. Jiang Y, Chen L, Zhang H, Xiao X (2019) Breast cancer histopathological image classification
using convolutional neural networks with small SE-ResNet module. PLoS ONE 14(3):1–21.
https://doi.org/10.1371/journal.pone.0214587
44. Hameed Z, Zahia S, Garcia-Zapirain B, Aguirre JJ, Vanegas AM (2020) Breast cancer
histopathology image classification using an ensemble of deep learning models. Sensors
(Switzerland) 20(16):1–17. https://doi.org/10.3390/s20164373
45. Yan R et al (2020) Breast cancer histopathological image classification using a hybrid deep
neural network. Methods 173(2019):52–60. https://doi.org/10.1016/j.ymeth.2019.06.014
46. Gheshlaghi SH, Nok Enoch Kan C, Ye DH (2021) Breast cancer histopathological image
classification with adversarial image synthesis. In: 2021 43rd annual international conference
of the IEEE engineering in medicine & biology society (EMBC), pp 3387–3390. https://doi.
org/10.1109/EMBC46164.2021.9630678
47. Zou Y, Zhang J, Huang S, Liu B (2022) Breast cancer histopathological image classification
using attention high-order deep network. Int J Imaging Syst Technol 32(1):266–279. https://
doi.org/10.1002/ima.22628
48. Ukwuoma CC, Hossain MA, Jackson JK, Nneji GU, Monday HN, Qin Z (2022) Multi-
Classification of breast cancer lesions in histopathological images using DEEP_Pachi: multiple
self-attention head. Diagnostics 12(5). https://doi.org/10.3390/diagnostics12051152
49. Ahmad N, Asghar S, Gillani SA (2022) Transfer learning-assisted multi-resolution breast
cancer histopathological images classification. Vis Comput 38(8):2751–2770. https://doi.org/
10.1007/s00371-021-02153-y
50. Obayya M, et al (2023) Hyperparameter optimizer with deep learning-based decision-support
systems for histopathological breast cancer diagnosis. Cancers (Basel) 15(3). https://doi.org/
10.3390/cancers15030885
51. Alyoubi WL, Shalash WM, Abulkhair MF (2020) Diabetic retinopathy detection through deep
learning techniques: a review. Informatics Med Unlocked 20:100377. https://doi.org/10.1016/
j.imu.2020.100377
52. Raja Memon DW, Lal DB, Aziz Sahto DA (2017) Diabetic retinopathy; frequency at level of
hba1c greater than 6.5%. Prof Med J 24(2):234–238. https://doi.org/10.17957/tpmj/17.3616
53. Wu L (2013) Classification of diabetic retinopathy and diabetic macular edema. World J
Diabetes 4(6):290. https://doi.org/10.4239/wjd.v4.i6.290
54. Jan S, Ahmad I, Karim S, Hussain Z, Rehman M, Shah MA (2018) Status of diabetic retinopathy
and its presentation patterns in diabetics at ophthalomogy clinics. J Postgrad Med Inst 32(1):24–
27
158 M. Sethi et al.

55. Seoud L, Hurtut T, Chelbi J, Cheriet F, Langlois JMP (2016) Red lesion detection using dynamic
shape features for diabetic retinopathy screening. IEEE Trans Med Imaging 35(4):1116–1126.
https://doi.org/10.1109/TMI.2015.2509785
56. Quellec G, Charrière K, Boudi Y, Cochener B, Lamard M (2017) Deep image mining for
diabetic retinopathy screening. Med Image Anal 39:178–193. https://doi.org/10.1016/j.media.
2017.04.012
57. Pratt H, Coenen F, Broadbent DM, Harding SP, Zheng Y (2016) Convolutional neural networks
for diabetic retinopathy. Procedia Comput Sci 90(July):200–205. https://doi.org/10.1016/j.
procs.2016.07.014
58. Ardiyanto I, Nugroho HA, Buana RLB (2017) Deep learning-based Diabetic Retinopathy
assessment on embedded system. In: Proc. annu. int. conf. IEEE eng. med. biol. soc. EMBS,
pp 1760–1763. https://doi.org/10.1109/EMBC.2017.8037184
59. Gao Z, Li J, Guo J, Chen Y, Yi Z, Zhong J (2019) Diagnosis of diabetic retinopathy using
deep neural networks. IEEE Access 7(c):3360–3370. https://doi.org/10.1109/ACCESS.2018.
2888639
60. Jiang H, Yang K, Gao M, Zhang D, Ma H, Qian W (2019) An interpretable ensemble deep
learning model for diabetic retinopathy disease classification. In: Proc. annu. int. conf. IEEE
eng. med. biol. soc. EMBS, pp 2045–2048.https://doi.org/10.1109/EMBC.2019.8857160
61. Qummar S et al (2019) A deep learning ensemble approach for diabetic retinopathy detection.
IEEE Access 7:150530–150539. https://doi.org/10.1109/ACCESS.2019.2947484
62. Gangwar AK, Ravi V (2021) Diabetic retinopathy detection using transfer learning and deep
learning, vol 1176. Springer, Singapore
63. Nguyen QH, et al (2020) Diabetic retinopathy detection using deep learning. In: ACM int.
conf. proceeding ser., pp 103–107. https://doi.org/10.1145/3380688.3380709
64. Khan Z et al (2021) Diabetic retinopathy detection using VGG-NIN a deep learning architecture.
IEEE Access 9:61408–61416. https://doi.org/10.1109/ACCESS.2021.3074422
65. Sebti R, Zroug S, Kahloul L, Benharzallah S (2022) A Deep Learning Approach for the Diabetic
Retinopathy Detection. In: 2022 2nd international conference on intelligent technologies,
CONIT 2022, no April, 2022, pp 459–469
66. Bilal A, Zhu L, Deng A, Lu H, Wu N (2022) AI-based automatic detection and classification
of diabetic retinopathy using u-net and deep learning. Symmetry (Basel) 14(7). https://doi.org/
10.3390/sym14071427
67. Chandrasekaran R, Loganathan B (2022) Retinopathy grading with deep learning and wavelet
hyper-analytic activations. Vis Comput. https://doi.org/10.1007/s00371-022-02489-z
68. Mondal SS, Mandal N, Singh KK, Singh A, Izonin I (2023) EDLDR: an ensemble deep
learning technique for detection and classification of diabetic retinopathy. Diagnostics 13(1):1–
14. https://doi.org/10.3390/diagnostics13010124
Healthcare Data Security Using AI
and Blockchain: Safeguarding Sensitive
Information for a Safer Society

Jyoti Upadhyay, Sunil Kumar Singh, Naresh Kumar Kar,

Manoj Kumar Pandey, Priyanka Gupta, and Preeti Tiwari

Abstract The synergy of AI and blockchain presents transformative opportuni-

ties, such as enabling secure sharing of medical data across institutions, facilitating
research collaborations, and streamlining insurance claims and billing processes.
Blockchain technology is known for its secure and transparent data management
addresses critical challenges in healthcare data interoperability, security, and privacy.
By creating a tamper-proof and decentralized ledger of patient records, medical
practitioners can access a patient’s comprehensive medical history, leading to more
informed decision-making. However, several hurdles including regulatory compli-
ance, technical integration, and ethical concerns must be overcome to fully harness
the potential of this combined approach. This chapter explores the current state, usage
of blockchain and AI in healthcare, trends of utilizing AI, benefits, challenges, and
future prospects of employing AI and blockchain in healthcare, highlighting their
potential to reshape the industry’s landscape and enhance patient care quality.

J. Upadhyay
GD Rungta College of Science and Technology, Bhilai, India
e-mail: [email protected]
S. K. Singh
CMREC, Hyderabad, India
e-mail: [email protected]
N. K. Kar
GITAM Deemed to be University, Hyderabad, India
e-mail: [email protected]
M. K. Pandey (B)
Pranveer Singh Institute of Technology, Kanpur, India
e-mail: [email protected]
P. Gupta
Research Scholar, Guru Ghasidas Central University, Bilaspur, India
e-mail: [email protected]
P. Tiwari
IBITF, India Institute of Technology, Bhilai, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 159
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_8
160 J. Upadhyay et al.

Keywords Blockchain · AI · Healthcare · Security · Privacy · Electronic health

record

1 Introduction

With increasing amount of data on the web, it is very essential to come up with some
technologies that can ensure the security and privacy on digital data. It is also very
important to note that the E-data related to the healthcare is utmost important to keep
it protected from unauthorized users. The form and sensitivity of digital data has
been transformed day by day and along with that the technology to misinterpret that
data is also improved so there is a need to improve the security of digital data using
the latest technology. The digital data is also very much aligned with the security of
the society and the nation therefore it is essential to provide maximum security to the
digital data. If we look into the past decades, our necessity is also increased to work
with the digital data due to the evolvement of artificial intelligence (AI) and with the
help of AI now we are more powerful and sophisticated to achieve some good results
and decision as compare to our traditional methods so both the terminology i.e. AI
and security of digital contents are equally important in order to be more effective
and precisely in any field [1]. The machine learning is term associated with the AI
that enables machine to think like human brains and act accordingly but to be more
effective and precise it needs enough digital data to train a model through which a
decision or result can be made [2]. When machine learning is not enough to reach
a decision and shows that it is less effective then deep learning is applied to draw
a decision by applying more complex calculations [3]. It is very clear as of now
that AI and security of digital data both are equally important for any society and
nation. Looking at the two or three decades back one can observe that the privacy
and integrity of digital data has become more vulnerable to digital attacks as the
decades passes because of the improvement in the technology so it is very clear the
with increasingly usages of the technology, the security of digital content is also at
high risk. This chapter includes the various aspects of how integration of AI and
blockchain can provide good infrastructure to build up good health infrastructures
for any nation.
There are various types of electronic data that is associated with a patient like
X-Ray data, Blood test data, MRI scan data, personal data, and many more, and all
these data need to be secured and processed effectively for better decision making.
Nowadays machine vision technique also plays an important role in the healthcare
industry by recognizing various objects and acting upon them like human does.
Machine vision is a technique which enables a machine to mimic computer as human
vison and to make them understand what object is placed in front of the machine [4].
Machine vision can promote the automation of system in case of emergency that can
save a human life. A patient data can be partition into text, image, audio and videos,
and for all these types of data different machine learning and deep learning based
application need to be developed or enhanced in order to automate the diagnosis
Healthcare Data Security Using AI and Blockchain: Safeguarding … 161

whenever needed. The digital data produced by patient in form of text and image
like prescription, written document, summary, lab report, and clinical reports are
very much important and these data can be effectively used for providing good
health advice and decision making if one can process it effectively. Natural language
processing (NLP) is a sub field of AI, which enable a machine to comprehend written
documents like a human does. With use of machine vision and NLP machine can
interpret, deduce, summarize, translate, and synthesize exact text for better human
diagnosis [5]. Image segmentation is also a type of processing in which medical image
is portioned into different regions in order to find region of interest [6]. Segmentation
plays an important role in medical diagnosis, and semantic segmentation is a type
of segmentation in which pixel belonging to one object is given one colour and
pixel belonging to another object is given other colour. Figure 1 shows the various
requirements of healthcare industry which shows that security and remote monitoring
are among the most requirements of the industry.
Audio data related to patient like heart-beat, crying, breathing, coughing etc.
plays an excellent role in diagnosis respiratory diseases, pulmonary diseases, and
cardiac diseases. AI based medical system can assist to analyse these sounds and
classifying audio according to their characteristics in order to diagnosis patient.
There is various state-of-art methods are present for audio signal processing that can
aid in medical industry. Though all these types of automation need the usages of AI
and machine learning concepts where the training data plays a significant role and
also the integration of these data is also an important concern. Collecting, analysing,
integrating and protecting all these forms of medical digital data are also an important

Fig. 1 Healthcare industry requirements

162 J. Upadhyay et al.

concern for the industry. There are few threats to enabling AI in the medical industry
are like, firstly, a patient data is very precious for any medical centre and securing
data and maintain the integrity of data needs specialized work force and resources.
Secondly integrating data from various resources in order to maximize the quantity
of training data is also a challenging task. It is also very obvious that these data may
be stolen or vulnerable to different attacks for different reasons therefore protecting
it along with the integration is a serious issue. Third, let us assume that training is
done for models and now AI is ready to support the clinical decision, but AI based
result are black box in nature and one cannot take risk of human life because of
AI recommendation therefore the simultaneous intervention of specialized doctors
is also needed. Fourth, there should be secure resource sharing for overcoming the
threat of the rouge devices. Fifth the sharing of knowledgeable data among the
researchers and clinical staff has also a security issues and threats. Hence we need to
work to overcome these threats collectively to make AI to assist medical diagnosis.
Basic terminologies of the chapter are as follows, and Sect. 1 covers the introduction
part, Sect. 2 covers the block chain in healthcare, Sect. 3 covers the related study,
Sect. 4 covers the analysis of various related work, and Sect. 5 contains the conclusion
and future work.

1.1 Artificial Intelligence (AI)

AI is a technique which enable a machine to think and mimic like human brains,
and it has now been a part of everyone life. The term artificial intelligence (AI)
describes how computers, particularly computer systems, may simulate human intel-
ligence processes. Learning, reasoning, problem-solving, perception, and language
comprehension are some of these processes. Artificial Intelligence (AI) seeks to
create machines that are capable of tasks that normally require human intelligence,
like pattern recognition, decision-making, and situational adaptation. In the context
of medical diagnosis, AI involves using advanced algorithms to analyse medical
data and assist healthcare professionals in diagnosing diseases, predicting outcomes,
and recommending treatments. The idea of building intelligent machines that could
express human intelligence was first investigated by researchers in the 1950s, which
is when artificial intelligence (AI) first emerged. The Dartmouth Workshop in 1956 is
credited with coining the phrase “artificial intelligence” [7], which officially estab-
lished AI as a discipline of research. Early AI research focused on symbolic AI,
using rules and logic to model human cognitive processes. However, progress was
slower than initially anticipated during the 1970 and 1980s, characterized by reduced
funding and enthusiasm due to unmet expectations. The field regained momentum
in the 1990s with advancements in machine learning and neural networks. After the
year 2000, with the rise of electronic health records and digital medical imaging, AI
began to make strides in medical diagnosis. Computer-aided detection systems were
developed to assist radiologists in identifying abnormalities in medical images. In
the year 2010, deep learning, a subset of AI, gained prominence due to improved
Healthcare Data Security Using AI and Blockchain: Safeguarding … 163

computational resources. Deep learning algorithms demonstrated remarkable capa-

bilities in image and speech recognition, which paved the way for breakthroughs in
medical image analysis. At present, AI in medical diagnosis has expanded to various
domains, including radiology (detecting tumours and fractures), pathology (identi-
fying cellular abnormalities), and genomics (analysing genetic data for disease risks).
AI-driven tools assist in early detection of conditions like diabetic retinopathy and
lung diseases, improving disease management and patient outcomes.

1.2 Blockchain Technology

Blockchain is a digital ledger that has no central administrator and records trans-
actions in an open, secure, and permanent manner. Since its initial introduction in
2008 as the core technology behind Bitcoin, it has been embraced by numerous
other industries for a variety of purposes [8]. Blockchain is a combination of peer-
to-peer communication and cryptographic technology. It is made up of a number
of blocks connected together by cryptographic hash functions. The blockchain is a
simple but brilliant way to distribute and receive data securely and automatically.
Block production by one of the parties is required to start a transaction. Thousands
of devices dispersed over the internet are verifying this block. After that, the verified
block is added to a chain and maintained online, producing a record that is distinct,
unique, and has a clear past. Therefore, on a blockchain, a transaction is regarded
valid when a block’s consensus through a contract is reached. Every member of a
blockchain network has a copy of the ledger because blockchain technology is decen-
tralized, which also means that trust is decentralized. Each block in a blockchain
network is made up of a collection of transactions that have been approved and
added to the network by a consensus mechanism. A chain of data that cannot be
changed is created by connecting the blocks in chronological order. Blockchain is
extremely secure because it is decentralised; no single entity controls the network.
The unchangeable nature of the ledger allows for speedy and effective transaction
verification, processing, and tracking, as well as convenient auditing of transactions.
Blockchain has established the ground work [9–12] for cryptocurrencies such as
Ripple, Bitcoin, and Ethereum, among others. Notably, Bitcoin is acclaimed as the
inaugural cryptocurrency. Given the contemporary reliance on digital authentication
for business transactions, blockchain introduces prospects for decentralized plat-
forms and services that are accessible to all. This innovative technology furnishes
avenues for conducting a diverse array of imaginative financial tools like micro-
payments and peer-to-peer lending, streamlining transactions while reducing associ-
ated costs. Each transaction functions as a digital block, necessitating consensus from
numerous participants within the network to be verified and incorporated. In response
to the susceptibilities witnessed by conventional databases, blockchain has emerged
as a safeguard, ensuring data security through cryptographic validation. Blockchain
technology [13–15] has undergone distinct evolutionary versions and each version
164 J. Upadhyay et al.

marking significant advancements. In its first iteration, Blockchain 1.0, the revolu-
tionary concept of decentralized ledgers birthed cryptocurrencies like Bitcoin, lever-
aging Proof of Work for consensus. Blockchain 2.0 followed, introducing smart
contracts and platforms like Ethereum, enabling self-executing agreements. The
third version, Blockchain 3.0, emphasized interoperability and scalability through
platforms like Polkadot. Privacy and security took the forefront in Blockchain 4.0,
exemplified by privacy-focused coins. The quest for sustainability led to Blockchain
5.0, championing energy-efficient consensus mechanisms. Presently, Blockchain 6.0
is emerging, integrating Artificial Intelligence and the IoT into the blockchain based
ecosystem, ushering in a new era of enhanced capabilities and applications across
industries.
While artificial intelligence may improve our ability to recognize and react
promptly to disease diagnoses, blockchain technology, which was initially devel-
oped to support the cryptocurrency ecosystem, is now being used in many other
industries to achieve extraordinary levels of security [16], improve the security of
medical records, and protect the privacy of record owners [17]. Blockchain tech-
nology is applicable to a number of industries, including e-commerce, cross-border
payments, digital identities, healthcare, IoT, and online voting. When utilised in data
management systems, blockchain technology can help solve important problems
with data transparency, traceability, immutability, auditing, safe data provenance, and
healthcare system confidence. Additionally, this can be effectively used to enhance
the management of medical records [18]. Figure 2 shows the various versions of
blockchain with its technology.

Fig. 2 Various versions of blockchain

Healthcare Data Security Using AI and Blockchain: Safeguarding … 165

2 Blockchain in Healthcare

Healthcare industry is one of the most prominent and significant area for any nation
and the failure of traditional methods has significantly leads to the adaption of
blockchain technology in the healthcare. The healthcare sector is going through
a revolutionary change thanks to blockchain technology, which has ushered in a
new era of data security, interoperability, transparency, and patient-centric treatment.
Traditionally plagued by fragmented systems, data silos, and security vulnerabilities,
healthcare has found a potent ally in blockchain’s decentralized and tamper-proof
architecture [19]. One of the most significant impacts of blockchain lies in data secu-
rity. Patient records, medical histories, and sensitive health information are vulnerable
to breaches and unauthorized access in centralized databases. Blockchain’s crypto-
graphic techniques offer a robust solution by ensuring that patient data remains
encrypted and accessible only to authorized individuals. Interoperability, a long-
standing challenge in healthcare, is another realm revolutionized by blockchain.
Health systems are often hindered by incompatible data formats and disparate infor-
mation sources, resulting in inefficient care coordination. Blockchain’s distributed
ledger eliminates this obstacle by providing a standardized, transparent platform
where healthcare providers, insurance companies, pharmacies, and even patients
can securely share and access information [20]. While blockchain’s potential in
healthcare is vast, challenges remain. Integration with legacy systems, scalability
concerns, and regulatory alignment require careful navigation. Collaborative efforts
among stakeholders, including healthcare providers, technology companies, and poli-
cymakers, are essential to fully harness blockchain’s potential. Figure 3 shows the
usages of blockchain in healthcare industry. Here one can notice that the stack holders
can utilize the cloud system for storing and retrieving of medical records using the
interface called blockchain hand-shaker, which ensure the secure way of communi-
cation by applying blockchain mechanism and cryptography if required. The inter-
face blockchain hand-shaker is also connected with the distributed ledger and smart
contact in bidirectional way for fetching any required details.
It is very clear that the involvement of AI and blockchain in healthcare industry
revolutionized the medical industry and the continuous development in these sectors
can boost the medical for the wellbeing of human kind. The convergence of Artifi-
cial Intelligence (AI) and blockchain technology holds the promise of a revolutionary
transformation in the medical healthcare system, addressing critical challenges and
unlocking new avenues for improved patient care, data management, research, and
operational efficiency. This synergistic approach harnesses the strengths of both
technologies to create a more secure, transparent, and patient-centric healthcare
ecosystem. Figure 4 shows the AI usages in healthcare system. Here one can observe
that the electronic patient record which is composed of audio, image and text can
be processed very efficiently using various AI technology. This figure also shows
the common applications and common challenges associated with healthcare system
using AI.
166 J. Upadhyay et al.

Fig. 3 Usages of Blockchain in healthcare

Fig. 4 Various usages of AI in healthcare

Blockchain technology combined with artificial intelligence (AI) has the potential
to transform the healthcare industry, overcoming obstacles and changing the game.
Through enhanced data security, interoperability, diagnostic precision, personalized
treatment, and streamlined research processes, this convergence empowers patients,
improves outcomes, and ushers in an era of efficient, transparent, and patient-centred
healthcare.
Healthcare Data Security Using AI and Blockchain: Safeguarding … 167

3 Related Study

In this segment, we will address several recent research investigations concerning the
security of healthcare data using the assistance of AI methodologies and blockchain
technologies. Challenges prevalent within the healthcare sector encompass issues
like interoperability, unavailable medical records, and the absence of thorough and
protected population health information. This chapter aims to identify cutting-edge
approaches utilizing AI techniques and blockchain technology to ensure healthcare
data security.
Shinde et al. [4] had delved into the notion that utilizing blockchain can enhance
the dependability and credibility of AI-driven healthcare. The proposed work indi-
cates that blockchain having the ability to tackle privacy and security concerns within
the healthcare community. Additionally, implementing blockchain could enable veri-
fying medical data and user origin. Furthermore, they identified specific defensive
strategies designed to counter certain adversarial attacks on healthcare data and also
mentioned how AI and blockchain technology together can be used for the healthcare
industry using various AI techniques.
Andrew et al. [21] have detailed several attributes and practical applications
of blockchain across various contexts, including its relevance to achieving inter-
operability in the healthcare sector. The discussion encompassed an overview of
blockchain architecture, platforms, and categorizations to aid in selecting an appro-
priate forum for healthcare purposes. They provided a thorough analysis to demon-
strate the blockchain technology’s importance to the healthcare sector from both an
application and a technological standpoint. Furthermore, the presentation contained
a comprehensive examination of security breaches targeting blockchain proto-
cols, including the classification of threat models. They also provided a compar-
ative assessment of detection and safeguarding methods. Various perspective of
using blockchain and AI along with is mentioned and safeguard measure of using
blockchain is also mentioned. Finally, they proposed several strategies to bolster the
security and confidentiality of blockchain networks.
Ali et al. [22] provided an architectural proposal comprising three distinct envi-
ronments: doctor’s, patient’s, and the meta-verse environment. Within this meta-
verse setting, doctors and patients engage with the support of blockchain technology,
ensuring the security, safety, and privacy of their information. The meta-verse envi-
ronment stands as the central component of the proposed architecture. By registering
on the blockchain, doctors, patients, and nurses enter this environment and assume
avatar forms. Every interaction and consultation between doctors and patients are
meticulously documented, encompassing diverse data such as images, speech, text,
videos, and clinical records. These data are then collected, transmitted, and securely
stored using blockchain technology. Explainable artificial intelligence (XAI) models
use the collected data to predict and diagnose diseases. Blockchain strengthens the
security of patient data while enabling information to be transparent, traceable, and
unchangeable. These intrinsic attributes of blockchain cultivate patient trust in their
168 J. Upadhyay et al.

data’s integrity, ultimately ensuring transparency, trustworthiness in disease diag-

nosis, and robust patient data security within the scope of this proposed architectural
framework.
Attaran [23] suggested the utilization of Blockchain solutions to enhance data
security and manage identities. This technology offers the potential to mitigate threats
and safeguard private information from unauthorized access. As data is added to
the blockchain, it becomes encrypted, rendering it immutable and indecipherable.
Transactions are validated through a private key, which is known exclusively to
the all the individual person. Unlike contemporary healthcare data systems, this
approach would restrict a healthcare provider’s access to a patient’s medical data
unless granted explicit permission through the blockchain record. Enhanced collab-
oration of data among providers enhances the accuracy of diagnoses, increases treat-
ment effectiveness, and enables healthcare facilities to provide cost-efficient care.
Through Blockchain, patient information can remain secure while still allowing
sharing with selected service providers. The technology establishes proof of medical
records ownership and ensures authenticity to counteract counterfeit techniques.
Farouk et al. [24] suggested novel attributes inherent to blockchain and delved
into the potential advantages and future trajectories of its integration into the health-
care domain. Healthcare entails combining blockchain with Internet of Things (IoT)
technologies, offering healthcare establishments a robust and precise system for
record management, a pivotal requirement. The suggested framework is segmented
into three contract categories: Registrar Contract (RC), Patient-Provider Relationship
Contract (PPR), and Summary Contract (SC). RC serves the purpose of associating
participants’ identification with their corresponding Ethereum address identity. PPR
is established between system nodes to facilitate the storage and administration of
patients’ medical records. SC retrieves patients’ complete medical history, show-
casing all past and ongoing interactions between participants and other nodes within
the system.
De Moraes Rossetto [25] put forward a framework for managing Healthcare Data
Privacy using Blockchain. They introduced a structure aimed at safeguarding the
confidentiality of medical information. This data is stored and exchanged within a
decentralized blockchain network. This protection is achieved by employing encryp-
tion methods such as RSA, ECC, and AES. Tests were conducted to assess how
cryptography influenced the suggested framework regarding computational work-
load, memory consumption, and processing duration. The findings reveal that the
primary impacts were on execution time and heightened computational burden when
transmitting data to the blockchain. These effects are reasonable, given the enhanced
privacy and security offered by the architectural design and encryption techniques.
Haddad et al. [26] presented a classification of suggested AI-Blockchain reme-
dies for EHR challenges. They extensively examined the advantages of implementing
Blockchain technology for health record administration. They also included a recent
examination of privacy and security concerns linked to EHR management. Further-
more, they proposed prospective avenues for research, aiming to establish a proficient
and dependable AI-Blockchain EHR ecosystem that surpasses existing centralized
Healthcare Data Security Using AI and Blockchain: Safeguarding … 169

models. This approach envisions a patient-focused data-sharing platform, granting

individuals full authority over their data.
Singh et al. [27] suggested strategies to enhance blockchain security by consoli-
dating pivotal aspects that could be leveraged to create diverse blockchain systems
and security utilities, thereby mitigating security susceptibilities. They elucidated a
range of blockchain based attacks prevalent in IoT based networks and presented
remedies for such vulnerabilities. Their paper systematically laid out a methodical
exploration of security constraints, weak points, hurdles, and complexities related
to blockchain technology. The same thorough treatment was given to security chal-
lenges within blockchain enterprises. Existing security resolutions for blockchain
in varied contexts were also elucidated. The discourse further delved into security
tools capable of addressing these vulnerabilities. A comprehensive analysis was
undertaken to underscore both the concerns related to security and advantages of
blockchain components, notably decentralization, that poses significant obstacles
concerning data privacy and transparency, leading to intricacies within the network.
Tanwar et al. [28] introduced a system framework and algorithm designed to adopt
a patient-centred strategy, facilitating access control policies through symmetric key
cryptography to diverse healthcare providers. They additionally suggested the utiliza-
tion of blockchain for a permission-oriented EHR sharing system, incorporating the
chain code concept. The proposed system was assessed to evaluate its alignment
with the requisites of patients, healthcare providers, and other stakeholders. Lastly,
an optimal strategy for enhancing the performance of the blockchain system was
identified, addressing metrics such as latency and throughput, network scalability,
and security considerations.
Hylock and Zeng [29] have presented a conceptual framework detailing the appli-
cation of blockchain technology to ascertain patients’ health status. This frame-
work emphasizes the integration of concurrent execution and AI within healthcare
networks. The approach focuses on evaluating the comprehensive well-being of
patients, including diagnosis and recovery procedures. Moreover, the study delves
into the analysis of pertinent surgical interventions, employing parallel operations and
computational assessments of clinical decision-making. By examining patient care
quality and diagnostic feasibility, both within real-world and simulated healthcare
systems, the proposed methodology underwent practical evaluation.
Feng et al. [30] provided a detailed analysis of the distributed blockchain database
using state machine replication, with atomic database changes known as transactions
organised into blocks and the transaction log’s integrity and tamper-resistance guar-
anteed by hash links between blocks. The idea of a blockchain was first suggested
in relation to Bitcoin and decentralised electronic money. The core blockchain
building blocks have been the topic of countless reports, and with the success of
Bitcoin, blockchain technology may now be utilised to conduct secure and reliable
transactions over unreliable networks without the need for a third party.
By integrating blockchain into several sources, Andoni et al. [31] analyses the
need for sharing of electronic healthcare record (EHR). It is designed primarily for
personal health data and electronic medical records, but it also investigates many
other types of data. They combined the application with on-chain and off-chain
170 J. Upadhyay et al.

authentication procedures to guarantee that it matched good standards for validity

and privacy. The adaption of blockchain for the interchange of EHR, obscurity, and
shield between clinical staff and organisations can be greatly increased.
Khurshid [32] has suggested adopting a similar mechanism, proposing the imple-
mentation of an inventive and organized infrastructure using Blockchain technology.
This innovative approach aims to enhance private patient records’ security and
address core data security concerns. Additionally, it seeks to establish a foundational
blockchain software for hospitals. The potential of Blockchain technology extends
to the medical and therapeutic domains as well. By employing blockchain solutions,
there is a possibility to store all clinical authorizations, schedules, and protocols on a
blockchain before the commencement of a clinical trial or examination. This would
result in more up-to-date, secure, and timestamped clinical trial data, which can be
readily accessed.
Wahl et al. [33] utilized a method to address or reply to an inquiry; these compu-
tational systems employ a diverse array of suggested algorithms, decision-making
skills, and extensive information. The technologies capable of assisting us in our
day-to-day existence encompass our surroundings. The efficiency of autonomous
automobiles, progress in medical investigations, and wearable gadgets tailored for
individuals have showcased the immense capability of technology and the remark-
able strides it has taken in the recent past. Jaiswal et al. [34] developed a GLSTM
model. The strength of this research lies in its creative combination of GANs and
LSTM, providing a robust framework for handling diverse datasets. The use of GANs
to generate synthetic PID diabetes data is particularly commendable, as it addresses
the challenges associated with limited real-world data.

4 Analysis of AI and Blockchain Technique Used

for Healthcare

The present progression of AI’s application in healthcare is expected to form a

cautious and well-balanced framework. Han et al. [35] have examined several
aspects of healthcare provision, indicating that AI, when combined with collabo-
ratively trained medical professionals, is likely to follow the current trajectory for
a considerable period. This trend demonstrates AI’s capacity for accurate patient
diagnosis. As depicted in Fig. 5, the year-by-year trends in AI’s healthcare utiliza-
tion reveal a continuous enhancement in AI technology, undeniably bolstering the
healthcare sector’s performance. However, this technological advancement comes
with an elevated risk to the security of medical record data, necessitating robust
quality measures.
The integration of AI and blockchain has undoubtedly strengthened every aspect
of the medical healthcare system as the evolvement of AI has brought various tech-
nologies together to process various electronic medical records like X-Ray, medical
Healthcare Data Security Using AI and Blockchain: Safeguarding … 171

Fig. 5 Trends of AI in healthcare year wise

prescription, ECG data, reports etc., in order to extract some useful pattern and infor-
mation related to the patient for automating the suggestion in case of emergency. It
also helps during various decision-making process related to the healthcare. As we
all know that the lots of data is being generated and every data related to the medical
healthcare is of utmost important and significant so blockchain is an emerging tech-
nology which ensures the exchange of data in efficient manner. There are various
related research papers published every year by Indian authors and authors from other
countries than then the India and it is seen that every year the number of publication
in this field is being increased rapidly. Figure 6 shows the worldwide publication in
the domain of healthcare using AI and blockchain.

Fig. 6 Worldwide publication details year-wise

172 J. Upadhyay et al.

Fig. 7 Publication by India authors year-wise

After observing Fig. 6 it is very clear that the number of publication in this domain
is continuously increasing worldwide and one can notice that in year 2017 the total
publication in this domain was 1780, but in 2018 it is increased by 124% than the year
2017. Likewise for the year 2019 it is increased by 77% than the 2018 to reach 7100
publication worldwide. For year 2020 the number of publication worldwide is 11600
with 63% of growth than the previous year. For year 2021 the total publication is
17200, which shows 48% growth than the previous year. Like for year 2022, the total
number of publication is 22000, which is 27% more than the previous year. For year
2023, till 25 August the total number of publication in this domain is 18600 and it
is expected to be continuously increasing. Figure 7 shows the number of publication
done by the Indian authors.
Looking at Fig. 6 one can easily say that the contribution done by the Indian
authors are good as well and continuously the growth is very good by the Indian
authors and for each year it is showing growth of more than 60%. Figure 8 shows a
comparison between the publication by India authors and publication by worldwide
authors and it shows a significant growth in every year which shows that publication
and research in this domain is continuously growing every year.
Table 1 contains the recent state-of-art related to healthcare industry which utilizes
the blockchain and the AI for providing better healthcare infrastructure. After going
through various literature and recent literature as mentioned in the Table 1, it is very
clear that there is room for the improvement in the technology and AI, but along with
this securing digital health related data is also a challenging issue.
Referencing to Table 1 it is very clear that the researchers have used different
blockchain technology like Consensus mechanisms specifically, Blockchain tech-
nology with smart contracts—Modified SHA256 for data security, with AI and IOT.
Healthcare Data Security Using AI and Blockchain: Safeguarding … 173

Fig. 8 Comparison between

2023 (Till 25 August)
numbers of publication
2022
Indian
2021
Worldwi
2020 de

2019

2018

2017

0 10000 20000 30000

As mentioned by Farahat et al. [43] the version 2.0 of blockchain is very fast to
create data block. We can say that Blockchain technology is an effective method for
enhancing the security of healthcare data and the use of blockchain in healthcare
ensures that sensitive data is protected from hackers and unauthorized access. The
integration of blockchain technology with IoT devices in the healthcare sector has
significantly impacted security, privacy, and efficiency.

5 Conclusion and Future Scope

The integration of AI and blockchain technologies has emerged as a promising solu-

tion to address the critical issue of healthcare data security, fostering a safer society in
the process. The exponential growth of sensitive healthcare data and the increasing
sophistication of cyber threats necessitate innovative approaches to safeguarding
this information. AI algorithms offer advanced data monitoring, anomaly detection,
and predictive analytics, enhancing the real-time identification of potential breaches.
Concurrently, blockchain’s decentralized and tamper-resistant nature ensures the
integrity, traceability, and confidentiality of healthcare data, thereby building trust
among stakeholders. The synergy of AI and blockchain not only bolsters the protec-
tion of patient records, diagnoses, and treatments but also streamlines data interop-
erability across healthcare institutions. This interoperability enhances the continuity
and quality of patient care, enabling timely and well-informed medical decisions.
Additionally, the collaboration between these technologies holds the potential to
mitigate the administrative burden, reduce costs, and improve the efficiency of health-
care systems globally. Future scope includes utilizing latest technology and security
mechanism to ensure good healthcare infrastructure.
174 J. Upadhyay et al.

Table 1 Comparison of recent state-of-the-art on healthcare using blockchain

Literature Objective Method used Result Limitation
Mahammad Blockchain Blockchain The paper proposes Security and
et al. [36] technology for security with a Blockchain-based storage
secure storage and smart contracts architectural challenges with
maintenance of Integration of framework for EHR
healthcare data but EHR data storing EHR data—excessive
does not mention (Electronic health administrative
the use of AI for data) data in and operative
healthcare data Bigdata storage costs for data
security systems management
Alruwaill Blockchain based Symmetric and Blockchain-based Not available
et al. [37] healthcare system asymmetric keys smart healthcare
using IoMT devices for data system using IoMT
for continuous concealment devices -
patient monitoring Hash function Location-based
and addresses SHA256 for data authentication for
security concerns alteration authenticity and
with the use of detection data source
symmetric and
asymmetric keys
Dayana The paper discusses Use of Solutions for Not available
et al. [38] healthcare data decentralized healthcare data
security using blockchain security using
blockchain technology blockchain
technology, but Implementation of technology
does not mention permissioned Benefits include
the use of AI blockchain with secure data storage
authorized access and reduced fear of
data theft
Devi et al. The paper discusses Blockchain Enhanced proof of Not available
[39] enhancing method for data stake increases
healthcare data security security in
security using Consensus blockchain
blockchain mechanisms: PoW Protects medical
technology, but (Proof Of Work), records from
does not mention PoS (Proof of hackers
the use of AI Stake), DPoS,
(Delegated Proof
of Stake)
Sharma Blockchain Blockchain-based Blockchain-based Enhances
et al. [40] technology Proposed IoT systems security, privacy,
enhances security of Application (PA), enhance security transparency,
healthcare data by Smart contracts and privacy and efficiency in
providing features for confidentiality, Proposed healthcare
like confidentiality, authentication, Application (PA) Offers a remedy
decentralization, and access control provides more that is more
and privacy effective solution efficient than
current plans
(continued)
Healthcare Data Security Using AI and Blockchain: Safeguarding … 175

Table 1 (continued)
Literature Objective Method used Result Limitation
Abdulatif This work proposes Conducting a Identification of Lack of
et al. [41] an AI and comprehensive security challenges integration of
blockchain-based review of security in smart healthcare modern
secure architecture challenges systems technologies in
to analyze malware Proposing an AI Proposal of AI and healthcare
and network attacks and blockchain-based security
on wearable devices blockchain-based secure architecture Research gap in
in the smart secure exploring
healthcare system architecture countermeasures
for security
challenges
Ali et al. The proposed Amalgamation of Proposed Not available
[42] architecture AI and blockchain architecture
combines AI and in the metaverse integrates AI,
blockchain in the Use of explainable blockchain, and
metaverse to ensure AI models for metaverse
data security and disease prediction Ensures trust,
privacy in and diagnosis security, and
healthcare transparency in
healthcare
Farahat The paper discusses Amalgamation of Proposed Standard privacy
et al. [43] the use of AI and blockchain architecture for techniques are
blockchain in the metaverse integrating AI and not secure
technology to Use of explainable blockchain in the enough
secure patient AI models for metaverse for Blockchain
records and ensure disease prediction healthcare version 2.0
privacy in and diagnosis Ensures performs better
healthcare transparency, trust, than subsequent
applications and data security in versions
healthcare
Sarker et al. It shows the Review of Review of Limitations of
[44] utilization of blockchain blockchain previous
blockchain technology in technology in approaches in
technology in healthcare healthcare healthcare
healthcare Evaluation of Identification of blockchain
applications, but healthcare limitations and technology
does not mention technologies future research
the use of AI in based on direction
conjunction with blockchain
blockchain
AlGhamdi The healthcare AI, IoMT, and Adoption of AI, Implementation
et al. [45] industry is adopting blockchain IoMT, and requires
AI and blockchain technologies in blockchain in collaboration
technologies to healthcare healthcare between multiple
enhance patient Role, applications, Advantages and stakeholders
outcomes, reduce obstacles, and challenges in Ensuring patient
costs, and improve future research implementing these privacy and data
operational areas are technologies security
efficiencies mentioned
176 J. Upadhyay et al.

Conflict of Interest There is no conflict of interest and all authors agree for the publication of the
chapter.

References

1. Nilsson N (2009) The quest for artificial intelligence: a history of ideas and achievements.
Cambridge University Press, New York. ISBN 978-0-521-12293-1
2. Ethem A (2020) Introduction to machine learning, 4th edn. MIT, pp xix, 1–3, 13–18. ISBN
978–0262043793
3. Ciresan D, Meier U, Schmidhuber J (2012) Multi-column deep neural networks for image
classification. In: 2012 IEEE conference on computer vision and pattern recognition, pp 3642–
3649. arXiv:1202.2745, https://doi.org/10.1109/cvpr.2012.6248110
4. Shinde R, Patil S, Kotecha K, Potdar V, Selvachandran G, Abraham A (2022) Securing AI-
based healthcare systems using blockchain technology: a state-of-the-art systematic literature
review and future research directions. https://arxiv.org/pdf/2206.04793
5. Bengio Y, Ducharme R, Vincent P, Janvin C (2003) A neural probabilistic language model. J
Mach Learn Res 3:1137–1155—via ACM Digital Library
6. Jaiswal S, Pandey MK (2021) A review on image segmentation. In: Rathore VS, Dey N, Piuri V,
Babo R, Polkowski Z, Tavares JMRS (eds) Rising threats in expert applications and solutions.
advances in intelligent systems and computing, vol 1187. Springer, Singapore. https://doi.org/
10.1007/978-981-15-6014-9_27
7. McCarthy J, Minsky M, Rochester N, Shannon C (1955) A proposal for the dartmouth summer
research project on artificial intelligence. AI Mag 27:12–14. http://www-formal.stanford.edu/
jmc/history/dartmouth/dartmouth.html
8. Narayanan A, Bonneau J, Felten E, Andrew M, Steven G (2016) Bitcoin and cryptocurrency
technologies: a comprehensive introduction. Princeton University Press, Princeton, New Jersey.
ISBN 978-0-691-17169-2
9. Armknecht F, Karame GO, Mandal A, Youssef F, Zenner E (2015) Ripple: overview and
outlook. Lect Notes Comput Sci 9229:163–180. https://doi.org/10.1007/978-3-319-22846-4_
10
10. Radziwill N (2018) Blockchain revolution: how the technology behind bitcoin is changing
money. Bus World 25(1). https://doi.org/10.1080/10686967.2018.1404373
11. Dannen C (2017) Introducing ethereum and solidity: foundations of cryptocurrency and
blockchain programming for beginners, vol 1. Springer. https://doi.org/10.1007/978-1-4842-
2535-6
12. Nakamoto S (2008) Bitcoin: a peer-to-peer electronic cash system. Decentralized Bus Rev
Article 21260
13. Lindman J, Rossi M, Tuunainen VK (2017) Opportunities and risks of blockchain technologies
in payments—a research agenda. In: Proceedings of the 50th Hawaii international conference
on system science, pp 1533–1542. https://doi.org/10.24251/HICSS.2017.185
14. Lundqvist T, Blanche A, De Andersson HRH (2017) Thing-to-thing electricity micro payments
using blockchain technology. In: Global Internet of Things Summit (GIoTS), Geneva,
Switzerland, pp 1–6. https://doi.org/10.1109/GIOTS.2017.8016254
15. Tosh DK, Shetty S, Liang X, Kamhoua CA, Kwiat KA, Njilla L (2017) Security implications
of blockchain cloud with analysis of block withholding attack. In: Proceedings 17th IEEE/
ACM international symposium on cluster, cloud and grid computing, CCGRID, pp 458–467.
https://doi.org/10.1109/CCGRID.2017.111
16. Yaeger K, Martini M, Rasouli J, Costa A (2019) Emerging blockchain technology solutions
for modern healthcare infrastructure. J Sci Innov Med 2(1):1. https://doi.org/10.29024/jsim.7
17. Kumar R, Arjunaditya SD, Srinivasan K, Hu Y-C (2023) AI-powered blockchain technology
for public health: a contemporary review, open challenges, and future research directions.
Healthcare 11:81. https://doi.org/10.3390/healthcare11010081
Healthcare Data Security Using AI and Blockchain: Safeguarding … 177

18. Singh S, Sharma SK, Mehrotra P, Bhatt P, Kaurav M (2022) Blockchain technology for efficient
data management in healthcare system: opportunity, challenges and future perspectives. https://
doi.org/10.1016/j.matpr.2022.04.998
19. McFarlane TD, Dixon BE, Grannis SJ (2016) Client registries: identifying and linking
patients. In: Health Information Exchange (HIE): navigating and managing a network of health
information systems, pp 163–182. https://doi.org/10.1016/B978-0-12-803135-3.00011-6
20. Neelakandan S, Beulah JR, Prathiba L, Murthy GLN, Irudaya Raj EF, Arulkumar N (2022)
Blockchain with deep learning-enabled secure healthcare data transmission and diagnostic
model. Int J Model Simul Sci Comput 13(04):2241006. https://doi.org/10.1142/S17939623
22410069
21. Andrew J, Isravel DP, Sagayam KM, Bhushan B, Sei Y, Eunice J (2023) Blockchain for health-
care systems: architecture, security challenges, trends and future directions. J Netw Comput
Appl. https://doi.org/10.1016/j.jnca.2023.103633
22. Ali S, Abdullah, Armand TPT, Athar A, Hussain A, Ali M, Yaseen M, Joo MI, Kim HC (2023)
Metaverse in healthcare integrated with explainable AI and blockchain: enabling immersive-
ness, ensuring trust, and providing patient data security. Sensors (Basel, Switzerland) 23(2):565.
https://doi.org/10.3390/s23020565
23. Attaran M (2022) Blockchain technology in healthcare: challenges and opportunities. Int J
Healthc Manag 15(1):70–83. https://doi.org/10.1080/20479700.2020.1843887
24. Farouk A, Alahmadi A, Ghose S, Mashatan A (2020) Blockchain platform for industrial health-
care: vision and future opportunities. Comput Commun 154:223–235. ISSN 0140-3664, https://
doi.org/10.1016/j.comcom.2020.02.058
25. De Moraes Rossetto AG, Sega C, Leithardt VRQ (2022) An architecture for managing
data privacy in healthcare with blockchain. Sensors 22(21):8292. https://doi.org/10.3390/s22
218292
26. Haddad A, Habaebi MH, Islam MH, Hasbullah NF, Zabidi SA (2022) Systematic review on AI-
blockchain based e-healthcare records management systems. IEEE Access 10:94583–94615.
https://doi.org/10.1109/ACCESS.2022.3201878
27. Singh S, Hosen ASMS, Yoon B (2021) Blockchain security attacks, challenges, and solutions
for the future distributed IoT network. IEEE Access 9:13938–13959. https://doi.org/10.1109/
ACCESS.2021.3051602
28. Tanwar S, Parekh K, Evans R (2020) Blockchain-based electronic healthcare record system
for healthcare 4.0 applications. J Inform Secur Appl 50:102407, ISSN 2214-2126. https://doi.
org/10.1016/j.jisa.2019.102407
29. Hylock RH, Zeng X (2019) A blockchain framework for patient-centered health records and
exchange (healthChain): evaluation and proof-of-concept study. J Med Internet Res 21
30. Feng Q, He D, Zeadally S, Khan MK, Kumar N (2019) A survey on privacy protection
in blockchain system. J Netw Comput Appl 126:45–58. https://doi.org/10.1016/j.jnca.2018.
10.020
31. Andoni M et al (2019) Blockchain technology in the energy sector: a systematic review of
challenges and opportunities. Renew Sust Energ 21(100):143–174. https://doi.org/10.1016/j.
rser.2018.10.014
32. Khurshid A (2020) Applying blockchain technology to address the crisis of trust during the
COVID-19 pandemic. JMIR Med Inform 8(9):e20477. https://doi.org/10.2196/20477
33. Wahl B, Cossy-Gantner A, Germann S, Schwalbe NR (2018) Artificial intelligence (AI) and
global health: how can AI contribute to health in resource-poor settings? BMJ Glob Health
3(4):e000798. https://doi.org/10.1136/bmjgh-2018-000798
34. Jaiswal S, Gupta P (2023) GLSTM: a novel approach for prediction of real & synthetic PID
diabetes data using GANs and LSTM classification model. Int J Exp Res Rev 30:32–45. https://
doi.org/10.52756/ijerr.2023.v30.004
35. Han C, Rundo L, Murao K et al (2020) Bridging the gap between AI and healthcare sides:
towards developing clinically relevant AI-powered diagnosis systems. In: IFIP advances in
information and communication technology, vol 584. Springer, Cham, pp 320–333. https://doi.
org/10.1007/978-3-030-49186-4_27
178 J. Upadhyay et al.

36. Mahammad AB, Kumar R (2023) Scalable and security framework to secure and maintain
healthcare data using blockchain technology. In: International conference on computational
intelligence and sustainable engineering solutions (CISES), Greater Noida, India, pp 417–423.
https://doi.org/10.1109/CISES58720.2023.10183494
37. Alruwaill AM, Mohanty SP, Kougianos E (2023) HChain: blockchain based healthcare data
sharing with enhanced security and privacy location-based-authentication. In: Proceedings
of the Great Lakes symposium on VLSI 2023 (GLSVLSI ’23). Association for Computing
Machinery, New York, NY, USA, pp 97–102. https://doi.org/10.1145/3583781.3590255
38. Dayana R, Vadivukkarasi K (2023) Healthcare data security using blockchain technology.
https://doi.org/10.1109/ICISCoIS56541.2023.10100572
39. Devi T, Kamatchi SB, Deepa N (2023) Enhancing the security for healthcare data using
blockchain technology. In: International conference on computer communication and infor-
matics (ICCCI), Coimbatore, India, pp 1–7. https://doi.org/10.1109/ICCCI56745.2023.101
28545
40. Sharma P, Namasudra S, Crespo RG, Parra-Fuente J, Trivedi MC (2023) EHDHE:
enhancing security of healthcare documents in IoT-enabled digital healthcare ecosystems using
blockchain. Inf Sci. https://doi.org/10.1016/j.ins.2023.01.148
41. Alabdulatif A, Khalil I, Saidur Rahman M (2022) Security of blockchain and AI-empowered
smart healthcare: application-based analysis. Appl Sci. https://doi.org/10.3390/app122111039
42. Ali S, Abdullah, Armand TPT, Athar A, Hussain A, Ali M, Yaseen M, Joo M-I, Kim HC
(2023) Metaverse in healthcare integrated with explainable AI and blockchain: enabling
immersiveness, ensuring trust, and providing patient data security. https://doi.org/10.3390/
s23020565
43. Farahat IS, Aladrousy W, Elhoseny M, Elmougy S, Tolba AE (2022) Improving healthcare
applications security using blockchain. Electronics. https://doi.org/10.3390/electronics1122
3786
44. Sarker B, Sharif N, Rahman MA, Parvez AHM (2023) AI, IoMT and blockchain in healthcare.
J Trends Comput Sci Smart Technol 5:30–50. https://doi.org/10.36548/jtcsst.2023.1.003
45. AlGhamdi R, Alassafi MO, Alshdadi AA, Dessouky MM, Ramdan RA, Aboshosha BW (2022)
Developing trusted IoT healthcare information-based AI and blockchain. Processes. https://doi.
org/10.3390/pr11010034
Future of Electronic Healthcare
Management: Blockchain and Artificial
Intelligence Integration

Parag Verma, Chakka Mohana Rao, Prudhvi Kumar Chapalamadugu,

Rajeev Tiwari, and Shuchi Upadhyay

Abstract Blockchain and artificial intelligence (AI) technologies have emerged as

innovative solutions in the healthcare sector. Using a thorough risk management
strategy and a generalizable analytical technology, this research study explores how
the integration of these technologies may have a transformational effect on health-
care. Many reliable sources, like the Web of Science and Google polls carried out
by regulatory authorities, were examined to compile statistics on healthcare indices.
The evaluation concentrates on several facets of blockchain and AI and emphasizes
the advantages of merging these technologies in the healthcare industry. The study
addresses how to create trustworthy artificial intelligence models for e-Health by
utilizing blockchain, an open network for secure information sharing and authoriza-
tion. The integration of these cutting-edge technologies can result in improved service
efficiency, lower costs, and democratized healthcare by giving healthcare profes-
sionals access to the blockchain to display patient medical records and using AI’s
proposed algorithms and decision-making capabilities, along with vast amounts of
data. Additionally, blockchain makes storing the encrypted data needed for AI appli-
cations easier. By utilizing the breakthroughs in blockchain and AI, this research
offers a vision for the future of healthcare delivery, paving the road for a more
effective and inclusive healthcare system.

Keywords Healthcare · Risk management · Service efficiency · Cost reduction ·

Democratized healthcare

P. Verma (B) · C. M. Rao · P. K. Chapalamadugu

Datafoundry Private Limited, Bengaluru, Karnataka 560080, India
e-mail: [email protected]
R. Tiwari
IILM University, Greater Noida 201306, India
S. Upadhyay
Department of Allied and Health Sciences, School of Health Science and Technology, UPES,
Dehradun 248007, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 179
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_9
180 P. Verma et al.

1 Introduction

1.1 A Subsection Sample

According to Alotaibi and Federico [1], healthcare systems are complicated and
comprise a variety of procedures, workflows, and patient-care activities. For health-
care providers to be successful, internal controls must be improved, performance,
compliance, and consistency must be improved, and risk, workload, and overhead
must be decreased [2]. This study suggests a healthcare smart contract structure to
handle these issues based on cutting-edge healthcare blockchain analysis and a solid
method of healthcare administration [3].
Integration of technology like blockchain and artificial intelligence (AI) is
becoming increasingly important as governments and corporate sectors digitize
healthcare systems [4, 5]. This integration intends to promote medical research,
achieve patient centricity, and push novel data processing methodologies for better
results [6]. AI may be very useful in medication monitoring, prioritizing essential
patients, and data-driven decision-making [7]. The repurposing of pharmaceuticals
and accurate dosage measurement in clinical trials are also made possible by AI
and numerical drug design methodologies [8]. Governments must make use of their
resources and push for change in the fast-changing healthcare industry while also
assuring uniformity, compliance, and data protection [9]. Medical professionals,
healthcare providers, and payers may receive fast updates thanks to blockchain
technology’s safe and automated mechanism for storing health-related data.
Additionally, the integration of blockchain and AI algorithms allows AI to
discover and study patterns and trends in the field of health. AI can efficiently analyze
unstructured data and collect data from a variety of sources, including radiologists and
patients. Despite the potential advantages of AI, there are worries among healthcare
professionals about its effects on patient well-being and the need for prudence in its
use [10]. While AI has proven its worth in a number of fields, such as autonomous cars
and fraud detection, its maturity level in the healthcare sector is still being investigated
[11]. This study explores how blockchain technology and artificial intelligence may
be used to enhance patient outcomes, data security, and healthcare administration.

2 The Blockchain: Advancements and Applications

in Healthcare

The blockchain has drawn a lot of interest since it was first introduced as the founda-
tional technology behind Bitcoin, decentralized electronic money. It offers integrity,
tamper resistance, and trust in a distributed context, presenting a novel data manage-
ment paradigm. The blockchain functions as a distributed database that makes use of
replication using state machines. Blocks are collections of transactions, which repre-
sent atomic modifications to the database. A chain of transactions is then created by
Future of Electronic Healthcare Management: Blockchain and Artificial … 181

connecting these blocks together using cryptographic hash connections. The main
goal of the blockchain is to create a transparent, decentralized system for recording
and validating transactions. The blockchain does not rely on a single authority or
middleman for transaction confirmation and security, in contrast to conventional
centralized databases. Instead, it uses a network-wide consensus technique to reach
a consensus on the state of the database. The blockchain’s capacity to guarantee the
consistency and tamper-resistance of the transaction log is a key feature. A strong
cryptographic connection is made possible by the fact that each block in the chain
holds a hash of the one before it. Any modification to a prior block would cause a
mismatched hash, instantly alerting the network of possible meddling. Any modifi-
cation to a prior block would cause a mismatched hash, instantly alerting the network
of possible meddling. The immutability and openness of the blockchain contribute
to its high level of security and trust. The blockchain has the potential to transform
several industries, including banking, supply chain management, and healthcare, in
addition to its use in decentralized electronic currency. It is the perfect option for
ensuring data integrity, enabling safe transactions, and boosting confidence in digital
ecosystems because of its transparency and immutability.
The essential components of blockchain technology have been the subject of
substantial research and reporting [12]. A sequential collection of blocks that include
complete and accurate transaction data make up the core of the blockchain. A connec-
tion between these blocks and the preceding block, which is commonly represented
by a hash value, connects them and creates a chain-like structure [13]. The block
of Genesis is the first block in the chain, and the block header is used to identify
each succeeding block [14, 15]. Due to blockchain’s adaptability, there is increasing
interest in a wide range of applications, including data storage, financial markets,
computer security, the Internet of Things (IoT), nutritional science, healthcare, and
brain research. Blockchain technology has gained a lot of attention, especially in the
healthcare industry, and has made major strides in ensuring the safe and trustworthy
monitoring of medical records. By providing individualized, dependable, and secure
access to real-time clinical records and enabling a thorough and current perspective
of a patient’s well-being inside a secure healthcare ecosystem, it has the potential to
change healthcare in the future [16].
By integrating artificial intelligence healthcare networks and running many trans-
actions concurrently, the use of a blockchain platform in healthcare intends to eval-
uate the health state of patients’ ailments. The suggested strategy entails assessing
the general health of patients, providing correct diagnoses, assessing the recovery
system, and researching pertinent surgical treatments through concurrent procedures
and computer studies in clinical decision-making. With the aid of this thorough tech-
nique, it is possible to evaluate the effectiveness of the patient care given and the
viability of making correct diagnoses. The efficiency and practical applicability of the
suggested strategy have been tested in both actual and simulated healthcare systems
[17].
A customized blockchain-oriented healthcare knowledge-sharing network called
Blo CHIE was created by researchers [18]. By combining blockchains from many
182 P. Verma et al.

sources, this platform focuses on assessing the requirements for transferring health-
care data, such as personal health data, electronic medical reports, and several
other forms of data. The program combines both on-chain and off-chain authen-
tication procedures to guarantee the accuracy and privacy of the data transferred.
Blockchain technology makes it possible to improve data safety, anonymity, and
medical record interchange between clinical specialists and healthcare organizations.
Further evidence for the possibility of substantial advancements in the application of
blockchain in the healthcare sector can be found in the study undertaken by Andoni
et al. [19].
The adoption of a related mechanism that emphasizes a systematic and cutting-
edge infrastructure utilizing blockchain technologies to improve the security of
private patient records, address fundamental data protection concerns, and establish
a thorough blockchain software framework inside a hospital setting was proposed by
Cryan, M.A. [20]. Blockchain technology has shown a lot of promise in the domains
of pharmaceuticals and biological science. It is now feasible to use blockchain tech-
nology to keep all clinical clearances, schedules, and protocols on a blockchain even
before clinical research or trial begins. This strategy makes sure that crucial infor-
mation about clinical trials is current, safe, time-stamped, and transparent, which
improves the efficiency and openness of the research process.

3 Blockchain in Healthcare: e-Health Use Cases

Blockchain in healthcare offers promising e-Health use cases, revolutionizing the

industry. Its decentralized and secure nature ensures patient data integrity, stream-
lined medical record management, and efficient supply chain tracking. Smart
contracts enable automated insurance claim processing, while telemedicine platforms
benefit from secure and transparent patient-doctor interactions. Figure 1 showcases
high-potential blockchain use cases in the healthcare industry.

3.1 Clinical Trials and Research

By securely storing and distributing trial data, blockchain has the potential to simplify
the process of conducting clinical trials. It enables effective and transparent data
gathering, participant consent management, and data exchange across researchers,
assuring data integrity and lowering fraud. For instance, Patientory is a blockchain-
based platform that provides a complete solution for storing and managing patients’
medical records and data. Patientory uses blockchain technology to store patient data
in a decentralized, tamper-proof way, ensuring the confidentiality and privacy of the
data. The platform gives patients more control over their own health data and offers
them individualized medical care based on their unique requirements. Additionally,
Patientory can help with the integration of health insurance programs, enabling users
Future of Electronic Healthcare Management: Blockchain and Artificial … 183

Fig. 1 High-potential blockchain use cases in the healthcare industry

to view and manage their insurance data without any difficulty. Another platform
powered by blockchain that focuses on storing patient medical records and other
data on the blockchain is MediBloc. MediBloc guarantees the safety and integrity
of patient data by utilizing the built-in security features of blockchain technology.
In addition to serving as a safe place to store medical documents, the platform also
provides users with assistance and other services like medical consultations and
insurance alternatives. By giving patients access to a complete platform that satisfies
their medical requirements, data security concerns, and access to multiple healthcare
providers, this integrated approach improves the overall healthcare experience. The
transformational potential of blockchain technology in healthcare is highlighted by
both Patientory and MediBloc. These systems improve data security and privacy,
give patients authority over their health information, and provide individualized
medical services and insurance alternatives by utilizing the blockchain’s decentral-
ized and transparent nature. These developments support better patient outcomes
and experiences by developing a healthcare ecosystem that is more patient-centric
and effective. The review articles by Shah and Garg [21] and Pegoraro et al. [22]
give a general summary of how digital technologies have affected clinical trials.
The researchers talk about how to better data collecting, patient monitoring, and
remote data analysis by wearable technology, mobile health applications, and elec-
tronic data recording. They draw attention to the difficulties and possibilities that
184 P. Verma et al.

digital technology brings for improving the efficacy and efficiency of clinical trials.
A systematic study that looks at the ethical issues in clinical research was done by
Alemayehu et al. [23] and de Jongh et al. [24]. Key difficulties such as informed
permission, data privacy, participant recruitment, and conflict of interest are identi-
fied by the researchers through analysis of diverse studies and ethical frameworks.
They shed light on the moral dilemmas that confront scientists and make suggestions
for upholding moral principles in clinical trials. A study on novel methods for patient
recruitment in clinical trials was undertaken by Peipert et al. [25]. To reach a larger
and more varied patient group, the researchers talk about using social media, internet
platforms, and tailored advertising. They discuss the benefits and drawbacks of each
approach to patient recruitment and offer doable suggestions for enhancing patient
recruitment and retention in clinical trials. An investigation of the regulatory frame-
works governing clinical trials in various nations was done by Welch et al. [26]. The
researchers examine the legal procedures, moral standards, and times for approval in
various countries. They talk about the regulatory standards’ parallels and variations
and provide information on the difficulties and chances associated with performing
international clinical trials.

3.2 Electronic Health Records (EHR) Management

Blockchain improves EHR systems by collecting pertinent data, identifying mistakes,

and automating repetitive processes. The decision-making process enhances data
accuracy, enables data integration, and helps with clinical documentation and coding.
For instance, Ciox Health employs blockchain to increase the effectiveness of medical
record retrieval. Their AI-powered technology examines EHRs to find pertinent
patient records, saving time and effort compared to manual record retrieval. Studies
on the difficulties and factors to be taken into account when integrating blockchain
technology into EHR systems have been done by researchers [27, 28]. Scalability,
interoperability, regulatory compliance, and data governance are just a few of the
important topics they explore. The authors address the challenges of implemen-
tation by offering insightful research into potential fixes and suggestions for the
successful integration of blockchain in EHR administration. Researchers [10, 29]
launch Mediledger, a blockchain-based platform created with the express purpose of
connecting EHR and pharmaceutical supply chain management systems. The bene-
fits of using blockchain technology to guarantee the authenticity and traceability of
pharmaceutical items are discussed by the writers. Additionally, they go through how
Mediledger may improve patient safety and drug tracking while addressing impor-
tant facets of pharmaceutical supply chain integration. The systematic review of the
use of blockchain technology for EHR interoperability is carried out by Schmeelk
et al. [28] and Maitra et al. [30]. The authors analyze the potential of blockchain in
promoting smooth data interchange and sharing among healthcare institutions via the
examination of numerous research and projects. In tackling interoperability issues
within EHR systems, the assessment highlights the importance of blockchain, which
Future of Electronic Healthcare Management: Blockchain and Artificial … 185

presents a viable alternative for enhanced data interoperability and cooperation in

healthcare settings.

3.3 Drug Supply Chain Management

Blockchain enhances supply chain transparency and traceability for pharmaceuticals.

It makes it possible to follow a drug’s path from its maker to the patient, lowering
the chance that it will be counterfeited and verifying the genuineness and safety of
medicines. An illustration of blockchain-based medicine supply chain management
is IBM’s medicine Supply Chain Integrity Solution. The application of blockchain
technology in managing drug recalls was the subject of a systematic review by Han
et al. [31]. They looked at numerous research papers and initiatives that investi-
gated how blockchain technology may be used to enhance the speed and accuracy
of drug recall procedures. The assessment focused on how blockchain technology
may improve coordination, communication, and traceability during medication recall
incidents.
Blockchain technology was the subject of a study [32] that concentrated on
safe medicine authentication. They suggested a blockchain-based system that guar-
antees the integrity and validity of pharmaceutical items all the way through the
supply chain. The benefits of adopting blockchain to thwart medicine fraud and raise
patient safety were outlined by the researchers. Additionally, they implemented the
suggested system as a proof-of-concept and assessed it. Clauson et al. [33] gave
a summary of the application of blockchain technology in the management of the
medication supply chain. The advantages of using blockchain to improve trace-
ability, transparency, and security in the pharmaceutical supply chain were outlined
by the researchers. They looked at a few blockchain applications, including supply
chain visibility, inventory management, and product verification. The report also
discussed the difficulties and potential outcomes of using blockchain technology in
the management of the medication supply chain. The difficulties and factors to be
considered while integrating blockchain technology into the pharmaceutical supply
chain were examined by Humayun et al. [34]. The researchers looked at things
including scalability, regulatory compliance, stakeholder confidence, and data inter-
operability. They offered perceptions of prospective remedies and suggestions for
the effective application of blockchain in the management of the medication supply
chain.
The relevance of blockchain in allowing track and trace systems in pharma-
ceutical supply chains was investigated by researchers [35, 36]. The benefits of
using blockchain to provide real-time transparency and accountability in drug traf-
ficking were outlined by academics. To integrate blockchain into current supply chain
management systems, they examined the implementation problems and potential
solutions. Collectively, these research projects advance our knowledge of the poten-
tial uses of blockchain technology in the management of the medication supply chain.
186 P. Verma et al.

They address a few issues, including managing medicine recalls, supply chain prob-
lems, secure authentication, transparency, and traceability. These research results
and insights offer important information and pointers for developing blockchain
technology’s use in the pharmaceutical sector.

3.4 Telemedicine and Remote Patient Monitoring

For telemedicine consultations and remote patient monitoring, blockchain makes

it possible to securely and decentrally share patient health data. Between patients
and healthcare providers, it assures data protection, interoperability, and consent
management. Blockchain telemedicine platform examples include Gem’s GemOS.
Using a literature study, Liu et al. [37] created a summary of telemedicine and remote
patient monitoring in the medical field. They investigate the advantages of using
these technologies as well as the difficulties that may arise. The researchers look at a
few studies and programs that have looked into how telemedicine and remote patient
monitoring might be used to increase patient outcomes, cost-effectiveness, and access
to healthcare. The paper emphasizes how telemedicine and remote patient monitoring
have the potential to revolutionize healthcare delivery by increasing accessibility,
better patient outcomes, and maximizing resource use. A study report on the ethical
issues of telemedicine and remote patient monitoring was released by Field and
Grigsby [38]. They examine the ethical issues raised by telemedicine activities in
relation to patient autonomy, data security, privacy, and informed permission. To
guarantee responsible and patient-centered use of telemedicine and remote patient
monitoring, the researchers talk on the significance of ethical frameworks, rules, and
policies. The study offers insights into the most effective ways to uphold moral norms
in telemedicine, safeguard patient rights, and preserve patient-provider confidence.
A study was undertaken by researchers [39] to look into the use of telemedicine
in rural and underserved regions. The difficulties that patients and healthcare profes-
sionals experience while trying to receive medical care in rural areas are highlighted
by the study. They talk about how telemedicine might improve access to treatment,
lower healthcare inequities, and close the healthcare gap in rural regions. The paper
offers suggestions for the effective adoption and implementation of telemedicine in
rural healthcare settings, taking into account issues with infrastructure, education,
and funding. A literature study on the financial effects of telemedicine and remote
patient monitoring was undertaken by Lloyd and Lee [40] and Niu et al. [41]. The
economic effects, cost-effectiveness, and reimbursement practices of utilizing these
technologies in healthcare are all examined by the researchers. From an economic
standpoint, they talk about the difficulties and advantages that telemedicine and
remote patient monitoring have for healthcare practitioners and systems. The anal-
ysis sheds light on the advantages and factors that should be taken into account in
terms of cost reductions, resource allocation, and long-term financial viability.
Future of Electronic Healthcare Management: Blockchain and Artificial … 187

3.5 Health Insurance and Claims Processing

Blockchain simplifies procedures related to health insurance, such as fraud detection

and claims handling. Insurers, healthcare providers, and patients may share patient
data in a safe and transparent manner, which lowers administrative costs and boosts
productivity. Health insurance options based on blockchain have been investigated
by Aetna and Humana. A literature review that [42] did gives an overview of health
insurance and claims processing in the healthcare sector. The difficulties and ineffi-
ciencies in the current health insurance systems and claim-processing practices were
investigated by the researchers. They talked about how new technology, shifting
government regulations, and policy changes might affect how quickly, accurately,
and transparently health insurance and claims are processed. To promote affordable
and patient-centered healthcare, the assessment also looked at the possible advan-
tages of value-based payment models and alternative payment strategies. A study
on the application of blockchain technology to the processing of health insurance
claims was released in 2021 by Abekah-Nkrumah et al. [43]. They talked about the
usage of smart contracts, claims adjudication, and safe, transparent transactions on
the blockchain in the insurance sector. The potential of blockchain to lower fraud,
simplify billing procedures, and increase the precision of claims processing was
investigated by the researchers. The study highlighted the potential for blockchain to
alter existing procedures by shedding light on the opportunities and problems related
to its implementation in health insurance and claims processing.
Electronic Health Records (EHRs) and systems for processing health insurance
claims were the subject of a study by researchers [44, 45]. The researchers looked
at the possible advantages of integrating EHR data with claims processing practices,
including strengthening payment accuracy, lowering billing mistakes, and increasing
coding accuracy. Data privacy, interoperability, and regulatory compliance were some
of the topics they covered while talking about the difficulties and factors involved
in integrating EHRs and claims processing systems. The study shed light on the
possibility of EHR integration in expediting the handling of health insurance and
claims, resulting in more accurate and efficient billing operations.
A study on the use of data analytics in the processing of medical claims was
published by a researcher [46]. The researchers looked at how data analytics methods
may be used to examine huge amounts of insurance data, find fraud, and enhance
billing accuracy. They emphasized how data analytics may improve revenue cycle
management, spot billing mistakes, and boost financial performance in healthcare
enterprises. The study revealed the prospects and difficulties of using data analytics in
health insurance and claims processing, highlighting its potential to raise the general
efficacy and precision of these procedures. A study of the literature was undertaken
by Ho et al. [47] with a focus on the changing environment of health insurance
and claims processing. The researchers looked at how technology improvements,
payment reform initiatives, and legislative changes affected health insurance systems
and claims processing practices. They talked on the transition to value-based payment
models, the development of alternative payment options, and the significance of
188 P. Verma et al.

pricing transparency in the healthcare industry. The review offered vital light on
the reasons underlying the evolution of these processes while revealing patterns and
future directions in health insurance and claims processing.

3.6 Medical IoT and Device Data Management

The integrity and security of data produced by medical Internet of Things (IoT)
devices are improved by blockchain. It allows for the safe storage and transfer of
device data, preserving data security and integrity, and promoting device interoper-
ability. One example of a blockchain-based system for managing medical IoT data is
Chronicled’s MediLedger. A literature study that offers an overview of Medical IoT
(Internet of Things) and device data management was undertaken by Baumfeld Andre
et al. [48] and Pradhan et al. [49]. The adoption of IoT devices in healthcare and its
effects on data management procedures were investigated by the researchers. They
talked about how IoT can improve remote patient monitoring, enable real-time data
collecting, and monitor patient health. The evaluation also covered the issues with
interoperability, data integration, data security, and privacy in medical IoT contexts.
A study article concentrating on the security and privacy issues in Medical IoT and
device data management was released by Frikha et al. [50] and Azbeg et al. [51].
Their analysis focused on the security and privacy of patient data as well as other
risks and vulnerabilities linked to the usage of IoT devices in healthcare. To protect
patient data in Medical IoT contexts, the researchers talked about the significance
of secure communication protocols, data encryption, authentication systems, and
access control. The study included recommendations for the safest ways to protect
the privacy and security of medical IoT data. The interoperability issues in medical
IoT and device data management were investigated in a study by researchers [52].
The necessity for standardized communication protocols and data formats was inves-
tigated to facilitate smooth integration and data sharing across various IoT devices
and systems. To facilitate data interoperability in Medical IoT contexts, they explored
the role of healthcare standards and interoperability frameworks. To facilitate effi-
cient data exchange and usage throughout the healthcare ecosystem, the research
offered solutions for tackling interoperability difficulties in device data management.
The application of data analytics in medical IoT and device data management was
covered in a paper by Abounassar et al. [53]. The researchers investigated the use of
data analytics tools to examine the massive amounts of data produced by IoT devices
in the healthcare industry. They talked about how data analytics may be used to iden-
tify abnormalities, forecast health outcomes, and extract insightful information. The
study demonstrated how data analytics may enhance decision-making, customized
medicine, and patient outcomes in Medical IoT contexts, allowing healthcare prac-
titioners to get insightful knowledge and enhance patient care. A literature study on
the ethical issues of medical IoT and device data management was undertaken by
Pradhan et al. [49]. The researchers looked at the ethical issues surrounding data
privacy, permission, ownership, and openness in IoT contexts for medical purposes.
Future of Electronic Healthcare Management: Blockchain and Artificial … 189

To ensure the acceptable and ethical use of Medical IoT data, they emphasized the
significance of informed permission, data governance systems, and ethical norms.
The evaluation shed light on the ethical implications and issues that should be consid-
ered while managing device data in the healthcare industry, placing emphasis on the
necessity of giving patient privacy and data protection a priority.

3.7 Personalized Medicine and Genomics

Blockchain enables the safe management and exchange of genetic data for individu-
alized medication. While protecting privacy and data ownership, it enables patients
to manage their genetic data, share it with researchers, and take part in genomic
research projects. An illustration of a blockchain-based genomics platform is Nebula
Genomics. Rajeswari and Ponnusamy [54] and Hong and Oh [55] did a study of the
literature to give an overview of genomics and personalized medicine. The use of
genetic information to customize medical interventions and therapies for specific
individuals was investigated by the researchers. The development of genomic tech-
nology, such as next-generation sequencing, and its implications for personalized
treatment were highlighted. The study looked at several research and programs that
addressed the use of genetics in illness diagnosis, prognosis, and therapy choice,
underlining the potential of genomics to change healthcare by enabling more targeted
and precise treatments. Beccia et al. [56] and Offit [57] released a study on the
ethics of genetics and customized medicine. They looked at the moral issues raised
by genetic discrimination, data privacy, and genetic testing. In order to ensure the
responsible and fair use of genetic data in personalized medicine, the researchers
stressed the significance of ethical frameworks, rules, and regulations. The study
highlighted the necessity for patient autonomy, well-informed decision-making, and
privacy protection while offering insights into best practices for upholding ethical
norms and defending patient rights in the context of genomics. A study on the use
of genomes and clinical data for customized medicine was carried out by Santaló
and Berdasco [58]. They looked at the difficulties and possibilities of fusing genetic
data with clinical data and electronic health records. The researchers talked about
how integrated data analysis may help with illness risk assessment, therapy response
assessment, and medication development. To enable the application of genomics in
personalized medicine and enable the smooth integration of genetic information into
clinical practice, the study underlined the necessity for interoperability and data inte-
gration standards. A study of the literature was undertaken by researchers [59] with
a focus on the financial effects of genomics and customized medicine. They looked
at the economic effects, cost-effectiveness, and reimbursement practices of inte-
grating genomics into clinical practice. From an economic standpoint, the researchers
explored the difficulties and advantages of implementing customized medical tech-
niques, taking into account things like healthcare costs, resource allocation, and
reimbursement models. In order to maximize the value of genomic medicine, the
study offered insights into the possible advantages and considerations for adopting
190 P. Verma et al.

personalized medicine and genomics in healthcare systems. It also highlighted the

necessity for evidence-based decision-making and effective resource allocation.

3.8 Healthcare Payments and Billing

Blockchain reduces administrative expenses and eliminates middlemen in the billing

and payment of medical services. It makes it possible for payers, providers, and
patients to transact in a secure and open manner, increasing productivity and lowering
billing mistakes. A blockchain-based platform for healthcare provider payments and
credentialing is called ProCredEx. A literature study undertaken by Veenstra et al.
[60] offers a summary of the procedures for paying for and billing for medical
services. The difficulties and inefficiencies in the present healthcare payment systems
were investigated by the researchers. They talked about how electronic payment
systems and blockchain, two new technologies, can increase the speed, accuracy,
and security of healthcare transactions. The review also looked at the potential
advantages of value-based payment models and alternative payment strategies in
promoting affordable and patient-centered healthcare, highlighting the need for
creative methods of billing and reimbursement to better align incentives and advance
the entire healthcare system. A study on the use of blockchain technology in health-
care billing and payments was released by Meyer [61]. They talked about smart
contracts, the processing of medical claims, and the usage of blockchain for safe and
transparent transactions. The possibility of using blockchain to lessen fraud, reorga-
nize billing procedures, and boost claims adjudication accuracy was investigated by
the researchers. The study highlighted the significance of data integrity, privacy, and
security in the digital healthcare ecosystem and offered insights into the prospects
and problems of integrating blockchain in healthcare payments. The integration of
electronic health records (EHRs) with healthcare payment and billing systems was
the subject of research by researchers [62]. They looked at the possible advantages of
integrating EHR data with billing procedures, including increased coding precision,
decreased billing mistakes, and improved reimbursement accuracy. The researchers
talked about interoperability, data privacy, and regulatory compliance as well as the
difficulties and factors to be considered when integrating EHRs and billing systems.
To increase billing efficiency and accuracy, the study emphasized the necessity
for seamless data connectivity between clinical and financial systems. An essay on
the application of data analytics in healthcare billing and payments was written by
Britton [63]. They looked at ways to evaluate billing data, spot fraud, and increase
billing accuracy using data analytics approaches. The advantages of data analytics
in enhancing revenue cycle management, detecting billing problems, and enhancing
financial performance in healthcare businesses were underlined by the researchers.
The study shed light on the difficulties and possibilities of using data analytics for
healthcare billing and payments, with particular emphasis on the necessity of strong
data governance and analytical skills in managing financial operations. A literature
study on the changing nature of healthcare billing and payments was undertaken
Future of Electronic Healthcare Management: Blockchain and Artificial … 191

by Yaqoob et al. [64]. They looked at how technology developments, legislative

changes, and payment reform initiatives affected healthcare payment systems. The
researchers talked about how healthcare is moving toward value-based payment
models, the emergence of alternative payment options, and the significance of pricing
transparency. The analysis highlighted the need for flexibility and creativity to suit
the shifting demands of the healthcare business and offered insightful information
on current trends and potential directions for healthcare payments and billing.

3.9 Consent Management and Data Privacy

Blockchain allows for granular consent management and gives individuals authority
over their health data. Patients have more control over who has access to their sensi-
tive information and can give or cancel access to their data, preserving privacy. An
illustration of a blockchain-based consent management system is the Health Nexus
platform. In a literature study, Kommunuri [65] presented an overview of permission
management and data privacy in healthcare. The significance of informed consent
in preserving patient confidentiality and privacy was investigated by the researchers.
They talked about the difficulties and factors to be considered when handling consent,
as well as consent models, consent documents, and patient involvement. The analysis
highlighted the necessity for enterprises to adhere to legal obligations and safeguard
patient rights by examining the effects of data privacy laws like GDPR and HIPAA
on consent management procedures in healthcare. A study on the application of
blockchain technology to consent management and data privacy was released in
2019 by Asghar et al. [66]. They talked about how to leverage blockchain for data
exchange, auditing, and safe, open consent management. The researchers looked at
how blockchain may improve data privacy procedures and increase patient ownership
over their health data.
In addition to demonstrating the value of decentralized and verifiable permis-
sion systems in preserving patient confidence, the study offered insights into the
difficulties and prospects of integrating blockchain in consent management and
data protection. In research published in 2023, Rantos et al. [67] looked at patient
involvement in consent management and data privacy. They looked at how crucial
it is for consent procedures to involve patients in decision-making and empower
them. The necessity for clear and easily available information was emphasized as
the researchers explored the difficulties and factors to be considered when prop-
erly conveying privacy rules and consent alternatives to patients. The study offered
insights on tactics for increasing openness and patient-centered treatment while also
promoting patient engagement and improving data privacy standards. A paper on the
effects of data privacy laws on consent management in healthcare was written by
Kommunuri [65]. They looked at the demands and effects of laws like the GDPR,
HIPAA, and CCPA on consent procedures. The researchers highlighted the neces-
sity for businesses to set up strong consent frameworks and privacy policies as they
explored the difficulties and factors to be considered when coordinating consent
192 P. Verma et al.

management procedures with legal obligations. By highlighting the significance of

privacy by design and data governance, the study offered insights into best prac-
tices for ensuring compliance and sustaining data privacy in healthcare settings. A
literature study on consent management and data privacy ethics was undertaken by
Maher et al. [68]. Informed permission, data sharing, secondary use of data, and data
ownership were all topics of discussion by the researchers. They talked about the
value of ethical guidelines, procedures for getting informed consent, and openness
in preserving patient autonomy and safeguarding data privacy. The assessment shed
light on the ethical issues and concerns surrounding permission management and
data protection, highlighting the necessity for companies to give patient rights and
moral behavior priority in their data management procedures.

3.10 Public Health Surveillance and Disease Tracking

Blockchain enhances disease monitoring and public health surveillance by safely

gathering and distributing epidemiological data. It enables safe data exchange
between healthcare institutions and public health organizations, real-time moni-
toring, and early disease epidemic identification. Blockchain technology has been
applied to public health surveillance through Solve.Care platform. A literature study
by Martin et al. [69] gave an overview of disease monitoring and public health surveil-
lance. The significance of surveillance systems in observing and limiting the spread
of infectious illnesses was investigated by the researchers. They reviewed several
public health monitoring techniques and tools, including electronic health records,
syndromic surveillance, and data fusion.
The evaluation emphasized the necessity of effective surveillance systems to safe-
guard people’s health by highlighting the importance of monitoring in early detec-
tion, response planning, and public health interventions. A study concentrating on
the application of big data analytics in public health surveillance was released by
Zeng et al. [70]. They presented how to analyze sizable datasets for disease moni-
toring using data mining, machine learning, and predictive modeling approaches.
The advantages of big data analytics in detecting epidemics, identifying risk factors,
and enhancing public health decision-making were investigated by the researchers.
The study highlighted the potential for data-driven techniques to improve public
health monitoring skills while underlining the limitations and opportunities of using
big data analytics in disease tracking.
An investigation of the use of mobile health technology for illness tracking and
surveillance was done by Khoury et al. [71]. They looked at how mobile apps,
wearable technology, and sensors may be used to gather real-time health data for
monitoring. The researchers talked about how mobile health technology may improve
data collecting, track illness patterns, and encourage people to report their own health
conditions. In addition to highlighting the necessity for privacy and security measures
to secure sensitive health data, the study offered insights into the difficulties and
factors to be considered when using mobile health technology for illness tracking. In
Future of Electronic Healthcare Management: Blockchain and Artificial … 193

a 2018 publication, MacKinnon and Brittain [72] discussed the significance of data
security and privacy in public health monitoring.
They looked at the issues and factors involved in safeguarding private health
information gathered for monitoring. The ethical and legal concerns around data
privacy, permission, and sharing in public health surveillance were highlighted by the
researchers. The report included recommendations for the best ways to maintain data
security and privacy in surveillance systems, highlighting the necessity of extensive
privacy regulations, safe data storage, and stringent access restrictions. Iwaya et al.
[73], Aiello et al. [74] and Chiou et al. [75] performed a literature analysis with
an emphasis on the moral issues involved in disease monitoring and public health
surveillance. The researchers looked at the moral issues surrounding data sharing,
privacy, informed permission, and public health measures. To ensure responsible
and ethical surveillance techniques, they talked about the significance of ethical
frameworks, openness, and community involvement. The assessment highlighted
the need to strike a balance between public health goals, individual private rights,
and ethical standards by offering insightful information on the ethical implications
and concerns of public health surveillance.

4 Artificial Intelligence in Healthcare: e-Health Use Cases

AI in healthcare, as shown in e-Health use cases, offers significant potential for

transforming the industry. Figure 2 showcases the high potential of AI use cases in
the healthcare industry.

4.1 Public Health Surveillance and Disease Tracking

e-Health uses artificial intelligence (AI) to help in medical condition diagnosis.

Large patient data sets, including medical history, symptoms, and test results, may
be analyzed by machine learning algorithms to produce precise and rapid diagnoses.
AI-based diagnostic tools can aid medical personnel in improving patient outcomes
and decision-making. Researchers [76] ran two online tests to determine if people
should trust AI to make medical diagnoses. For high-risk disorders, participants
trusted human doctors more than AI doctors. Surprisingly, participants’ expectations
of AI and human doctors were similar, and their level of confidence did not rise
when AI performed better than humans. However, when participants were given the
option of choosing between human and AI doctors, their trust in AI diagnoses consid-
erably improved, highlighting the significance of patient control in fostering trust.
The use of molecular imaging in identifying and treating inflammatory illnesses was
highlighted by Juravle et al. [77]. Diabetes, cardiovascular disease, and immune-
mediated inflammatory disorders are just a few of the ailments that include inflam-
mation as a major contributing role. However, molecular imaging of inflammation
194 P. Verma et al.

Fig. 2 AI high potential of use cases in the healthcare industry

is not frequently performed using existing clinical imaging methods. To improve

illness diagnosis, direct therapeutics, and identify new research possibilities, the
review investigates the potential of known and developing imaging methods to image
inflammatory biomarkers, pathways, and cells. MacRitchie et al. [78] examined how
data science and digital technology have revolutionized diagnostic categorization,
with an emphasis on autism.
The study emphasizes data-driven approaches, including computer vision, feature
selection algorithms, and early indicators of autism. Concerns about privacy and the
use of people in the diagnosis process are also covered. The evaluation also covers
digital therapy treatments for data collection and result monitoring, paving the way
for more accurate and scalable psychiatric diagnostic methods. Washington et al.
[79] updated summary of categorization, etiopathogenesis, diagnosis, and therapy
for insomnia problems. Hyperarousal driven by a number of variables characterizes
insomnia, which is categorized as an “insomnia disorder.” It is divided into chronic,
transient, and other subtypes. Modified diagnostic criteria are needed, as well as
an assessment of related variables and co-existing disorders. Pharmacotherapy and
cognitive-behavioral therapy for insomnia (CBTI) are both available as treatments.
The use of phytochemicals and complementary alternative medicines (CAMs) such
as digital cognitive behavioral therapy and mindfulness-based techniques has shown
potential. Paul et al. [80] examined prenatal diagnostic methods’ ethical considera-
tions, focusing on preimplantation genetic testing (PGT). PGT gives those who are
Future of Electronic Healthcare Management: Blockchain and Artificial … 195

more at risk the chance to examine embryos for genetic abnormalities. PGT raises
questions about embryo safety and increased expenses even while it improves embryo
selection and reproductive results. The paper examines the obstacles, ethical ramifi-
cations, and demand for patient-centered decision-support technologies. It highlights
the value of making informed decisions, considering the concerns of patients, and
addressing ethical issues during counseling sessions.

4.2 Predictive Analytics

Predictive analytics, which uses previous data to estimate future health outcomes, is
made possible by AI in e-health. AI algorithms may identify people who are at risk
of contracting diseases by examining patient data such as demographics, lifestyle
variables, and genetic markers. Early intervention and individualized preventative
care are both made possible by this proactive strategy. To address the restricted use
of radio frequency identification (RFID) technology in the healthcare supply chain,
researchers [81] performed a study. The study sought to forecast RFID adoption by
combining the unified theory of acceptance and use of technology (UTAUT) with
individual variations, such as personality traits and demographic variables. Neural
network analysis was used to gather and examine data from 252 doctors and nurses.
The study suggested 11 criteria, with an emphasis on individual characteristics, to
forecast RFID adoption. The results showed that, in comparison to factors generated
from UTAUT, individual differences were more useful in predicting RFID adoption.
This study stresses the significance of considering individual aspects in technology
adoption and advances our understanding of RFID acceptability in the healthcare
sector. The hard issue of customer churn prediction (CCP) in the telecom business
was tackled by Chong et al. [82]. Six steps make up their suggested methodology:
pre-processing the data, feature analysis, gravitational search feature selection, and
separating the data into train and test sets. On the train set, several prediction models
were applied, including boosting and ensemble methods, logistic regression, naive
Bayes, support vector machines, random forests, and decision trees. For hyperpa-
rameter tuning, K-fold cross-validation was utilized, and the AUC curve and confu-
sion matrix were used to assess the outcomes. Adaboost and XGBoost classifiers
outperformed other models, achieving the greatest accuracy of 81.71 and 80.8%,
respectively, with an AUC score of 84%. This work advances machine learning
methodologies for forecasting customer turnover in the telecom sector. A predic-
tive model for the discontinuation of antihyperglycemic medication in patients with
type 2 diabetes after laparoscopic metabolic surgery was created and validated by
Coussement et al. [83].
In two big US healthcare databases, they employed machine learning techniques
and a shared data model. The model’s ability to help with patient selection and
enhance outcomes was demonstrated by its high accuracy in predicting drug discon-
tinuation. However, such models would need to be implemented in real-world deci-
sion support, which would require a sufficient technology foundation. To create
196 P. Verma et al.

predictive models for clinical decision-making, this study shows the potential of
machine learning with real-world healthcare data. Birth Match is described by John-
ston et al. [84] as an innovative policy approach to reduce newborn maltreatment
by utilizing data systems to forecast future risk. With this method, a child protec-
tion response is started using information from birth certificates and child welfare
records. The research explores the moral implications of Birth Match and emphasizes
the significance of openness and responsibility in its execution. While technology
has the potential to stop newborn maltreatment that is deadly, ethical issues and
trade-offs need to be carefully considered. The study highlights the requirement for
moral frameworks to direct the use of policy innovations and guarantee the avoid-
ance of baby abuse. Lanier et al. [85] investigates the use of artificial intelligence and
machine learning to detect high-risk suicidal patients using predictive analytics. They
stress the necessity for extensive, sensitive patient data to underpin difficult medical
choices. To illustrate its difficulties, the research compares suicide prediction to non-
medical and medical forecasts. A risk–benefit paradigm is used to examine clinical
and ethical issues, with an emphasis on the possible drawbacks of misclassifying
suicide risk. The authors urge a thorough evaluation of the hazards and advantages
in healthcare populations and offer useful guidelines to safeguard patient rights and
improve the therapeutic value of suicide prediction analytics technologies.

4.3 Patient Monitoring

Remote patient monitoring systems use AI technology to collect and analyze patients’
vital signs and health data in real time. Wearable technology using AI algorithms
and sensors can identify unusual trends, notify healthcare professionals, and enable
quick treatments. AI-powered remote patient monitoring increases accessibility to
healthcare, promotes disease management, and lowers hospital readmissions. For
instance, Biofourmis, a firm that develops digital therapies, employs AI to monitor
patients with heart failure. Their wearable gadget gathers physiological data, which
AI systems then analyze to forecast and identify heart failure exacerbations, allowing
for prompt therapies. Artificial intelligence (AI) is discussed in papers by researchers
[86, 87] that include telemedicine and remote patient monitoring. The researchers
look at how AI algorithms may be used to interpret data from remote patient moni-
toring, speed up diagnostic choices, and enhance individualized patient care. They
look at how AI may improve telemedicine practices’ effectiveness, accuracy, and
patient outcomes. The study sheds light on the difficulties and possible possibilities
of using artificial intelligence (AI) in telemedicine and remote patient monitoring,
emphasizing how AI could change healthcare delivery and enhance patient care.
A thorough analysis of remote patient monitoring (RPM) systems in healthcare is
provided by Jeddi and Bohr [87] with an emphasis on the use of artificial intelligence
(AI). RPM is a useful tool for keeping tabs on patients in a variety of contexts, and AI
has the ability to improve it. The paper examines the effects of AI on RPM, taking
into account cutting-edge technological applications, difficulties, and new trends.
Future of Electronic Healthcare Management: Blockchain and Artificial … 197

The assessment has a focus on patient-centric RPM systems that make use of wear-
able gear, sensors, and cutting-edge technologies like blockchain, fog, and edge.
In several facets of RPM, including activity categorization, chronic illness moni-
toring, and vital sign tracking, AI plays a crucial role. The results underscore the
revolutionary possibilities of AI-enabled RPM, including tailored monitoring, early
diagnosis of health decline, and learning human behavior patterns. The paper also
covers the difficulties and implementation problems that come with incorporating
AI into RPM systems and offers predictions about the future uses of AI in RPM
applications based on new developments and issues. Shaik et al. [88] emphasizes
the Internet of Things’ (IoT) considerable influence on healthcare, notably in the
tracking of important health indicators. IoT devices have completely changed how
health monitoring is done by measuring factors like blood pressure, body temper-
ature, pulse rate, and blood oxygen saturation. Additionally, these gadgets pick up
on physical movements to spot dangers like falls and injuries. For such gadgets,
mobility, minimal weight, and user-friendliness are essential design elements. An
inbuilt CPU is used in the system’s architecture to pre-process sensor signals and
gather data. A shared cloud platform is used to execute feature extraction, recognition
algorithms, and the data presentation of vital indicators, along with emergency call-
outs as needed. Elango et al. [89] examines how machine learning (ML) and artificial
intelligence (AI) are transforming social media in the healthcare industry. Telehealth,
remote patient monitoring, and general well-being may all benefit from the appro-
priate management of the massive amounts of data created on social media platforms
thanks to AI and ML algorithms. The paper identifies key trends in the adoption of
AI-ML, such as the use of sentiment analysis for improved social media marketing,
the use of social media as a tool for data collection with privacy protections, and
the use of chatbots and personalized content to build long-term relationships with
stakeholders.
In the context of telehealth and remote patient monitoring, the study identi-
fies research gaps, offers a conceptual framework to maximize AI-ML application,
handle ethical issues, and counteract false information on social media platforms.
The research focused on deploying wearable Internet of Things devices to monitor
COVID-19 patients was carried out by Leung [90]. The technology monitors vital
signs and uses real-time GPS data to notify medical authorities of potential confine-
ment breaches. A layer of wearable IoT sensors, a layer of mobile Android applica-
tions for alerts, and a cloud layer for data processing make up the proposed system’s
three tiers. The paper also presents a CNN-UUGRU deep neural network model for
identifying human activities that outperforms other models on the Kaggle dataset
with accuracy, precision, and F-measures of 97.75, 96.8, and 97.8%, respectively.
198 P. Verma et al.

4.4 Drug Discovery and Development

AI is essential for expediting the processes of drug discovery and development. Large
datasets of genomic and proteomic data may be analyzed by machine learning algo-
rithms to pinpoint new therapeutic targets and forecast the potency of medication
candidates. AI-powered platforms can cut expenses, accelerate the creation of novel
drugs, and streamline the drug discovery process. Researchers reviewed the most
recent approaches to medication development for uncommon disorders [91]. There
is a need to close the gap between fundamental research and therapeutic therapies
since there are millions of people in the United States who suffer from one of the
7,000 rare illnesses that are known to exist. The review focuses on whole genome
sequencing and pharmacogenetics for determining causes and creating therapies for
genetic uncommon disorders. High throughput screening, medication repurposing,
and the utilization of biologics like gene therapy and recombinant proteins are all
covered in this article. Explored are many disease models, such as induced pluripo-
tent stem cells and animal models. The importance of biomarkers in the development
and discovery of new drugs is also underlined. Sarkar et al. [7] emphasizes how Arti-
ficial Intelligence (AI) has the potential to revolutionize drug research. In computer-
facilitated drug development, the application of machine learning, in particular deep
learning (DL), in conjunction with massive data and improved processing capacity,
has shown promising outcomes. Artificial neural networks, for example, enable the
automated extraction of characteristics from input data and the establishment of
nonlinear connections.
The early pessimism regarding the use of AI in pharmaceutical discovery is
dissipating since it is anticipated that AI would speed up the search for new and
better medications when combined with contemporary experimental approaches.
The growth of AI-driven drug discovery will be greatly aided by technological break-
throughs and open data sharing. In this paper, the potential uses of AI to speed up
the drug discovery process are examined. Various methods used to increase pharma-
ceutical research and development success rates are discussed in Sun et al. [92]. The
review entails a discussion of genomes and proteomics, target-based and phenotypic
screening, drug repurposing, collaborative research, underdeveloped therapeutic
domains, outsourcing, and artificial intelligence-assisted pharmaceutical modeling.
These methods have improved target identification and validation, rational drug
design, cost, and time savings, higher returns on investment, cross-fertilization of
ideas, resource sharing, niche drug discovery, and effective computer-aided drug
design, all of which have resulted in the discovery of successful drugs. The applica-
tion of these tactics, either singly or in combination, has the potential to stimulate
pharmaceutical research and development. Kiriiri et al. [93] draws attention to the
substantial expansion of biological data and the expanding usage of machine learning
(ML) and artificial intelligence (AI) techniques in data mining for drug development.
Deep learning (DL) is one of the AI techniques that has shown promise in a variety of
tasks, including the production of chemical structures, scoring of binding affinities,
position prediction, and molecular dynamics.
Future of Electronic Healthcare Management: Blockchain and Artificial … 199

The paper evaluates the present state-of-the-art in AI-assisted drug development,

examines obstacles to be overcome, and tackles problems. It also looks at possible
directions for this quickly developing profession in the future. The difficulties in
drug design and development, such as low effectiveness, off-target delivery, time
requirements, and high costs, are highlighted by Cerchia and Lavecchia [94]. The
discipline has undergone a revolution thanks to the introduction of artificial intel-
ligence (AI) and machine learning (ML) technologies, which have made strides
in peptide synthesis, virtual screening, toxicity prediction, medication monitoring,
and other fields. With the use of data mining and management strategies, AI and
deep learning algorithms have shown to be successful in a few drug development
procedures. These developments have the potential to have a significant influence on
human health and present exciting options for rational medication design. The intri-
cate interaction of genomic, proteomic, and metabolic networks in regulating human
biological processes is discussed in Patel and Shah [95]. The discovery of biological
processes that are being reprogrammed and the creation of network-based models
to comprehend these intricate systems are both made possible by advancements in
the omic sciences. A comprehensive interpretation of the system is made possible by
the integration of molecular networks from many angles, which offers promise for
precision medicine. Pharmaco-omics studies, which combine pharmacology with
omics data, can produce treatment plans that are specifically suited to a patient’s
needs. The review emphasizes the fundamental concepts of network-based systems
biology, the incorporation of multi-omics data, and the value of network-based drug
discovery techniques for the treatment of breast cancer.
AlzGPS, a complete systems biology tool created to assist in the process of
drug discovery and development for Alzheimer’s disease (AD), is introduced by
Turanli et al. [96]. AlzGPS combines heterogeneous biological networks, multi-
omics data, clinical databases, and more to make it easier to identify targets and
create AD preventive and treatment plans that work. The platform features clinical
trial information, FDA-approved/investigational medication information, literature
references for therapeutic candidates, curated AD multi-omics datasets, endophe-
notype disease modules, and a user-friendly online interface for browsing and
displaying data. AlzGPS offers useful methods for advancing AD treatment research
by utilizing systems pharmacology and network-based analyses. An overview of
machine learning techniques for incorporating multi-omics data in cancer molecular
biology investigations may be found in Zhou et al. [97]. The review divides the tools
into approaches that are general-purpose and task-specific, including supervised and
unsupervised learning methods. Utilizing information from the Cancer Cell Line
Encyclopedia, benchmarking is done to assess runtime effectiveness, mean absolute
error in drug response prediction, and accuracy of cancer type categorization. The
results emphasize the need for continued study and development of the fresh method-
ology to improve drug discovery, clinical trial design, and individualized therapies.
They also provide direction to researchers in choosing effective machine-learning
methods for their applications.
200 P. Verma et al.

4.5 Virtual Assistants and Chatbots

AI-based chatbots and virtual assistants are used in e-health to schedule appoint-
ments, respond to questions, and deliver individualized health information. These
conversational interfaces utilize machine learning and natural language processing to
comprehend user questions and provide relevant answers. Virtual assistants improve
accessibility to healthcare services and patient involvement For instance, Buoy Health
has created an AI-powered chatbot that assists users in evaluating their symptoms and
offers tailored health recommendations. It leads users through a series of questions
to identify probable reasons and suggests the best course of action, such as engaging
in self-care or seeing a doctor. To pinpoint key trends and knowledge gaps in the
area of chatbots and stakeholder interactions, researchers [98] undertook a thorough
literature study.
The review analyzed 62 peer-reviewed English articles using a code book and
inductive analysis. Findings revealed that existing studies primarily focused on the
technical aspects of chatbots and their language skills, with limited consideration for
organizational and societal perspectives. The study emphasized the need for corporate
communication scholars to contribute more to the discussion of chatbot-stakeholder
interactions. The results provide valuable insights into the organizational capabilities
and affordances of chatbots, highlighting their importance in stakeholder engagement
strategies. Syvänen and Valentini [99] analyzes the advantages, limitations, ethical
considerations, and future prospects of ChatGPT and artificial intelligence (AI) in
healthcare. ChatGPT, a powerful language model, generates human-like responses
using deep learning techniques. 62 Peer-reviewed English publications were exam-
ined using inductive analysis and a code book. Findings showed that previous research
mostly concentrated on the technical features of chatbots and their linguistic capa-
bilities, with little attention paid to organizational and societal viewpoints. The study
highlighted the need for corporate communication experts to add more to the conver-
sation on chatbot-stakeholder interactions. The findings underscore the significance
of chatbots in stakeholder engagement strategies by offering insightful information
on the organizational capabilities and affordances of these tools. Syvänen and Valen-
tini [99] examines the benefits, drawbacks, moral issues, and potential applications of
ChatGPT and AI in the medical field. Deep learning methods are used by ChatGPT,
a potent language model, to produce replies that are human-like.
It has several uses in medicine, including supporting patient care, clinical diag-
nosis, medical education, and research. Copyright infringement, medico-legal prob-
lems, and transparency in AI-generated information are examples of ethical chal-
lenges, nevertheless. The study analyses these issues, highlighting ChatGPT’s
promise and difficulties in the healthcare industry. The Leora model is one example of
an AI-powered platform that [100] discusses as having the potential to help mental
health. These systems, which include conversational agents like Leora, can offer
tailored and easily available mental health help to people who are only mildly to
Future of Electronic Healthcare Management: Blockchain and Artificial … 201

moderately affected by anxiety and sadness. However, for the appropriate develop-
ment and use of AI in mental health treatment, ethical concerns relating to trust, trans-
parency, bias, and potential negative repercussions must be addressed. To guarantee
the efficacy of such models, intensive user testing is required for validation. A study
was undertaken to determine how eager people are to interact with AI-driven health
chatbots [101]. Participants’ responses to an online survey and semi-structured inter-
views were gathered for data. Three themes—"Understanding of chatbots,” “AI hesi-
tancy,” and “Motivations for health chatbots”—came out of the interviews, stressing
worries about accuracy, cyber-security, and the perceived lack of empathy in AI-led
services.
The poll found a modest level of acceptance for health chatbots, with character-
istics impacting adoption including IT proficiency, attitude, and perceived benefit.
Intervention designers should address patients’ concerns and improve user experi-
ence by utilizing a user-centered and theory-based approach to guarantee effective
uptake and use. Nadarzynski et al. [102] addresses the demand for proactive asthma
treatment as well as the difficulties in maintaining ongoing asthma monitoring and
control in conventional clinical settings. They unveil kBot, a customized chatbot
system made to help children with asthma. Through an Android app, kBot contin-
ually collects pertinent health and environmental data while tracking medication
adherence.
It makes use of contextualization through obtaining patient feedback and domain
expertise, as well as customization through surveys and regular interactions. KBot’s
excellent acceptability and utility during preliminary testing with physicians and
researchers suggested that it has the potential to be an important tool in the manage-
ment of asthma. The goal of [103] is to create a virtual caregiver system for individ-
ualized healthcare for the elderly. The system uses a mobile chatbot to communi-
cate with the user and gather details about their physical and emotional well-being.
The technology incorporates a rule-based virtual caregiver system dubbed “Mind
Monitoring” with physical, mental, and social questions within the chat application,
in contrast to conventional health monitoring techniques. The elderly individual
answers a question from the chatbot every day by pressing buttons or speaking via a
microphone. The technology quantifies the responses, creates illustrative graphs, and
offers summaries and recommendations that are personalized for each user. Positive
outcomes from an experimental examination with eight senior participants and 19
younger participants over a 14-month period included a response rate above 80%
and useful feedback messages. To better examine and enhance health outcomes,
interviews were also performed.

4.6 Medical Imaging Analysis

For the examination of medical imaging, artificial intelligence is widely employed

in the healthcare industry. Medical pictures like X-rays, CT scans, and MRIs may
be analyzed by AI algorithms to help in the identification and diagnosis of a variety
202 P. Verma et al.

of illnesses. AI-powered image analysis tools can accurately identify abnormali-

ties, cancers, and other anomalies, assisting radiologists with their interpretation
and enhancing diagnosis efficiency. For instance, Google’s DeepMind created an
AI system that can examine retinal scans to find signs of eye conditions like
diabetic retinopathy. The device helps ophthalmologists diagnose and treat patients
in real-world clinics. Deep learning approaches are being looked at by researchers
[104] for the analysis of medical pictures and the localization of areas of interest
(ROI). Due to these photos’ complicated look, conventional techniques are limited.
The study focuses on using a powerful deep learning architecture, especially the
FCN-UNET approach, to segment MRI images of skin lesions, retinal lesions, and
brain lesions. The suggested approach exhibits effective ROI recognition, making it
possible to diagnose conditions including skin cancer, eye abnormalities, and brain
tumors. Evaluation using databases that are open to the public reveals good accu-
racy and a dice coefficient that exceeds 90%. Adegun et al. [105] suggests a deep
learning-based multi-mode technique for fusing medical images. This technique
enables batch processing of diverse multi-modal medical pictures while addressing
the drawbacks of MRI, CT, and SPECT image fusion. It enhances time effectiveness,
image detail clarity, and fusion effect. Because of quantitative assessment criteria,
the suggested method outperforms existing techniques when applied to multimodal
medical pictures. This study advances medical picture fusion using deep learning and
makes contributions to pattern identification, image processing, and practical imple-
mentation in medical diagnostics. A deep learning-based radiomics model named
DRTOP is presented by Li et al. [106] for forecasting time-to-event clinical outcomes
in patients with lung cancer. DRTOP use raw pictures as opposed to manually created
radiomics and determines image-based chances of mortality or recurrence.
The model’s predictive ability for outcomes such as overall survival (OS), distant
control (DC), and local control (LC) outcomes using computed tomography (CT)-
based characteristics was assessed on a dataset of 132 patients. Features from
positron emission tomography (PET) were used to predict both overall survival (OS)
and recurrence-free survival (RFS). In terms of predicting clinical outcomes, the
deep learning-based radiomics model performed better than hand-crafted radiomics,
potentially providing advantages in the management of individual cases of lung
cancer. The significance of explainable machine learning (XML) techniques in
medical image analysis is covered in Afshar et al. [107]. Even though machine
learning models can predict clinical parameters with accuracy, their lack of inter-
pretability raises questions. ML algorithms may be explained and important predic-
tors in a prediction can be identified using XML techniques like Shapley values. The
integration of XML techniques into medical software solutions might increase confi-
dence in ML predictions and help doctors make wise judgments. The work presents
a unique XML imaging pipeline for explaining result predictions from sophisticated
medical imaging models using radiomics data and Shapley values. The outcomes are
shown on a dashboard designed specifically for clinicians and used to demonstrate
how to forecast genetic mutations in glioma patients using MRI data. An overview
of the development and uses of deep learning in medical image processing is given
in Severn et al. [108]. The paper focuses on the neurological system, cardiovascular
Future of Electronic Healthcare Management: Blockchain and Artificial … 203

system, digestive system, and skeletal system as particular examples of therapeutic

applications for convolutional neural network-based methodologies. Although deep
learning models have demonstrated remarkable performance, the research notes that
the practical usefulness of these models is constrained by the limited size of the
medical datasets. In order to advance the area, the future direction advises investi-
gating federated learning, gathering benchmark datasets, and adding domain-specific
topic knowledge. Overall, cutting-edge deep learning methods have shown signif-
icant promise in the interpretation of medical images, but further development is
required to handle data issues and improve scalability.

4.7 Patient Engagement and Behavioral Modification

Artificial intelligence (AI) technologies are used to increase patient engagement and
encourage behavioral change for better health outcomes. Applications and platforms
driven by AI may give individualized health recommendations, perform focused
interventions, and provide feedback and encouragement to people so they follow
their treatment programs, adopt better lives, and effectively manage chronic illnesses.
The Patient Health Engagement Scale (PHE scale) was produced through thorough
conceptualization and psychometric techniques, according to research [109]. The
PHE model was used to develop the scale, which was then tested on a large national
sample of chronic patients and showed to have strong psychometric characteristics
and reliability. A deeper understanding of the role that patient participation plays in
healthcare quality, outcomes, and cost containment is made possible by this valid and
reliable metric. The PHE scale shows potential in terms of helping to adapt interven-
tions and evaluate improvements following patient engagement efforts. An expert
agreement on digital behavior change interventions is presented by Graffigna et al.
[110], with an emphasis on conceptualizing and evaluating engagement. The report
places more emphasis on “effective engagement” than it does on merely increasing
involvement to achieve desired results. It emphasizes the necessity for reliable and
effective metrics to create multifaceted models of interaction. In addition, the article
advocates for an iterative, user-centered approach to intervention design, combining
mixed techniques and qualitative research to fine-tune treatments in accordance with
user needs and context.
Yardley et al. [111] emphasizes the significance of taking patient involvement
into account while designing rehabilitation trials to enhance the outcomes during
the implementation phase. The study suggests that by including patient participation
as a key factor in trials, we move from a therapist-focused to a patient-focused
strategy. Exercise prescription is used as an example to demonstrate how the Behavior
Change Wheel may be used to examine obstacles and facilitators relating to patients’
skills, opportunities, and motivations. In order to guarantee realistic, significant, and
transferable findings that enable actual effects on rehabilitation treatments at the
primary care level, the study suggests a framework based on the COM-B model.
The adoption and use of artificial intelligence (AI) technology in healthcare settings
204 P. Verma et al.

are highlighted as being critically important by Zheng et al. [112]. Twelve Canadian
patients participated in semi-structured interviews as part of the study to learn more
about their opinions on what skills healthcare professionals should possess for the
future of AI-enabled healthcare and how to better involve patients in the usage of AI
technology. In addition to establishing data governance and validating AI technology,
the panelists emphasized the necessity to create trust and patient involvement. To
address health disparities and improve the quality of care, the study focused on the
role of healthcare providers in including patients.
To capture various patient viewpoints, more study is required. The necessity of
comprehending and enhancing adolescent involvement with behavior modification
programs is emphasized by Jeyakumar et al. [113]. To evaluate and improve treat-
ments aimed at teenagers, the article advises combining artificial intelligence (AI)
and process-level data. The approach that has been developed focuses on four main
objectives: assessing engagement, modeling engagement, improving existing inter-
ventions, and developing new interventions. The framework is built on the illustration
of the INSPIRE narrative-centered intervention for hazardous alcohol consumption.
However, when using AI in treatments for children, ethical issues, such as privacy
concerns, must be considered. The report ends by outlining the many possibilities
for additional research in this quickly changing sector.

5 Blockchain-Based Various Tools Used in Healthcare

Blockchain technology was being explored and implemented in healthcare for various
purposes. Here is a list of some existing blockchain-based tools used in healthcare.

5.1 MedicalChain

It is a blockchain-based platform created to overcome the difficulties associated with

the safekeeping and exchange of patient data and medical records in the healthcare
sector. It makes use of blockchain technology to offer a decentralized and unchange-
able mechanism for handling private medical data. Patients may safely exchange their
health information with approved healthcare professionals, researchers, and organi-
zations using MedicalChain, giving them ownership over their personal health infor-
mation. The technology uses the immutability of blockchain to record every transac-
tion and access to medical records, ensuring data integrity, privacy, and transparency.
Advanced encryption and smart contracts are used to preserve patient privacy and
permission for data access. It makes it easier for healthcare practitioners to interact
seamlessly, cutting down on redundant testing and enhancing care coordination.
MedicalChain also makes it possible for patients to consult with medical experts
securely, remotely, and in real-time through telemedicine services. This function is
especially helpful in enhancing accessibility to medical knowledge and lessening
Future of Electronic Healthcare Management: Blockchain and Artificial … 205

the strain of in-person appointments. MedicalChain benefits patients and healthcare

providers alike by strengthening data security and expediting medical data sharing,
enabling a more effective and patient-centric healthcare ecosystem.

5.2 SimplyVital Health

A blockchain-based platform called SimplyVital Health aims to revolutionize health-

care by improving care coordination between healthcare professionals and securely
exchanging patient data. The platform uses blockchain technology to build a tamper-
and immutable-proof ledger of patient records, assuring the security and integrity
of the data. Healthcare practitioners may safely communicate patient data, such as
medical records, test results, and treatment plans, with other authorized parties with
SimplyVital Health. The smooth coordination of treatment is promoted by this safe
and open data interchange, which also improves patient satisfaction. Incorporating
smart contracts into the platform also enables healthcare providers to join into value-
based care agreements and simplify payment procedures. Smart contracts ensure fast
and correct payment for services by automatically executing when certain criteria
are satisfied. SimplyVital Health lessens the administrative complexity associated
with data exchange and improves data interoperability among healthcare systems by
utilizing blockchain’s decentralized nature. This results in better patient outcomes,
lower healthcare expenses, and greater ecosystem efficiency. SimplyVital Health’s
overall goal is to improve the healthcare sector by promoting teamwork, data security,
and patient-centered treatment.

5.3 Patientory

A blockchain-based platform called Patientory focuses on giving patients owner-

ship over their health data while assuring a secure and effective transmission of
that information. Patients may save their test results, medical records, and other
health-related data on the platform in a decentralized, unhackable way. Patients may
use it to provide healthcare practitioners, researchers, and other authorized parties
access to their health data, facilitating safe and secure data exchange within the
healthcare ecosystem. By utilizing blockchain technology, the platform provides
data security, privacy, and integrity, lowering the possibility of data breaches and
illegal access. Smart contracts, which allow for automatic and open communication
between patients and healthcare professionals, are also incorporated into Patien-
tory. These smart contracts simplify payment procedures and support value-based
care agreements, improving care coordination and lowering administrative burdens.
Patientory enables people to make educated decisions about their health and take
an active role in their care by placing patients at the heart of their health data
206 P. Verma et al.

management. The platform’s blockchain-based methodology promotes data inter-

operability, cooperation, and trust among all stakeholders, improving healthcare
outcomes and making the healthcare system more patient-centric.

5.4 Nebula Genomics

Nebula Genomics is a revolutionary platform that leverages blockchain technology

to empower individuals with control over their genomic data. Users may choose
how they wish to share their genetic data with academics, healthcare professionals,
or other interested parties via the platform, which securely stores their data. By
offering a decentralized and unchangeable storage solution using blockchain, Nebula
Genomics protects the confidentiality and security of its consumers’ genetic data.
This promotes openness and trust by allowing users to retain ownership of their data
while providing authorization for access to parties. The opportunity for consumers to
profit from their genetic data is one of Nebula Genomics’ distinctive features. Users
can receive prizes in the form of cryptocurrency for taking part in research projects or
clinical trials, encouraging data sharing while retaining their privacy. The platform
also makes it simple for people and research organizations to share data, which
speeds up genetic analysis and advances in precision medicine. The blockchain-
powered ecosystem of Nebula Genomics intends to transform the genomics sector
by fostering data privacy, ownership, and cooperation for a more individualized and
efficient approach to healthcare.

5.5 Healthereum

Healthereum is a pioneering blockchain-based platform designed to enhance

patient engagement and incentivize positive healthcare actions. The platform uses
blockchain technology to provide a safe and open environment where stakeholders
such as patients, healthcare professionals, and others may communicate easily.
Patients are encouraged to actively participate in their healthcare journey through
Healthereum. For fulfilling specific tasks, such as making appointments, following
treatment programs, or taking part in health-related surveys, they are rewarded with
Halthereum tokens. On the platform, these tokens may be used for a variety of things,
like getting access to upscale medical treatment or getting deals on hospital bills.
The blockchain-based technology used by Healthereum guarantees the security and
immutability of patient data and interactions, fostering a culture of trust. The tech-
nology also makes it easier for patients and healthcare professionals to communicate,
which improves patient-provider connections and allows for real-time feedback.
Healthereum aims to encourage individuals to adopt a more pro-active attitude to
Future of Electronic Healthcare Management: Blockchain and Artificial … 207

their health by gamifying healthcare encounters and offering concrete benefits. This
cutting-edge platform’s ultimate goals are to increase patient involvement, enhance
healthcare results, and promote constructive behavior changes within the healthcare
sector.

5.6 Solve.Care

Solve.Care is a comprehensive blockchain platform that revolutionizes the way

healthcare is administered and coordinated.
It makes use of blockchain technology to raise the level of efficiency, security,
and transparency throughout the healthcare industry. By establishing connections
between patients, healthcare providers, insurers, and other stakeholders through
a decentralized network, Solve.Care’s primary goal is to enhance care coordina-
tion. All stakeholders engaged in a patient’s treatment path may communicate and
collaborate easily because to the platform’s secure patient data sharing capabilities.
By automating activities including appointment scheduling, claims processing, and
insurance verifications, Solve.Care simplifies administrative procedures. Through
automation, administrative tasks are simplified, mistakes are reduced, and processing
times are shortened. SOLVE, the platform’s native utility token, is employed to
streamline internal rewards and payments. Incentives for following treatment plans
may be given to patients, and the token can be used by providers and insurers to
speed up payment procedures. Solve.Care establishes data integrity, privacy, and
auditability using blockchain technology, fostering confidence in the healthcare
system. The platform’s cutting-edge method of managing and coordinating care has
the ability to greatly raise the standard of healthcare delivery while lowering costs
and inefficiencies.

5.7 MedRec

MedRec is a blockchain-based system developed by researchers at the Massachusetts

Institute of Technology (MIT) to address the challenges in managing electronic
health records (EHRs). It makes use of blockchain technology to ensure that patient
data is stored securely and impervious to tampering, giving patients and health-
care professionals transparent and auditable access to medical information. The
platform improves privacy and security by giving patients more control over their
health data and the decision of who may access it. By streamlining data exchange
and enhancing interoperability between healthcare organizations, MedRec hopes to
produce healthcare services that are more effective and patient-focused (Table 1).
208 P. Verma et al.

Table 1 Comparative table of blockchain tools support the healthcare system

Platform Purpose and features Medical Data sharing Genomic Patient Care
records and data engagement coordination
storage coordination sharing
MedicalChain Blockchain platform ✓ ✓ ✗ ✗ ✗
for secure storage
and sharing of
medical records and
patient data
SimplyVital Blockchain-based ✓ ✓ ✗ ✗ ✓
Health platform for
healthcare providers
to securely share
patient data and
coordinate care
Patientory Blockchain-based ✓ ✓ ✗ ✗ ✗
platform for patient
data management
and health
information
exchange
Nebula Blockchain-powered ✗ ✗ ✓ ✗ ✗
Genomics genomic data
sharing platform that
allows individuals to
control and monetize
their genetic data
Healthereum Blockchain-based ✗ ✗ ✗ ✓ ✗
platform for patient
engagement and
incentivizing
healthcare actions
Solve.Care Blockchain platform ✓ ✓ ✗ ✗ ✓
for care
coordination,
administration, and
payments in
healthcare
MedRec Blockchain system ✓ ✓ ✗ ✗ ✗
developed by MIT
for managing
electronic health
records and
providing secure
data access
Future of Electronic Healthcare Management: Blockchain and Artificial … 209

6 Integration of AI and Blockchain in Healthcare

To overcome the issues with both technologies, it has been suggested that blockchain
and AI integration in healthcare be used [114]. Data is a major component of how AI
systems learn, understand, and make decisions. Machine learning algorithms work
best when the data they use comes from a reliable, trustworthy, and secure platform.
The blockchain acts as a distributed ledger, guaranteeing that data is kept and shared
by all participating nodes in an unhackable and cryptographically secure way. The
outcomes may be trusted and validated with high integrity and resilience by using
smart contracts for decision-making and analytics using machine learning algorithms.
With improved security and dependability, this connection has the potential to trans-
form healthcare data management and analytics. For handling sensitive data gath-
ered, saved, and used by AI-driven techniques, the integration of blockchain with AI
offers a safe, unchangeable, and decentralized solution [115]. The medical, personal,
banking, financial, trade, and legal sectors, among others, all benefit greatly from
these developments in data and information security [116]. According to Mamoshina
et al. [117], this integration enables intelligent decentralized autonomous agents, or
DAOs, to evaluate data, value, and asset transfers autonomously and quickly among
several authorities. As a result, the combination of AI with blockchain technology has
the promise of revolutionizing data management and transactional procedures while
providing unmatched efficiency and security. To analyze incoming data, spot trends,
and enable more efficient operations and insights into patients’ habits and health,
AI is essential. Analytics may be strategically applied by AI to improve decision-
making. AI systems can identify life-threatening illnesses and prescribe treatments
and length of hospital stays for patients by studying algorithms and making diffi-
cult judgments. With the help of AI and Google’s enormous computing capacity,
massive datasets can be processed with astounding precision. Hanover, a Microsoft
AI product, is also used to mine healthcare data and unlock the promise of AI for
improvements in the healthcare industry. The use of AI in healthcare has a lot of
potential to enhance patient care and healthcare outcomes in general. With the use
of cutting-edge technology, machines are now able to read, understand, and retain
medical research articles, providing prospective treatments specific to each patient
[118]. Pharmaceutical firms use blockchain technology to protect the integrity of their
supply chains, collect data in real time, and share it effectively to improve patient
outcomes [119]. Blockchain aids in the fight against medicine fraud, which claims
millions of lives annually, by issuing distinctive serial numbers to pharmaceuticals.
By prohibiting unlawful swaps and assuring prompt delivery of high-quality drugs,
our serialization technology guarantees precise drug tracking. For pharmaceutical
firms, the full traceability of pharmaceuticals utilizing blockchain technology has
proven advantageous since it improves drug management and safety [120].
210 P. Verma et al.

7 Advancing Healthcare with AI and Blockchain

Integration: Overcoming Challenges

Blockchain and AI work well together to improve safety, integrity, and treatment
alternatives that are specifically suited to each patient. Healthcare will improve and
accuracy will increase because of AI’s superior capacity to detect abnormalities in
medical imaging. With the goal of using patient data, improving diagnoses, and
providing suggestions based on evidence, major tech giants like Google, Microsoft,
Apple, and Amazon, as well as startups, are aggressively investigating AI’s uses in
healthcare. Robotic surgery and digital counseling via smartphone applications are
innovations that go hand in hand with one another to bring about considerable cost
reductions in the healthcare industry. Security and interoperability issues must be
resolved for healthcare AI to completely realize its potential. The fusion of complex
algorithms, enormous databases, and effective machines offers a cutting-edge and
revolutionary method for integrating healthcare [121].

8 Enhancing Healthcare Data Storage: Integrating

Blockchain and AI

The use of blockchain technology and artificial intelligence in the storage of health-
care data presents intriguing answers to the problems posed by the administration
of massive amounts of medical records and data. The blockchain can be used to
store medical data uploaded via Electronic Health Record (EHR) systems, however,
processing such data directly within the blockchain network has computational cost
and storage constraints owing to its tiny block sizes. Additionally, privacy issues come
up and put data breaches at danger. A strong architecture is used to solve these chal-
lenges, relying on trustworthy third parties to manage enormous volumes of sensitive
data while blockchain is used for secure on-chain storage. In managing healthcare
information, this strategy achieves a balance between effectiveness, security, and
privacy [122, 123].

9 Decentralizing Biomedical Research: Blockchain and AI

Integration

Blockchain and AI technology integration is changing how biological research and

healthcare are conducted. Ahuja [124] showed the effectiveness of deep learning,
a crucial AI method, in a variety of fields, including medication categorization and
medical diagnosis. Its potential as a fruitful tool for drug development and research
was proved by their study. In this situation, the MolAICal software is essential for
producing 3D structural ligands inside protein targets using both conventional and
Future of Electronic Healthcare Management: Blockchain and Artificial … 211

deep learning methods. In the first of its two modules, MolAICal uses fragments of
FDA-approved pharmaceuticals to create a deep learning model based on WGANs.
The protein pocket’s 3D ligands are then made using these produced fragments. In the
second module of the software, WGAN-based deep learning models are trained using
drug-like compounds from the zinc database, and then molecular docking is used
to calculate the affinities between produced compounds and proteins. Testing on the
membrane target GCGR and the non-membrane target SARSCoV-2 Mpro showed
the effectiveness of MolAICal’s drug design functionalities. Notably, the method was
effective in producing ligands that differed in their degree of 3D structural similarity
to the crystal ligands of GCGR and SARS-CoV-2 Mpro, providing helpful resources
for researchers looking for novel drug candidates.
Additionally, the integration of blockchain and AI in customized cardiovascular
therapy was investigated by Bai et al. [125]. This integration offers the potential
for better results by expanding the amount of data available for AI training, sharing
confidential AI techniques, and decentralizing databases. However, other concerns
about implementation, such as technical trust, data transfer among competing organi-
zations, compensation, and ethical implications, need more research. Disseminating
information for lung cancer patients across a decentralized network is another use
for blockchain and AI. Organizations may jointly create global models to find novel
patterns and symptoms by employing locally taught models for big data analytics.
With this novel strategy, patient care might be improved while maintaining privacy
concerns.

10 Conclusion

The use of blockchain and AI technology in the healthcare sector has enormous
potential to transform patient care and medical procedures. Healthcare workers now
have quick access to detailed patient information thanks to the usage of blockchain
to securely store and display medical records, enabling more informed and effective
decision-making. On the other side, AI’s sophisticated algorithms and data analysis
skills are essential for processing huge amounts of patient data and gleaning insightful
information. The medical system may gain from improved service efficiency, simpler
processes, and cost savings by integrating the benefits of various technologies.
Blockchain technology’s decentralized structure protects data confidentiality and
integrity, while AI’s capacity to decode and comprehend complicated medical data
supports individualized healthcare, diagnosis, and treatment planning. Additionally,
the combination of blockchain technology and artificial intelligence democratizes
the healthcare industry by giving individuals more ownership over their health data
and enabling frictionless data exchange between healthcare providers. As a result,
patient outcomes and overall healthcare quality are enhanced via collaboration and
interoperability. But in order to successfully use blockchain and AI in healthcare, a
number of issues must be resolved, including interoperability, data protection, and
ethical considerations. To overcome these obstacles and realize the full potential of
212 P. Verma et al.

this integration, stakeholders such as healthcare practitioners, academics, politicians,

and technology experts must work together. A more patient-centric, effective, and
accessible healthcare ecosystem will be made possible by embracing the transfor-
mational power of blockchain and AI as the healthcare sector continues to develop.
To continuously improve the capabilities of these technologies and take use of their
potential to improve healthcare in the future, further research and development in
this area are essential. Overall, the combination of blockchain and AI is a significant
step toward a more sophisticated, secure, and egalitarian healthcare environment.

Acknowledgements Authors would like to extend their appreciation to every member of the
Datafoundry Pvt. Ltd. team who contributed to this endeavor. The dedication and commitment of
Datafoundry Pvt. Ltd. have played a pivotal role in enhancing the depth and breadth of our research.
The seamless access to data, technical guidance, and collaborative discussions have enriched our
understanding and enabled us to present a comprehensive and well-informed study.

References

1. Alotaibi YK, Federico F (2017) The impact of health information technology on patient safety.
Saudi Med J 38(12):1173
2. Chapuis C et al (2010) Automated drug dispensing system reduces medication errors in an
intensive care setting. Crit Care Med 38(12):2275–2281
3. Campanella P et al (2016) The impact of electronic health records on healthcare quality: a
systematic review and meta-analysis. Eur J Public Health 26(1):60–64
4. Wong ZSY, Zhou J, Zhang Q (2019) Artificial intelligence for infectious disease big data
analytics. Infect Dis Health 24(1):44–48
5. Bragazzi NL, Dai H, Damiani G, Behzadifar M, Martini M, Wu J (2020) How big data and
artificial intelligence can help better manage the COVID-19 pandemic. Int J Environ Res
Public Health 17(9):3176
6. Sahoo MS, Baruah PK (2018) Hbasechaindb—a scalable blockchain framework on Hadoop
ecosystem. In Supercomputing frontiers: 4th Asian conference, SCFA 2018, Singapore, March
26–29, 2018, proceedings 4. Springer, pp 18–29
7. Sarkar C et al (2023) Artificial intelligence and machine learning technology driven modern
drug discovery and development. Int J Mol Sci 24(3):2026
8. Cha Y et al (2018) Drug repurposing from the perspective of pharmaceutical companies. Br
J Pharmacol 175(2):168–180
9. Siyal AA, Junejo AZ, Zawish M, Ahmed K, Khalil A, Soursou G (2019) Applications
of blockchain technology in medicine and healthcare: challenges and future perspectives.
Cryptography 3(1):3
10. Hang L, Choi E, Kim D-H (2019) A novel EMR integrity management based on a medical
blockchain platform in hospital. Electronics (Basel) 8(4):467
11. Akkiraju R et al (2020) Characterizing machine learning processes: a maturity framework.
In: Business process management: 18th international conference, BPM 2020, Seville, Spain,
September 13–18, 2020, proceedings 18. Springer, pp 17–31
12. Feng Q, He D, Zeadally S, Khan MK, Kumar N (2019) A survey on privacy protection in
blockchain system. J Netw Comput Appl 126:45–58
13. Lin C, He D, Huang X, Khan MK, Choo K-KR (2020) DCAP: a secure and efficient decentral-
ized conditional anonymous payment system based on blockchain. IEEE Trans Inf Forensics
Secur 15:2440–2452
Future of Electronic Healthcare Management: Blockchain and Artificial … 213

14. Ahmad SS, Khan S, Kamal MA (2019) What is blockchain technology and its significance
in the current healthcare system? A brief insight. Curr Pharm Des 25(12):1402–1408
15. Zhang P, White J, Schmidt DC, Lenz G, Rosenbloom ST (2018) FHIRChain: applying
blockchain to securely and scalably share clinical data. Comput Struct Biotechnol J
16:267–278
16. Linn LA, Koo MB (2016) Blockchain for health data and its potential use in health it and
health care related research. In: ONC/NIST use of blockchain for healthcare and research
workshop. ONC/NIST, Gaithersburg, Maryland, United States, pp 1–10
17. Bryatov SR, Borodinov AA (2019) Blockchain technology in the pharmaceutical supply
chain: researching a business model based on Hyperledger fabric. In: Proceedings of the
international conference on information technology and nanotechnology (ITNT), Samara,
Russia, pp 1613–1673
18. Singh M, Kim S (2018) Branch based blockchain technology in intelligent vehicle. Comput
Netw 145:219–231
19. Andoni M et al (2019) Blockchain technology in the energy sector: a systematic review of
challenges and opportunities. Renew Sustain Energy Rev 100:143–174
20. Khurshid A (2020) Applying blockchain technology to address the crisis of trust during the
COVID-19 pandemic. JMIR Med Inform 8(9):e20477
21. Shah VN, Garg SK (2015) Managing diabetes in the digital age. Clin Diabetes Endocrinol
1:1–7
22. Pegoraro V et al (2023) Cardiology in a digital age: opportunities and challenges for e-Health:
a literature review. J Clin Med 12(13):4278
23. Alemayehu C, Mitchell G, Nikles J (2018) Barriers for conducting clinical trials in developing
countries—a systematic review. Int J Equity Health 17:1–11
24. de Jongh D et al (2022) Early-phase clinical trials of bio-artificial organ technology: a
systematic review of ethical issues. Transpl Int 35:10751
25. Peipert BJ, Spinosa D, Howell EP, Weber JM, Truong T, Harris BS (2021) Innovations
in infertility: a comprehensive analysis of the ClinicalTrials.gov database. Fertil Steril
116(5):1381–1390
26. Welch MJ et al (2015) The ethics and regulatory landscape of including vulnerable populations
in pragmatic clinical trials. Clin Trials 12(5):503–510
27. Kassab M, DeFranco J, Malas T, Laplante P, Destefanis G, Neto VVG (2019) Exploring
research in blockchain for healthcare and a roadmap for the future. IEEE Trans Emerg Top
Comput 9(4):1835–1852
28. Schmeelk S, Kanabar M, Peterson K, Pathak J (2022) Electronic health records and blockchain
interoperability requirements: a scoping review. JAMIA Open 5(3):ooac068
29. Maitra S, Yanambaka VP, Puthal D, Abdelgawad A, Yelamarthi K (2021) Integration of
Internet of Things and blockchain toward portability and low-energy consumption. Trans
Emerg Telecommun Technol 32(6):e4103
30. Han Y, Zhang Y, Vermund SH (2022) Blockchain technology for electronic health records.
Int J Environ Res Public Health 19(23):15577
31. Agrawal D, Minocha S, Namasudra S, Gandomi AH (2022) A robust drug recall supply chain
management system using hyperledger blockchain ecosystem. Comput Biol Med 140:105100
32. Clauson KA, Breeden EA, Davidson C, Mackey TK (2018) Leveraging blockchain tech-
nology to enhance supply chain management in healthcare: an exploration of challenges and
opportunities in the health supply chain. Blockchain Healthc Today
33. Humayun M, Jhanjhi NZ, Niazi M, Amsaad F, Masood I (2022) Securing drug distribution
systems from tampering using blockchain. Electronics (Basel) 11(8):1195
34. Ghadge A, Bourlakis M, Kamble S, Seuring S (2022) Blockchain implementation in
pharmaceutical supply chains: a review and conceptual framework. Int J Prod Res 1–19
35. Dasaklis TK, Voutsinas TG, Tsoulfas GT, Casino F (2022) A systematic literature review of
blockchain-enabled supply chain traceability implementations. Sustainability 14(4):2439
36. Liu X, Barenji AV, Li Z, Montreuil B, Huang GQ (2021) Blockchain-based smart tracking
and tracing platform for drug supply chain. Comput Ind Eng 161:107669
214 P. Verma et al.

37. Field MJ, Grigsby J (2002) Telemedicine and remote patient monitoring. JAMA 288(4):423–
425
38. Pirtle CJ, Payne K, Drolet BC (2019) Telehealth: legal and ethical considerations for success.
Telehealth Med Today
39. Lloyd J, Lee CJ (2022) Use of telemedicine in care of hematologic malignancy patients:
challenges and opportunities. Curr Hematol Malig Rep 17(1):25–30
40. Niu B, Mukhtarova N, Alagoz O, Hoppe K (2022) Cost-effectiveness of telehealth with remote
patient monitoring for postpartum hypertension. J Matern Fetal Neonatal Med 35(25):7555–
7561
41. De Guzman KR, Snoswell CL, Taylor ML, Gray LC, Caffery LJ (2022) Economic evaluations
of remote patient monitoring for chronic disease: a systematic review. Value Health 25(6):897–
913
42. Abekah-Nkrumah G, Antwi M, Attachey AY, Janssens W, Rinke de Wit TF (2022) Readi-
ness of Ghanaian health facilities to deploy a health insurance claims management software
(CLAIM-it). PLoS One 17(10):e0275493
43. Thenmozhi M, Dhanalakshmi R, Geetha S, Valli R (2021) WITHDRAWN: implementing
blockchain technologies for health insurance claim processing in hospitals. Elsevier
44. Deluca JM, Enmark R (2000) E-health: the changing model of healthcare. Front Health Serv
Manage 17(1):3–15
45. Desai RJ et al (2021) Broadening the reach of the FDA sentinel system: a roadmap for
integrating electronic health record data in a causal analysis framework. NPJ Digit Med
4(1):170
46. Ho CWL, Ali J, Caals K (2020) Ensuring trustworthy use of artificial intelligence and big
data analytics in health insurance. Bull World Health Organ 98(4):263
47. Baumfeld Andre E, Reynolds R, Caubel P, Azoulay L, Dreyer NA (2020) Trial designs using
real-world data: the changing landscape of the regulatory approval process. Pharmacoepi-
demiol Drug Saf 29(10):1201–1212
48. Pradhan B, Bhattacharyya S, Pal K (2021) IoT-based applications in healthcare devices. J
Healthc Eng 2021:1–18
49. Frikha T, Chaari A, Chaabane F, Cheikhrouhou O, Zaguia A (2021) Healthcare and fitness
data management using the IoT-based blockchain platform. J Healthc Eng 2021
50. Azbeg K, Ouchetto O, Andaloussi SJ (2022) BlockMedCare: a healthcare system based on
IoT, blockchain and IPFS for data management security. Egypt Inform J 23(2):329–343
51. Esposito C, De Santis A, Tortora G, Chang H, Choo K-KR (2018) Blockchain: a panacea for
healthcare cloud-based data security and privacy? IEEE Cloud Comput 5(1):31–37
52. Abounassar EM, El-Kafrawy P, Abd El-Latif AA (2022) Security and interoperability issues
with internet of things (IoT) in healthcare industry: a survey. In: Security and privacy
preserving for IoT and 5G networks: techniques, challenges, and new directions, pp 159–189
53. Rajeswari S, Ponnusamy V (2022) AI-based IoT analytics on the cloud for diabetic
data management system. In: Integrating AI in IoT analytics on the cloud for healthcare
applications. IGI Global, pp 143–161
54. Hong K-W, Oh B-S (2010) Overview of personalized medicine in the disease genomic era.
BMB Rep 43(10):643–648
55. Beccia F et al (2022) An overview of personalized medicine landscape and policies in the
European Union. Eur J Public Health 32(6):844–851
56. Offit K (2011) Personalized medicine: new genomics, old lessons. Hum Genet 130:3–14
57. Santaló J, Berdasco M (2022) Ethical implications of epigenetics in the era of personalized
medicine. Clin Epigenet 14(1):1–14
58. McGowan ML, Settersten RA Jr, Juengst ET, Fishman JR (2014) Integrating genomics into
clinical oncology: ethical and social challenges from proponents of personalized medicine.
In: Urologic oncology: seminars and original investigations. Elsevier, pp 187–192
59. Veenstra DL, Mandelblatt J, Neumann P, Basu A, Peterson JF, Ramsey SD (2020) Health
economics tools and precision medicine: opportunities and challenges. In: Forum for health
economics and policy. De Gruyter, p 20190013
Future of Electronic Healthcare Management: Blockchain and Artificial … 215

60. Meyer MA (2023) A patient’s journey to pay a healthcare bill: it’s way too complicated. J
Patient Exp 10:23743735231174760
61. Al Barazanchi I et al (2022) Blockchain: the next direction of digital payment in drug purchase.
In: 2022 International congress on human-computer interaction, optimization and robotic
applications (HORA). IEEE, pp 1–7
62. Britton JR (2015) Healthcare reimbursement and quality improvement: integration using the
electronic medical record: comment on “fee-for-service payment—an evil practice that must
be stamped out?” Int J Health Policy Manag 4(8):549
63. Yaqoob I, Salah K, Jayaraman R, Al-Hammadi Y (2021) Blockchain for healthcare data
management: opportunities, challenges, and future recommendations. Neural Comput Appl
1–16
64. Kommunuri J (2022) Artificial intelligence and the changing landscape of accounting: a
viewpoint. Pac Account Rev 34(4):585–594
65. Asghar MR, Lee T, Baig MM, Ullah E, Russello G, Dobbie G (2017) A review of privacy and
consent management in healthcare: a focus on emerging data sources. In: 2017 IEEE 13th
international conference on e-Science (e-Science). IEEE, pp 518–522
66. Rantos K, Drosatos G, Kritsas A, Ilioudis C, Papanikolaou A, Filippidis AP (2019) A
blockchain-based platform for consent management of personal data processing in the IoT
ecosystem. Secur Commun Netw 2019:1–15
67. Maher M, Khan I, Prikshat V (2023) Monetisation of digital health data through a GDPR-
compliant and blockchain enabled digital health data marketplace: a proposal to enhance
patient’s engagement with health data repositories. Int J Inf Manag Data Insights 3(1):100159
68. Martin C et al (2022) The ethical considerations including inclusion and biases, data protec-
tion, and proper implementation among AI in radiology and potential implications. Intell
Based Med 100073
69. Zeng D, Cao Z, Neill DB (2021) Artificial intelligence–enabled public health surveillance—
from local detection to global epidemic monitoring and control. In: Artificial intelligence in
medicine. Elsevier, pp 437–453
70. Khoury MJ, Armstrong GL, Bunnell RE, Cyril J, Iademarco MF (2020) The intersection of
genomics and big data with public health: opportunities for precision public health. PLoS
Med 17(10):e1003373
71. MacKinnon GE, Brittain EL (2020) Mobile health technologies in cardiopulmonary disease.
Chest 157(3):654–664
72. Iwaya LH, Fischer-Hübner S, Åhlfeldt R-M, Martucci LA (2018) mhealth: a privacy threat
analysis for public health surveillance systems. In: 2018 IEEE 31st international symposium
on computer-based medical systems (CBMS). IEEE, pp 42–47
73. Aiello AE, Renson A, Zivich P (2020) Social media- and internet-based disease surveillance
for public health. Annu Rev Public Health 41:101
74. Chiou H, Voegeli C, Wilhelm E, Kolis J, Brookmeyer K, Prybylski D (2022) The future of
infodemic surveillance as public health surveillance. Emerg Infect Dis 28(Suppl 1):S121
75. Mello MM, Wang CJ (1979) Ethics and governance for digital disease surveillance. Science
368(6494):951–954
76. Juravle G, Boudouraki A, Terziyska M, Rezlescu C (2020) Trust in artificial intelligence for
medical diagnoses. Prog Brain Res 253:263–282
77. MacRitchie N, Frleta-Gilchrist M, Sugiyama A, Lawton T, McInnes IB, Maffia P (2020)
Molecular imaging of inflammation—current and emerging technologies for diagnosis and
treatment. Pharmacol Ther 211:107550
78. Washington P et al (2020) Data-driven diagnostics and the potential of mobile artificial intelli-
gence for digital therapeutic phenotyping in computational psychiatry. Biol Psychiatry Cogn
Neurosci Neuroimaging 5(8):759–769
79. Paul S, Vidusha K, Thilagar S, Lakshmanan DK, Ravichandran G, Arunachalam A (2022)
Advancement in the contemporary clinical diagnosis and treatment strategies of insomnia
disorder. Sleep Med 91:124–140
216 P. Verma et al.

80. Kaye DK (2023) Addressing ethical issues related to prenatal diagnostic procedures. Matern
Health Neonatol Perinatol 9(1):1–9
81. Chong AY-L, Liu MJ, Luo J, Keng-Boon O (2015) Predicting RFID adoption in healthcare
supply chain from the perspectives of users. Int J Prod Econ 159:66–75
82. Coussement K, Lessmann S, Verstraeten G (2017) A comparative analysis of data preparation
algorithms for customer churn prediction: a case study in the telecommunication industry.
Decis Support Syst 95:27–36
83. Johnston SS, Morton JM, Kalsekar I, Ammann EM, Hsiao C-W, Reps J (2019) Using machine
learning applied to real-world healthcare data for predictive analytics: an applied example in
bariatric surgery. Value Health 22(5):580–586
84. Lanier P, Rodriguez M, Verbiest S, Bryant K, Guan T, Zolotor A (2020) Preventing infant
maltreatment with predictive analytics: applying ethical principles to evidence-based child
welfare policy. J Fam Violence 35:1–13
85. Luk JW, Pruitt LD, Smolenski DJ, Tucker J, Workman DE, Belsher BE (2022) From everyday
life predictions to suicide prevention: clinical and ethical considerations in suicide predictive
analytic tools. J Clin Psychol 78(2):137–148
86. Jeddi Z, Bohr A (2020) Remote patient monitoring using artificial intelligence. In: Artificial
intelligence in healthcare. Elsevier, pp 203–234
87. Shaik T et al (2023) Remote patient monitoring using artificial intelligence: current state,
applications, and challenges. Wiley Interdiscip Rev Data Min Knowl Discov 13(2):e1485
88. Elango S, Manjunath L, Prasad D, Sheela T, Ramachandran G, Selvaraju S (2023) Super
artificial intelligence medical healthcare services and smart wearable system based on IoT
for remote health monitoring. In: 2023 5th International conference on smart systems and
inventive technology (ICSSIT). IEEE, pp 1180–1186
89. Leung R (2023) Using AI–ML to augment the capabilities of social media for tele-health and
remote patient monitoring. In: Healthcare, MDPI, p 1704
90. Palanisamy P, Padmanabhan A, Ramasamy A, Subramaniam S (2023) Remote patient activity
monitoring system by integrating IoT sensors and artificial intelligence techniques. Sensors
23(13):5869
91. Sun W, Zheng W, Simeonov A (2017) Drug discovery and development for rare genetic
disorders. Am J Med Genet A 173(9):2307–2322
92. Kiriiri GK, Njogu PM, Mwangi AN (2020) Exploring different approaches to improve the
success of drug discovery and development projects: a review. Futur J Pharm Sci 6(1):1–12
93. Cerchia C, Lavecchia A (2023) New avenues in artificial-intelligence-assisted drug discovery.
Drug Discov Today 103516
94. Patel V, Shah M (2022) Artificial intelligence and machine learning in drug discovery and
development. Intell Med 2(3):134–140
95. Turanli B, Karagoz K, Gulfidan G, Sinha R, Mardinoglu A, Arga KY (2018) A network-based
cancer drug discovery: from integrated multi-omics approaches to preci-sion medicine. Curr
Pharm Des 24(32):3778–3790
96. Zhou Y et al (2021) AlzGPS: a genome-wide positioning systems platform to catalyze multi-
omics for Alzheimer’s drug discovery. Alzheimers Res Ther 13(1):1–13
97. Cai Z, Poulos RC, Liu J, Zhong Q (2022) Machine learning for multi-omics data integration
in cancer. iScience
98. Syvänen S, Valentini C (2020) Conversational agents in online organization–stakeholder inter-
actions: a state-of-the-art analysis and implications for further re-search. J Commun Manag
24(4):339–362
99. Dave T, Athaluri SA, Singh S (2023) ChatGPT in medicine: an overview of its applica-
tions, advantages, limitations, future prospects, and ethical considerations. Front Artif Intell
6:1169595
100. van der Schyff EL, Ridout B, Amon KL, Forsyth R, Campbell AJ (2023) Providing self-led
mental health support through an artificial intelligence-powered chat bot (Leora) to meet the
demand of mental health care. J Med Internet Res 25:e46448
Future of Electronic Healthcare Management: Blockchain and Artificial … 217

101. Nadarzynski T, Miles O, Cowie A, Ridge D (2019) Acceptability of artificial intelligence (AI)-
led chatbot services in healthcare: a mixed-methods study. Digit Health 5:2055207619871808
102. Kadariya D, Venkataramanan R, Yip HY, Kalra M, Thirunarayanan K, Sheth A (2019)
kBot: knowledge-enabled personalized chatbot for asthma self-management. In: 2019 IEEE
international conference on smart computing (SMARTCOMP). IEEE, pp 138–143
103. Miura C, Chen S, Saiki S, Nakamura M, Yasuda K (2022) Assisting personalized healthcare
of elderly people: developing a rule-based virtual caregiver system using mobile chatbot.
Sensors 22(10):3829
104. Adegun AA, Viriri S, Ogundokun RO (2021) Deep learning approach for medical image
analysis. Comput Intell Neurosci 2021:1–9
105. Li Y, Zhao J, Lv Z, Li J (2021) Medical image fusion method by deep learning. Int J Cognit
Comput Eng 2:21–29
106. Afshar P et al (2020) DRTOP: deep learning-based radiomics for the time-to-event outcome
prediction in lung cancer. Sci Rep 10(1):12366
107. Severn C, Suresh K, Görg C, Choi YS, Jain R, Ghosh D (2022) A pipeline for the implemen-
tation and visualization of explainable machine learning for medical imaging using radiomics
features. Sensors 22(14):5205
108. Liu X et al (2021) Advances in deep learning-based medical image analysis. Health Data Sci
2021 (2021)
109. Graffigna G, Barello S, Bonanomi A, Lozza E (2015) Measuring patient engagement: devel-
opment and psychometric properties of the Patient Health Engagement (PHE) scale. Front
Psychol 6:274
110. Yardley L et al (2016) Understanding and promoting effective engagement with digital
behavior change interventions. Am J Prev Med 51(5):833–842
111. Zheng Z et al (2022) Patient engagement as a core element of translating clinical evidence into
practice-application of the COM-B model behaviour change model. Disabil Rehabil 1–10
112. Jeyakumar T et al (2023) Preparing for an artificial intelligence-enabled future: patient
perspectives on engagement and health care professional training for adopting artificial
intelligence technologies in health care settings. JMIR AI 2(1):e40973
113. Giovanelli A et al (2023) Supporting adolescent engagement with artificial intelligence—
driven digital health behavior change interventions. J Med Internet Res 25:e40306
114. Campbell D (2018) Combining ai and blockchain to push frontiers in healthcare. http://www.
macadamian.com/2018/03/16/combining-ai-andblockchain-in-healthcare/, vol online
115. Marwala T, Xing B (2018) Blockchain and artificial intelligence. arXiv preprint arXiv:1802.
04451
116. Mamoshina P et al (2018) Converging blockchain and next-generation artificial intelligence
technologies to decentralize and accelerate biomedical research and healthcare. Oncotarget
9(5):5665
117. Magazzeni D, McBurney P, Nash W (2017) Validation and verification of smart contracts: a
research agenda. Computer (Long Beach Calif) 50(9):50–57
118. Bohr A, Memarzadeh K (2020) The rise of artificial intelligence in healthcare applications.
In: Artificial intelligence in healthcare. Elsevier, pp 25–60
119. Khezr S, Moniruzzaman M, Yassine A, Benlamri R (2019) Blockchain technology in
healthcare: a comprehensive review and directions for future research. Appl Sci 9(9):1736
120. Behner P, Hecht M-L, Wahl F (2017) Fighting counterfeit pharmaceuticals: new defenses for
an underestimated and growing menace, vol 12, p 2017
121. Tran V-T, Riveros C, Ravaud P (2019) Patients’ views of wearable devices and AI in healthcare:
findings from the ComPaRe e-cohort. NPJ Digit Med 2(1):53
122. Jennath HS, Anoop VS, Asharaf S (2020) Blockchain for healthcare: secur-ing patient data
and enabling trusted artificial intelligence
123. Ahuja AS (2019) The impact of artificial intelligence in medicine on the future role of the
physician. PeerJ 7:e7702
218 P. Verma et al.

124. Bai Q, Tan S, Xu T, Liu H, Huang J, Yao X (2021) MolAICal: a soft tool for 3D drug
design of protein targets by artificial intelligence and classical algorithm. Brief Bioinform
22(3):bbaa161
125. Krittanawong C et al (2020) Integrating blockchain technology with artificial intelli-gence
for cardiovascular medicine. Nat Rev Cardiol 17(1):1–3
Impact of Neural Network on Malware
Detection

Aaeen Alchi, Kiranbhai Dodiya, and Varayogula Sai Niveditha

Abstract Attacks by malware have significantly increased during the last several
years, endangering the security of computer systems and networks. The continually
shifting landscape of malware assaults makes it challenging for traditional techniques
like a rule- and signature-based detection to stay up. Consequently, researchers are
now looking at more cutting-edge methods, including neural networks, to increase
the precision of malware detection. In assessing new malware by comparable traits,
neural networks may gather data from previously discovered malware. This method
makes it possible to quickly and reliably identify malware, which is essential for
stopping infestation spread. Finding patterns and connections among massive data
sets is one of the main benefits of utilising neural networks to detect malware.
However, there are some issues with employing neural networks to find viruses.
One of the main difficulties is that the network needs a lot of labelled data training.
Effective evasion attacks are also feasible, in which fraudsters try to control neural
networks by feeding them harmful material designed to elude detection. Despite
these difficulties, researchers are still exploring neural networks’ potential for virus
detection. This book chapter provides a detailed analysis of how neural networks
affect malware detection. Along with an outline of the current status of the research
area, updates on neural network topologies and training methods are provided. We’ll
discuss possible advancements and future possibilities in the industry, including
reinforcement learning and generative adversarial networks for malware detection.
The first section of the chapter summarises the essential characteristics of neural
networks, including their capacity to identify patterns and abnormalities in massive
amounts of data. The benefits and drawbacks of using neural networks to detect
malware are highlighted. Finally, the extensive usage of neural networks in this
book chapter’s malware detection analysis is highlighted. Researchers may greatly
enhance computer and network security and better protect against the rising danger
of assaults by malware by using neural networks’ complicated capabilities.

A. Alchi (B) · K. Dodiya

Department of Biochemistry and Forensic Science, Gujarat University, Ahmedabad, Gujarat, India
e-mail: [email protected]
V. S. Niveditha
Digital Forensics, Rashtriya Raksha University, Lavad, Gandhinagar, Gujarat, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 219
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_10
220 A. Alchi et al.

Keywords Malware · Signature-based · Detection · Examination · Large data

sets · Neural Network

1 A Summary of Malware Detection’s Current State

The complexity and volume of malware assaults are swellings, which pose a grim
threat to the safety of computer structures and networks and are a contemporary
hallmark of malware detection. A broad spectrum of harmful software, including
viruses, worms, Trojan horses, and ransomware, are called Malware. Malware attacks
may cause data loss, system outages, financial losses, and brand damage.
The ever-evolving nature of malware attacks has rendered conventional malware
detection methods like rule-based and signature-based detection ineffective. Experts
may identify malicious software and stop using signature-based detection, which
relies on known malware patterns but can easily be evaded by attackers using new or
updated malware versions. Rule-based detection finds and disables Malware using
predetermined criteria but may produce false positives and negatives [13].
As a result, researchers and security professionals have turned to more cutting-
edge technology to improve the precision and efficacy of malware detection.
However, these approaches rely on massive extents of data to teach computers to
spot patterns and anomalies in Malware, and they may also react to new, unforeseen
threats.
However, there are still challenges in malware detection, including the necessity
for extensive and diverse training datasets, the possibility that attackers would evade
detection by using cunning tactics, and the need to strike a balance concerning detec-
tion accuracy and processing speed. In addition, researchers and security specialists
must keep coming up with fresh ideas and methods to spot and stop malware attacks
as they increase in frequency.

1.1 More Complex Techniques Are Required

Experts must develop malware detection systems with ever-greater sophistication due
to the complexity and culture of malware attacks. Malware developers continuously
formulate innovative strategies to avoid detection and entering computer systems and
networks. The upshot is that conventional signature-based and rule-based detection
techniques often must be in sync with the changing threat environment.
More complicated technologies like machine learning and neural networks have
emerged to satisfy this need. These technologies can improve malware detection
accuracy and efficacy by utilising enormous datasets and training algorithms to
uncover patterns and irregularities associated with Malware.
Furthermore, the number of possible entry points for malware attacks grows as
connected devices and the Internet of Things (IoT) increase. As a result, there is
Impact of Neural Network on Malware Detection 221

a particular need for more complicated detection algorithms capable of identifying

and stopping threats across several platforms and devices.
In addition to these technological challenges, regulatory and compliance demands
necessitate enhanced detection approaches. In Europe, for example, the General
Data Protection Regulation (GDPR) requires commerce to put in place appropriate
technological and organisational protections to guarantee the security of personal
data. As a result, powerful malware detection techniques must be used to avoid data
breaches and other security calamities.
Overall, the rising complexity and sophistication of malware assaults and modern
information security’s legal and regulatory requirements drive the demand for more
sophisticated malware detection technologies.

1.2 Artificial Neural Networks

Neural networks are machine learning algorithms that simulate the human brain’s
structure and function. They comprise layers of linked nodes called neurons that
process input data and anticipate output. Each neuron in the preceding layer gets
input from other neurons, computes, and then sends the result to the next layer.
Since they can understand patterns and abnormalities associated with Malware
even in vast and complicated datasets, neural networks have become popular in detec-
tion. They can also constantly adapt to new and undiscovered dangers by learning
from new data.
Feedforward neural networks, recurrent neural networks, and convolutional neural
networks are among the neural networks utilised in malware detection. Liable on
the nature of the data and the detection job, each kind has different strengths and
disadvantages (Jakub Gotthans et al., n.d.).
One of the benefits of neural networks is their capacity to do feature extraction,
which entails finding essential data properties relevant to the detection job. This is
especially beneficial in detecting Malware since characteristics like code obfuscation
and packing can make it challenging to detect Malware using standard approaches.
Yet, there are several drawbacks to using neural networks for malware detection.
One of the difficulties is the requirement for extensive and diverse training datasets,
which might be challenging when dealing with uncommon or novel malware strains.
Moreover, training and evaluating neural networks may be computationally costly,
which might be a limiting issue for real-time detection systems.
Overall, neural networks have emerged as a viable solution to malware detection,
promising more remarkable accuracy and the capability to adapt to new and devel-
oping threats. Their success, however, depends on the quality of the training data and
the neural network’s unique design and configuration.
222 A. Alchi et al.

1.3 Neural Network Overview

Neural networks are machine learning algorithms that mimic the human brain’s
structure and function. They comprise layers of linked nodes called neurons that
process input data and anticipate output.
Each neuron in the preceding layer takes input from other neurons in that layer,
computes using an instigation function, and then transmits the outcome to the next
layer. The neural network’s prediction is the output of the last layer.
Classification, regression, and pattern recognition are just a few tasks that neural
networks can do. Therefore, professions that involve managing intricate and non-
linear relationships between input and output variables can be significantly enhanced
by individuals who possess exceptional skills in this area.
There are several forms of neural networks, each with its structure and function.
Among the most regularly utilised kinds are (Fig. 1):
Feedforward neural networks are the most basic neural networks, with infor-
mation flowing in just one technique, from the input to the output layer. They are
often employed in classification and regression problems [10].
Backpropagation neural networks: Recurrent networks have feedback connec-
tions that allow them to store and comprehend consecutive input across time. They
are often used for time-series data applications, for instance, speech recognition and
natural dialectal processing.
Convolutional neural networks analyse image data using a two-dimensional
pixel input matrix. They may be used for object identification and picture classifica-
tion and employ filters to extract photo information [12].
Backpropagation is a technique for training neural networks that involves
changing the weights and Preferences of the neurons to close the gap between
predicted and observed output. To do this, one must first calculate the gradient of
the loss function concerning the weights and biases before changing the parameters
[12].
Some areas where neural networks have shown to be very successful are image
identification, natural language processing, and voice commands. Nevertheless,

Fig. 1 Form of neural

network
Feedforward Backpropagation
neural networks neural networks

Convolutional
Backpropagation
neural
Impact of Neural Network on Malware Detection 223

they need a substantial amount of training data to achieve high accuracy and are
computationally costly to train.

1.4 Types of Neural Networks

Neural networks come in many different forms, each with its specific structure and
function. Listed below are a few of the most typical types (Fig. 2):

1. Feedforward neural networks: A primary type of neural network that allows

information to flow in a single direction, from the input to the output layer. They
are often used to solve classification and regression problems (Jakub Gotthans
et al., n.d.).
2. Recurrent neural networks: Thanks to their feedback connections, these
networks can store and analyse sequential input across time. They are often used
in time-series data applications, including natural language processing and speech
recognition (Jakub Gotthans et al., n.d.).
3. Convolutional neural networks: These networks interpret picture data and
use a two-dimensional matrix of pixel values as input.
4. Autoencoders: The purpose of these networks is to learn a condensed version
of the input data, which can then be applied for tasks such as data compression,
denoising, and detecting anomalies. These networks acquire a compact under-
standing of the input data, which can be effective until serious charges such as data
compression, denoising, and anomaly detection. They do this by collecting infor-
mation from images using filters, which can be used for functional classification
and object detection (Jakub Gotthans et al., n.d.).
5. GANs (generative adversarial networks): A GAN consists of a generator and
a discriminator. The discriminator separates the created samples from the actual
data, while the generator produces new examples comparable to the original data.
Among other things, GANs may be used to create images and transfer fashion.

Types of neural
networks

Feedforward Recurrent neural Convolutional

Autoencoders GANs
neural networks networks neural networks

Fig. 2 Types of neural networks

224 A. Alchi et al.

6. LSTM networks: These networks are a form of the recurrent neural network
intended to handle consecutive input over extended periods. They’re typically
employed for voice recognition, language translation, and sentiment analysis [4].

Deep belief networks are feedforward neural networks constructing a hierar-

chical input data representation. They are commonly applied in feature extraction,
unsupervised learning, and recommendation systems.
Each neural network style has its benefits and weaknesses and is best suited for
specific tasks and data. Therefore, the neural network utilised is dictated by the
problem’s nature and the data’s characteristics.

1.5 Advantages of Employing Neural Networks to Identify

Malware

The use of neural networks for malware detection has various advantages (Fig. 3):

1. Learning patterns: As neural networks can learn ways and identify abnormal-
ities in big datasets, they are highly suited for malware detection. Even when
standard signature-based approaches fail, they may uncover minute changes in
code or behaviour that may suggest Malware.
2. Adaptability: Neural networks can adjust to new kinds of Malware and evolving
threats. They may learn from new instances and increase their detecting skills.
3. Speed: Since neural networks can handle enormous volumes of data fast and
effectively, they are well suited for real-time detection and reaction.
4. Scalability: Since neural networks can be scaled up or down based on the amount
and complexity of the data collection, they can be used in large-scale systems.
5. Accuracy: When trained on vast and diverse datasets, neural networks may reach
significant levels of accuracy in malware detection.
6. Automated feature extraction: Neural networks can extract features from data
automatically, removing manual feature engineering requirements. This can save
time and lower the likelihood of human mistakes.

Fig. 3 Advantages of Learning patterns

Employing Neural Networks
Advantages
to Identify Malware Adaptability

Speed

Scalability

Accuracy

Automated feature extraction

Low false positive rate

Impact of Neural Network on Malware Detection 225

7. Low false positive rate: since neural networks have a low false positive rate,
they are less likely to mistake innocuous software for Malware.
Overall, using neural networks for malware detection is a viable way to tackle the
rising sophistication of malware assaults and enhance computer system and network
security.

1.6 Neural Network Limitations for Malware Detection

While neural networks provide various advantages for malware detection, there are
several limits to be aware of (Fig. 4):
1. Black box model: Neural networks are sometimes called “black box” models
since it might be challenging to comprehend how the network makes choices.
This makes interpreting the data and identifying the exact traits utilised to detect
Malware challenging.
2. Data bias: A neural network model’s accuracy depends on the training data’s
quality and representativeness. If the training data are skewed or inadequate, the
model may fail to recognise new varieties of Malware.
3. Overfitting: When training data are excessively fed to neural networks, they
become overly specific and incapable of adapting to new information. This can
ultimately result in poor detection of unique forms of Malware that were not
present during their initial training.
4. Adversarial attacks: Malware makers can build their software to avoid detec-
tion by neural networks, potentially resulting in an arms race between malware
creators and malware detection systems.

Fig. 4 Neural network

limitations for malware Black box
detection model

Regulatory
Data bias
problems

Limitations

Computatio
nal Overfitting
resources

Adversarial
attacks
226 A. Alchi et al.

5. Computational resources: Since neural networks are computationally expen-

sive, significant resources are required for training and deployment, which can
be a hurdle for smaller firms with low resources.
6. Regulatory problems: Using neural networks to identify Malware may raise
worries about data privacy and security, as well as ethical questions regarding
the use of AI in security-related applications.
Thus, while neural networks hold great potential for enhancing malware detection,
they also have limits that must be examined and addressed to be successful and ethical.

2 The Use of Neural Networks in Malware Detection

2.1 Data Pre-Processing for Neural Network Input

Pre-processing is required to ensure the quality and relevance of the input data before
feeding it to a neural network for malware detection. Pre-processing is a sequence of
procedures that clean, standardise, and convert input so the neural network can use
it. Some popular neural network input pre-processing processes include (Fig. 5):
Data cleaning includes deleting unneeded or superfluous data and dealing with
missing or incorrect values.
Data normalisation is scaling the data to a specified range or distribution to
guarantee that each attribute is given equal weight.
Feature selection entails picking a subset of essential characteristics from the data
to decrease the model’s complexity and enhance performance.
Data transformation entails changing the data into a more acceptable format for
neural network input. Text data, for example, could need to be tokenised and trans-
formed to a numerical representation, while image data would need to be scaled or
converted to grayscale.
Data augmentation is new synthetic data generated by applying changes to the
original data, such as rotation or cropping. Data augmentation can assist in increasing
the neural network’s accuracy and generalisation [3].
The pre-processing stages for neural network input will vary depending on the
type of data utilised for malware detection and the neural network model’s needs.
Therefore, it is critical to carefully evaluate the pre-processing stages to ensure the

Fig. 5 Data pre-processing Data cleaning

for neural network input
Data normalization

Feature selection

Data transformation

Data augmentation
Impact of Neural Network on Malware Detection 227

input data’s quality and relevance and enhance the neural network’s performance for
malware detection.

2.2 Neural Network Training for Malware Detection

Many steps are involved in training a neural network for malware detection,
including:
Designing the network architecture entails selecting the proper type of neural
network, calculating the number of layers and nodes, and deciding on the activation
functions for each layer.
• Preparing the training data entails pre-processing the data to ensure that it is in a
format appropriate for training the neural network.
• Dividing the data into training and validation sets entails partitioning it into two
groups: training the neural network and assessing its performance during training.
• Setting hyperparameters entails adjusting the learning rate, batch size, and other
training-related factors.
• Training the network entails entering training data into the neural network and
changing the network’s weights and biases to minimise the error between expected
and actual output.
• Assessing the network entails running the trained neural network on the validation
set to evaluate its performance and, if necessary, make improvements.
• Testing the network entails running the trained neural network on a different test
set to assess its performance on previously unknown data.
• Tuning the network entails adjusting the hyperparameters and network design to
optimise performance on test data.
To ensure the accurate detection of Malware and prevent false positives, it is
essential to train a neural network, even though it may require significant time and
computational resources. Therefore, the quality and relevance of the training data,
as well as the choice of appropriate hyperparameters and network design, are all
crucial elements that can substantially influence the neural network’s performance
for malware detection.

2.3 Performance Evaluation of Neural Networks

Assessing neural network performance is essential to ensuring the network detects

Malware effectively while avoiding false positives. Many measures are used to
determine a neural network’s effectiveness in malware detection, including:
Accuracy: This parameter measures the neural network’s proportion of correct
classifications.
228 A. Alchi et al.

Precision is the fraction of successfully detected malware samples among all

malware samples categorised by the neural network.
Remember that this statistic counts the proportion of successfully recognised
malware samples among all malware samples in the dataset.
The F1 score calculates the harmonic mean of accuracy and recall, providing a
fair evaluation of both factors. The ROC curve is a graph that shows the effectiveness
of a neural network in different classification scenarios. The AUC, or area under the
ROC curve, is a widely used overall performance measure.
Confusion matrix: The breakdown of categorisation findings is presented in
this matrix, providing a detailed account of the true positives, true negatives, false
positives, and false negatives.
The false positive rate (FPR) and false negative rate (FNR) are metrics that assess
the proportion of false positives and false negatives among all malware samples
categorised by the neural network [1].
The malware detection application’s unique requirements determine the selection
of relevant assessment metrics. In specific applications, for example, avoiding false
positives may be more essential than maximising recall, but in others, the converse
may be true. It is critical to carefully consider the assessment criteria and employ suit-
able approaches to analyse the neural network’s performance for malware detection
effectively [11].

2.4 Evaluation of Neural Network Performance

in Comparison to Alternative Malware Detection
Approaches

Neural networks have exhibited promising results in malware detection, outper-

forming standard malware detection approaches in several circumstances. Following
are some performance comparisons of neural networks to different malware detection
techniques:
1. Signature-based detection compares a file’s code or behaviour to a database of
known malware signatures. While signature-based detection can be successful
against known Malware, it is vulnerable to zero-day assaults and new malware
variants. By understanding the fundamental patterns of malware activity, neural
networks can detect new and undiscovered viruses, making them more successful
than signature-based detection [14].
2. Heuristic-based detection entails studying a file’s activity and determining its
maliciousness based on specified criteria or heuristics. While heuristic detec-
tion can be helpful, it can also generate many false positives. By learning from
a massive collection of benign and malicious samples, neural networks can
minimise the number of false positives [14].
3. Machine learning-based detection: This method entails developing a model to
identify files based on their attributes. While machine learning can be helpful,
Impact of Neural Network on Malware Detection 229

it frequently necessitates extensive feature engineering, which can be time-

consuming and may fail to collect all essential information. Neural networks
can learn characteristics from data automatically, making them more effective
than standard machine learning approaches [5].
4. Behaviour-based detection to assess a file’s maliciousness; behaviour-based
detection analyses its activity in a sandboxed environment. While behaviour-
based detection can be helpful, it is computationally demanding and may miss
Malware that can elude detection in a sandbox. Without a sandboxed environ-
ment, neural networks can evaluate file activity and highly accurately detect
malware [14].
Overall, neural networks outperformed traditional malware detection algorithms
in several circumstances. On the other hand, the success of neural networks is deter-
mined by the quality and relevancy of the input data and the design of the neural
network architecture. Therefore, it is critical to thoroughly examine the efficacy of
various malware detection algorithms and select the best solution for the individual
application.

2.5 The Present Status of Neural Network-Based Malware

Detection Research

Neural network-based malware detection is an active study area, with notable

breakthroughs in recent years. Among the current research areas are:
Transfer learning is employing a previously trained neural network for a different
job and fine-tuning it for malware detection. This method can minimise the training
data necessary while improving neural network performance.
Adversarial attacks entail altering the input to a neural network to avoid detection
or generate a false positive. As a result, researchers are looking for ways to increase
the robustness of neural networks in malware detection against adversarial assaults.
Explainability and interpretability: Neural networks may be difficult to inter-
pret, which can be an issue in malware detection applications where understanding
the reasoning for a particular categorisation result is critical. Researchers are working
on techniques to increase the explainability and interpretability of neural network
models used to identify Malware.
Hybrid models enhance malware detection performance by combining various
detection methods, including signature-based and neural network-based detection.
Edge computing is processing at the network’s edge, such as on a mobile device
or IoT device, rather than in a centralised server. Researchers are researching using
neural networks for edge-based malware detection to increase the detection process’s
efficiency and scalability.
The present level of research in neural network-based malware detection is focused
on enhancing neural network models’ accuracy, speed, and robustness. As neural
networks progress, they more successfully detect and prevent malware assaults.
230 A. Alchi et al.

2.6 Current Developments in Neural Network Designs

and Training Methods

Recent advances in neural network topologies and training methodologies have

helped to produce more powerful and effective malware detection models. Among
the noteworthy advancements in this field are:
CNNs (convolutional neural networks): CNNs are a neural network that has
proven highly successful in image recognition applications. They have been used
to identify Malware by portraying binary data as pictures and extracting characteris-
tics from these images using CNNs. This method has been demonstrated to be highly
efficient in identifying Malware, with the bonus of viewing the features used by the
model for categorisation [8].
RNNs are neural networks that can process sequential data, such as time series
or text data. They have been used to identify Malware by expressing a program’s
activities as a series of API or system calls. RNNs may then be trained on this data
to detect aberrant activity that may indicate Malware.
GANs (generative adversarial networks) are neural networks that can produce
synthetic data comparable to actual data. They’ve been used to identify Malware by
creating simulated malware samples that may be used to train and evaluate neural
network models. This method could create more diverse and representative data for
neural network training, improving their performance (Maria Rigaki & Sebastian
Garcia, 2018).
Transfer learning is employing a previously trained neural network for a different
job and fine-tuning it for malware detection. This method can minimise the training
data necessary while improving neural network performance.
Adversarial training adds adversarial samples to the training data to increase
the neural network’s robustness against hostile attacks. This method has been found
to improve the resilience of neural network models for malware detection.
Overall, advances in neural network topologies and training methodologies have
aided in creating increasingly powerful and effective malware detection models.
However, more complex and successful models are expected to be completed as
research in this field develops.

2.7 Current Research’s Challenges and Limits

Despite the encouraging outcomes of neural network-based malware detection,

several obstacles and limits in current research must be solved. Among these
difficulties are:
Malware data are frequently uneven, with a small number of malware samples
outnumbering many benign ones. This might result in skewed models and lower
detection rates for malware samples. Data augmentation and class weighting must
be used to b data and enhance model performance to address this issue.
Impact of Neural Network on Malware Detection 231

Malware makers can employ adversarial tactics to change malware code to avoid
detection by neural networks. This can lead to false negatives and lower detection
rates. There is continuing research to construct robust neural network models capable
of detecting adversarial assaults and improving detection rates.
Interpretability: Because neural networks are generally regarded as “black box”
models, it might be challenging to grasp how they make their predictions. This lack of
interpretability can make identifying and correcting model faults easier. Techniques
for improving the interpretability of neural network models for malware detection
are being developed.
Overfitting: Neural networks can overfit specific datasets, resulting in poor perfor-
mance on new, previously unknown data. Creating neural network models that
generalise effectively to new data is an ongoing issue in malware detection research.
Scalability: Since neural networks take significant computer resources and time
to train, they might be challenging to scale to massive datasets or real-time detec-
tion applications. Researchers are working to build more efficient neural network
topologies and training methodologies to solve this difficulty.
Overall, resolving these issues and constraints is crucial for progressing in neural
network-based malware detection and increasing computer system and network
security.

2.8 Case Studies of Effective Neural Network-Based

Malware Detection Systems

Numerous effective neural network-based malware detection systems have been

created in recent years. Here are a couple of such examples:
DeepMal: DeepMal is a deep learning-based malware detection solution that anal-
yses portable executable (PE) files using a convolutional neural network (CNN). The
method beat numerous classic machine learning-based malware detection systems
in detecting known and unknown malware samples [2].
N-BaIoT is a neural network-based solution that detects Malware in IoT devices.
To monitor network traffic and detect Malware, the system employs a hybrid neural
network design that combines a CNN and a long short-term memory (LSTM)
network. As a result, the system detected Malware with great accuracy and beat
numerous state-of-the-art malware detection methods.
MalwareNet is a deep neural network-based solution that detects Malware in
Android apps. Our system uses a deep neural network to detect malware by analysing
permissions, API calls, and file actions. It’s more effective than traditional methods,
identifying known and unknown samples.
DeepHunter: DeepHunter is a malware detection solution that uses deep learning
to detect new and undiscovered malware samples. The system analyses the binary
properties of the malware sample and sees similarities with known malware families
232 A. Alchi et al.

using a stacked autoencoder neural network. The approach surpassed various state-
of-the-art malware detection techniques to detect unknown malware samples [9].
Overall, these neural network-based malware detection systems indicate the
promise of deep learning in enhancing malware detection accuracy and efficacy,
and they serve as a foundation for future study in this field.

3 Future Directions and Possible Developments

GANs (generative adversarial networks) have lately been investigated as a viable

solution for virus detection. GANs comprise two neural networks: a generator and a
discriminator.
The generator produces new data samples that mimic the distribution of actual
malware samples, while the discriminator differentiates between authentic and fabri-
cated models. Both networks are trained together, with the generator striving to create
convincing malware samples that can deceive the discriminator and the discriminator
trying to detect which examples are legitimate and artificially generated. The use of
GANs for malware detection offers various possible benefits. One advantage is that
the generator may generate synthetic malware samples comparable to real-world
Malware even if the training dataset is restricted. Another advantage of GANs is that
they can detect new forms of Malware that were not included in the training dataset
by producing samples that were never seen before (Maria Rigaki & Sebastian Garcia,
2018).
Unfortunately, there are significant obstacles and limits to using GANs for
malware detection. One disadvantage is that GANs can be challenging to train since
the two networks must be properly balanced to avoid one network overwhelming the
other. Another issue is that GANs are vulnerable to adversarial assaults, in which an
attacker creates samples tailored to deceive the discriminator and avoid detection.
Notwithstanding these obstacles, research into the use of GANs for malware
detection is underway, and this technique can potentially increase the accuracy and
efficacy of malware detection systems.

3.1 Detecting Malware via Reinforcement Learning

The process of educating an agent to make choices in situations where rewards and
penalties are given is known as reinforcement learning or RL. The viability of using
RL to enhance malware detection has been examined. Using RL-based malware
detection, an agent learns to identify malicious software or processes based on their
behaviour and the environment in which they operate.
Impact of Neural Network on Malware Detection 233

The agent receives bonuses for correctly identifying Malware and suffers conse-
quences for false positives or negatives. The objective is to teach the agent to accu-
rately determine if a specific program is Malware, even if the infection is intricate
and capable of eluding conventional detection techniques.
RL-based malware detection has the benefit of being adaptable to new and
changing disease strains. The agent may adapt its decision-making process when
the infected environment changes due to its experiences.
RL for malware detection has a variety of drawbacks and restrictions, which is
unfortunate. For example, as the agent must investigate various speculative possibil-
ities and learn from its experiences, RL-based systems may be challenging to train.
Another issue is that RL-based systems could be susceptible to adversarial assaults,
in which a perpetrator tries to influence the surroundings to deceive the agent into
doing the wrong action.
Notwithstanding these difficulties, research into the use of RL for malware detec-
tion is ongoing, and this te potentially improves the precision and effectiveness of
malware detection systems.

3.2 Prospects for Neural Network-Based Malware Detection

Future developments in neural network-based malware detection might increase the

accuracy and effectiveness of these systems.
Advanced neural network structures, such as deep neural networks and convolu-
tional neural networks, are a crucial area of study in this field. They are specifically
crafted to handle the complex tasks of malware detection with utmost efficiency. In
addition, the efficiency of neural network-based malware detection may be enhanced
by similar architectures, which have shown promising results in other areas, including
picture and voice recognition.
Another area of research is using more challenging training approaches, such as
transfer learning and meta-learning, which may allow neural networks to incorpo-
rate knowledge gained from earlier tasks or experiences to improve their ability for
malware detection. To increase its ability to detect Malware, a neural network trained
on a large dataset of benign software, for instance, would be able to transmit part of
that knowledge. With neural network-based malware detection systems, combining
several data types, such as network traffic data, system call data, and file information,
is possible.
By mixing several data types, neural networks could spot patterns and anomalies
that aren’t apparent from a single data form.
Lastly, using reinforcement learning and machine learning to create adaptive
malware detection systems that can react to threats shows promise.
To constantly changing virus types. However, with ongoing learning and
adaptation, these systems can stay ahead of the always-evolving danger scenario.
234 A. Alchi et al.

Overall, neural network-based malware detection has several potential advance-

ments, and ongoing research in this area is expected to result in future improvements
in the precision and efficacy of these systems.

3.3 Neural Network Forensic Application for Malware

Detection

The forensic use of neural networks for malware detection entails employing these
networks to examine digital data and identify malware risks. In addition, forensic
analysis is critical for determining the breadth and effects of a malware attack and
implementing appropriate mitigation solutions.
Neural networks may be trained on enormous datasets of known Malware and
benign files to understand the traits and properties of each type of file. Once intro-
duced, these networks may categorise new files as harmful or harmless. This method
can assist forensic investigators in swiftly identifying potentially dangerous data and
prioritising their investigation.
Neural networks may also be used for behavioural analysis, which entails moni-
toring system activity to detect malicious conduct. For example, neural networks
can recognise abnormal movements and highlight possible malware risks by evalu-
ating patterns in system activity. This method can assist investigators in identifying
Malware that has eluded standard detection techniques and give insights into attacker
behaviour and strategies.
The network must be trained on a diverse and comprehensive dataset to
ensure successful malware detection in forensic applications using neural networks.
Outdated or insufficient datasets may hinder the network’s ability to accurately iden-
tify new or advanced threats due to the constant evolution of malware. Nevertheless,
neural networks are simply one tool in the forensic investigator’s toolbox, and its
output must be understood in the context of other evidence and information.
Overall, the forensic use of neural networks for malware detection shows promise
as a solid and efficient technique for identifying and reducing malware risks.
As technology advances, neural networks are anticipated to play an increasingly
crucial role in digital forensics.
Impact of Neural Network on Malware Detection 235

4 Conclusion

4.1 A Description of the Effect of Neural Networks

on Malware Detection

Finally, neural networks have had a significant impact on malware detection. Tradi-
tional malware detection methods can no longer keep up with the increasing sophis-
tication of malware attacks. Because of their ability to learn patterns and detect
irregularities in massive datasets, neural networks have emerged as a potential
method for finding and detecting Malware. When identifying malware, one sound
choice is to deploy neural networks, given its numerous benefits above traditional
methods, including efficient management of large datasets through adaptation when
encountering new threats. Although there are a few drawbacks to deploying this
algorithm l, like exacting superior quality training data while still being prone to
generating false positive or negative outcomes, it is still more efficient compared
with common approaches for detecting malicious activities online. Furthermore,
the recent advancements made through improved tactics and enhanced topologies
depict positive signs concerning the future effectiveness in identifying different types
of malware.
This trend is anticipated to continue as researchers develop new strategies and
methods to improve the precision and effectiveness of neural network-based malware
detection systems.

4.2 Future Prospects and Potential Effects on the Computer

System and Network Security.

1. Improved Detection Accuracy: Neural networks are exceptionally good at iden-

tifying intricate patterns and separating essential elements from big datasets. This
feature enables neural networks to analyse massive volumes of data, including
file structures, behavioural patterns, network traffic, and code fragments, for
malware identification. Furthermore, neural networks may learn complex corre-
lations between these features by training on various malware samples, leading
to the development of precise models for recognising malicious software. As a
result, detection accuracy increases, making it possible to identify known and new
malware strains. In malware detection, neural networks have greatly improved
detection accuracy. Signature-based approaches traditionally used predetermined
patterns and signatures to identify known malware variants. These techniques,
however, could not locate polymorphic and obfuscated malware, which may
change its signature or display variations in its code structure. Neural networks
have overcome this restriction thanks to their ability to learn intricate patterns and
extract valuable characteristics from big datasets. Furthermore, neural networks
236 A. Alchi et al.

may identify subtle markers and associations within the data that suggest mali-
cious behaviour by training on various malware samples. Thanks to this, they
can accurately identify malware versions that are both known and undiscovered.
The ability of neural networks to analyse multiple data sources, including file
structures, behavioural patterns, network traffic, and code fragments, is one of
their most vital points. Neural networks can find hidden patterns and correla-
tions by analysing these factors that might not be visible through manual anal-
ysis or conventional methods. As a result, malware detection is more successful
and has better detection accuracy. Additionally, neural networks are excellent at
managing massive datasets, enabling them to handle enormous volumes of data
quickly. Given the exponential growth in malware samples, this capacity is vital
for malware identification. Since neural networks can analyse this data quickly,
they can identify malware more quickly and precisely.
Furthermore, improved detection accuracy results from neural networks’
capacity to adapt to and learn from new data. Neural networks may update their
models and incorporate this further information into their detection algorithms
as new malware strains appear. Because of their flexibility, neural networks can
efficiently recognise even the most recent and sophisticated malware strains and
stay current with the most recent threats.
2. Robust Feature Extraction: Conventional signature-based approaches mainly
rely on preset patterns and signatures, which limits their efficacy against polymor-
phic and obfuscated malware. On the other hand, neural networks may automat-
ically discover and extract pertinent features from unprocessed data, negating
the need for intentional feature engineering. By picking up on tiny signs of
harmful behaviour, this adaptability enables neural networks to adapt to new
malware types and significantly increases detection rates. The capacity of neural
networks to perform reliable feature extraction is one of their primary advan-
tages in malware detection. In this case, malicious and benign software, signif-
icant traits or patterns must be extracted from raw data to distinguish between
distinct groups or categories. Traditionally, selecting specific qualities consid-
ered indicative of malware required manual selection and subject knowledge.
The complexity and diversity of malware, as well as the laborious process of
feature selection, can put a limit on this technique. On the other hand, neural
networks are excellent at automatically learning and extracting features from
unprocessed data. As a result, there is no longer a requirement for explicit feature
engineering, and the neural network can detect subtle and complicated patterns
that human-designed feature sets would miss. Neural networks can analyse a
wide range of data in the context of malware detection, including binary file
formats, API calls, network traffic, system logs, and behaviour patterns. They can
record data at various abstraction levels, from simple byte sequences to complex
behaviour sequences. Neural networks analyse the input data and derive mean-
ingful representations using layers of interconnected nodes (neurons). These
networks modify the weights and biases of the neurons through a process known
as backpropagation to reduce the discrepancy between the projected output and
the actual output. This optimisation process lets the network learn and recognise
Impact of Neural Network on Malware Detection 237

the most pertinent features for differentiating between malware and good soft-
ware. The ability of neural networks to automatically learn intricate, hierarchical
representations of data gives them an edge in feature extraction. As a result,
the network can recognise clear indicators and abstract, higher-level features
that reveal malware’s fundamental makeup and behaviour. Neural networks,
for instance, can spot recurrent code patterns, strange system calls, or unusual
network traffic that point to malicious activities. Additionally, neural networks
can modify their feature extraction skills to account for newly emerging and
changing malware kinds. Neural networks may generalise their learned features
and use them to identify malware samples that had not previously been observed
by training on various datasets that cover a wide range of malware families and
variants. Nevertheless, it is important to remember that the calibre and represen-
tativeness of the training data significantly impact how well feature extraction
in neural networks performs. Sufficient and diverse training datasets are essen-
tial to guarantee that the network learns pertinent properties that may generalise
successfully to real-world malware samples.
3. Behavioural Analysis: Neural networks are excellent at analysing malware’s
intricate behavioural patterns. Neural networks may learn to distinguish between
legitimate and harmful behaviour by watching how malware interacts with a
system, including file alterations, network communications, and process execu-
tions. This behavioural analysis method works exceptionally well in spot-
ting complex malware that may use sophisticated evasion strategies or display
zero-day traits. Anomalies that deviate from typical system behaviour can be
detected using neural networks, which can also provide alerts. Neural networks
have successfully tackled the vital component of malware detection known as
behavioural analysis. Behavioural analysis entails watching and examining how
software or code interacts and behaves to spot trends or deviations from expected
behaviour. It focuses on comprehending the behaviour displayed by malware in
the context of malware detection to distinguish it from legitimate software.
The capacity of neural networks to learn and recognise intricate patterns in data
makes them particularly effective in behavioural analysis. They can be trained on
big datasets that record both malware and good software behaviours, allowing them
to comprehend the everyday interactions and behaviours of various software kinds.
Here is how neural networks use behavioural analysis to detect malware:
Various aspects of a system’s behaviour, including file modifications, registry
changes, network communications, process executions, and API requests, can be
observed using neural networks. Neural networks can detect abnormalities or varia-
tions from typical behaviour that can suggest the presence of malware by recording
and examining these behavioural occurrences.
Anomaly Detection: Neural networks are good at spotting patterns and can spot
changes in how a system behaves typically. In the training phase, neural networks
build a baseline of typical behaviour to identify unexpected or suspicious behaviours
that deviate from the expected patterns. This helps them to identify malware that is
zero-day-aware or malicious variants that display novel behaviours.
238 A. Alchi et al.

Evasion Methods: Advanced malware frequently uses sophisticated evasion

methods to prevent detection. These evasive behaviours, such as code obfuscation,
anti-analysis tactics, or attempts to conceal or modify their presence, can be iden-
tified by neural networks with practice. Furthermore, neural networks can improve
the detection of such evasive malware strains by examining the behavioural patterns
linked to evasion.
Dynamic Analysis: Based on continuing behavioural analysis, neural networks
can process real-time data and make judgements. They can respond quickly to devel-
oping malware threats thanks to their ability to adapt to changing circumstances.
Neural networks can instantly detect and respond to dangerous behaviours in real-
time by continuously observing and analysing system behaviour, offering a proactive
defence against malware.
Contextual Understanding: Neural networks can record and examine the context
of system behaviour. This comprises elements like the chronology of events, timing,
interdependencies between activities, and connections between various parts. As a
result, neural networks can gain a more thorough grasp of the behaviour displayed by
malware and improve the accuracy of their detections by considering these contextual
elements.
4. Adaptability and Generalisation: Neural networks can apply their under-
standing to other virus families. They can accurately detect new and undiscovered
malware types after being trained on various datasets. This adaptability is essen-
tial in today’s quickly changing threat environment, as attackers continuously
create new malware strains to circumvent established defences. Neural networks
provide a proactive defence system that can adapt to new threats.
In malware detection, neural networks’ primary advantages include adaptability
and generalisation. These characteristics allow neural networks to efficiently iden-
tify and categorise known and unidentified malware variants and adapt to changing
cybersecurity threats.
Neural networks can modify and update their knowledge in response to new
input. As a result, neural networks can be retrained on updated datasets to include
understanding these new dangers as new malware types are found. This adaptability
guarantees that the network stays current with the most recent malware trends and
may precisely identify emerging threats that may have eluded conventional signature-
based detection techniques.
1. 2. Generalising Across Malware Families: Neural networks gain knowledge
from various datasets covering various malware families and variants. By training
on multiple samples, neural networks can identify underlying patterns and traits
shared by several malware strains. By identifying similarities and shared features
between malware families, they can use this generalisation to find new malware
variants with similar structures or behaviours.
2. Determining Unknown Malware: Neural networks can identify zero-day
malware, software that has yet to be discovered. This is accomplished through
their ability to pick up on the traits and actions displayed by well-known malware
Impact of Neural Network on Malware Detection 239

strains. In addition, neural networks may detect departures from the norm even
without detailed knowledge about a specific malware strain by comprehending
the underlying patterns and traits that separate malicious software from benign
software.
3. Transfer Learning: To adapt their expertise from one malware detection task to
another, neural networks can use transfer learning. Transfer learning uses skills
developed during training on one dataset to enhance performance on another
related dataset. This indicates that, in the context of malware detection, a neural
network trained on a substantial and varied malware dataset can transfer its
learned features, representations, and behaviours to enhance detection accuracy
on different malware detection problems.
4. Variability Resistance: Neural networks are renowned for their capacity to deal
with erratic and noisy input. This robustness is helpful for malware identification
when dealing with polymorphic malware that can alter its structure or behaviour
to avoid detection. In addition, neural networks can capture the constant under-
lying patterns and characteristics across many malware strain changes, allowing
them to correctly categorise such malware even when it manifests in various
ways.
Neural networks offer a proactive and adaptable malware detection method by
fusing flexibility with generalisation. They can manage variances in malware strains,
generalise their knowledge across many malware families, detect unknown malware,
and use transfer learning. They can also adapt to new threats. Because of these
features, neural networks are valuable weapons in the continuous conflict against
sophisticated and constantly changing malware threats.
1. Limitations and Challenges: While neural networks have many benefits,
detecting malware presents particular difficulties. The requirement for sizable and
representative training datasets is a significant obstacle. Gathering and tagging
various malware samples can take time and resources. Additionally, when harm-
less files are mistakenly labelled as malware or vice versa, false positives and
negatives may occur when using neural networks. The neural network models
must be continuously monitored and improved to lessen these difficulties. While
neural networks have much to offer regarding malware detection, several restric-
tions and challenges must be considered. The following are some significant
drawbacks and problems related to the application of neural networks in this
field:
1. Large and Representative Training Datasets: For neural networks to perform
at their best, large volumes of training data are necessary. Gathering and cate-
gorising a broad and representative array of malware samples might take time
and effort. The quantity and calibre of training data can significantly impact how
well neural networks detect malware.
2. False Positives and False Negatives: Although resistant, neural networks are
not impervious to them. False positives happen when suitable files or actions are
mistakenly categorised as malicious, resulting in pointless alarms and potential
disruptions. False negatives, on the other hand, occur when malware slips through
240 A. Alchi et al.

the cracks and lets dangerous activity continue unchecked. Therefore, the neural
network models must be adjusted and fine-tuned to reduce false alarms and
missed detections.
3. Interpretability and Explainability: Because neural networks are frequently
viewed as “black boxes,” it might be challenging to understand how they make
decisions. This lack of interpretability can make determining if a file or action
is malicious or benign complex. The capacity to defend a choice is crucial in
several circumstances, including legal inquiries and regulatory compliance.
4. Avoiding Adversarial Attacks: Neural networks can be subject to adversarial
attacks, in which evildoers purposefully alter or create inputs to trick the web
and avoid detection. For example, malware samples can be changed to evade
detection or adversarial instances that look harmless but are mistakenly identified
as malicious can be made. As a result, neural network-based malware detection
faces continual challenges in creating effective defences against such assaults.
5. Resource Needs: Neural network training and deployment for malware detection
can be computationally demanding, needing significant computational time and
resources. Due to their intricacy, deep learning models and other neural network
topologies might require a lot of processing power and memory. For deployment
to be realistic, it is essential to implement effective and scalable ways to handle
these resource requirements.
6. Dynamic and Evolving Danger Environment: As new malware variants and
attack strategies often appear, the dangerous climate for malware is constantly
changing. To respond to evolving threats effectively, neural networks must be
regularly updated and trained on new datasets. It’s a constant challenge to stay on
top of malware developments and ensure that neural network models accurately
represent current malicious behaviours.
Cyber security experts, data scientists, and machine learning specialists must
continue researching, developing, and collaborating to address these constraints
and obstacles. However, these challenges can be overcome, and malware detection
capabilities can be strengthened by ongoing enhancement and refinement of neural
network-based malware detection systems.

References

1. Narayanan BN, Djaneye-Boundjou O, Kebede TM (2016) Performance analysis of machine

learning and pattern recognition algorithms for Malware classification
2. Yang C, Xu J, Liang S, Wu Y, Wen Y, Zhang B, Meng D (2021) DeepMal: maliciousness-
preserving adversarial instruction learning against static malware detection
3. Catak FO, Ahmed J, Sahinbas K, Khand ZH (2021) Data augmentation-based malware
detection using convolutional neural networks
4. Goodfellow IJ Pouget-Abadie J, Mirza M, Xu B, Warde-Farley D, Ozair S, Courville A, Bengio
Y (2014) Generative adversarial networks
5. Singh J, Singh J (2021) A survey on machine learning-based malware detection in executable
files. J Syst Arch
Impact of Neural Network on Malware Detection 241

6. Gotthans J, Gotthans T, Novak D (n.d.) Improving TDOA radar performance in jammed areas
through neural network-based signal processing
7. Rigaki M, Garcia S (2018) Bringing a GAN to a knife-fight: adapting malware communication
to avoid detection
8. Recent advances in convolutional neural networks (2018)
9. Wei R, Cai L, Zhao L, Yu A, Meng D (2021) DeepHunter: a graph neural network based
approach for robust cyber threat hunting
10. Rita (2022) Neural networks for pattern recognition
11. Anandhi V, Vinod P, Menon VG, Aditya KM (2022) Performance evaluation of deep neural
network on malware detection: visual feature approach
12. Wise A, Kruglyak KM (2019) Variant classifier based on deep neural networks
13. Lee Y-S, Lee J-U, Soh W-Y (2018) The trend of malware detection using deep learning
14. Bazrafshan Z, Hashemi H, Fard SMH, Hamzeh A (2013) A survey on heuristic malware
detection techniques
Evaluating Different Malware Detection
Neural Network Architectures

Harinadh Varikuti and Valli Kumari Vatsavayi

Abstract In today’s digital world, cutting-edge technologies in machine learning,

cybersecurity, data science, and blockchain are emerging rapidly. Every day billions
of devices are connected together through internet for data exchange. Due to the
existence of some vulnerabilities present in the device, attackers perform malicious
activities through such devices. Detecting the malwares using anti-malware engines
and signature-based methods is becoming more and more complex day by day.
Various techniques are used by the attackers in the generation of malwares such
as encryption, obfuscation, etc., to escape the detection. Few years ago, in tradi-
tional machine learning-based malware detection algorithms, feature engineering
was the most complex activity to perform. Inefficient extraction of features gives
poor performance in the detection of malware samples. Neural network techniques
in deep learning give better results in the identification of malware data. This chapter
presents a survey on the performance of various deep learning-based neural networks
for malware detection and classification. It is also noted that these models require a lot
of computational time during training. Transfer learning is a technique used to reduce
the training time by using pre-trained models such as VGG-16, ResNet-50, Incep-
tion V3, etc. This chapter explains how the evolution of neural networks changed the
malware analysis giving exceptional results during detection of malware samples.
It classifies, compares, and evaluates several deep learning-based malware detec-
tion approaches published in literature elaborating their advantages, disadvantages,
applicability, assumptions, and scope.

Keywords Neural networks · Malware · Detection · Classification · Machine

learning · Deep learning · Transfer learning

H. Varikuti (B) · V. K. Vatsavayi

Andhra University, Visakhapatnam, India
e-mail: [email protected]
V. K. Vatsavayi
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 243
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_11
244 H. Varikuti and V. K. Vatsavayi

1 Introduction

The cybersecurity industry is growing rapidly with the growth in the number of users
and connectivity. A large number of devices are interconnected for the exchange of
data, operations, etc. Insecure connection of devices causes lot of vulnerabilities and
security issues in the network. Meanwhile, attackers have been trying to gain access
to targeted systems by sending malicious files. With the new technologies at hand,
attackers attempt new obfuscation techniques to create a variety of malware files
that escape detection using traditional malware detection systems. Attackers perform
cybercrimes which include data theft, data destruction, money theft, denial of service,
hacking the operations of the system, etc. According to cybercrime reports prediction
[1], 8 trillion USD will be lost annually due to cybercrime damage in 2023. The total
loss is more than most of the developed countries’ economies. The growth in cyber
threat damage demands focus on research towards overcoming security vulnerabil-
ities and building resilient secured systems. A telecommunication company named
T-Mobile suffered data breach [2] recently in January 2023. In this attack, over 37
million customers’ data was stolen by the attacker through Application Program-
ming Interface (API). Due to some vulnerability identified in the API, the attacker
retrieved customer personal information through it. These types of data breaches
affect the customer’s trust in the organization. Ransomware attacks are the popular
malware attacks that happen more frequently throughout the world. Attacker [3]
takes advantage of vulnerabilities present in the security layer of organizations and
encrypts all the files related to it. Organizations include private businesses, health-
care, and government institutions. Recent ransomware attackers used triple extortion
techniques [4] in the attacks to get more profits.
According to the malware analysis market [5], the predicted global malware anal-
ysis market size is likely to reach USD 24.1 billion by 2026 with a Compound Annual
Growth Rate (CAGR) of 28.5% between 2019 and 2026.

1.1 Malware Analysis

Detecting the malware is done by analyzing the malware samples. Malware analysis
depends on the functionality and behavior of the code present in it. Daily a large
number of new malwares are generated by the attackers using various techniques such
as obfuscation, compression, etc. to escape detection. Malware analysis is performed
in three ways (i) static analysis, (ii) dynamic analysis, and (iii) hybrid analysis.
(i) Static analysis is about studying the code present in malware file, for example,
in the Windows executable file without its execution. It examines the code structure
related to functions, libraries, and operations used by the executable file. This type of
analysis initially converts executable files into low-level assembly language code files
to get information about what exactly the code performs. Sometimes, extraction of
strings from the executables gives hints about the file functionality. Strings include
Evaluating Different Malware Detection Neural Network Architectures 245

DLL files, domain names, attacker file paths, IP addresses, etc. Tools used in the
process of static analysis are Dependency Walker, PEView, IDA freeware, IDA pro,
and Ghidra, etc. The dependency walker scans the entire executable file and represents
a hierarchical tree with all dependent modules. PEView is used to view portable
executable files with header information of the file. IDA freeware is an open tool
to perform analysis on malware binaries with limited functionality, whereas IDA
Pro is an enterprise tool, which is commonly used by reverse engineers and security
analysts. Attackers use obfuscation techniques in the malware generation to evade
detection. In the obfuscation method, encryption and compression have been used
to generate the code which is difficult to get the information using traditional tools
mentioned. Static analysis depends mostly on the domain knowledge of the analyst
and also misinterpretation of data may lead to errors in detecting the malware.
While considering the limitations of static analysis, dynamic analysis gives reli-
able results in the detection of malicious files. Dynamic analysis is the process of
performing the execution of malware executable files in a safe isolated environment.
It tells the way in which the malicious code interacts with the system. This type
of analysis has been performed by isolated physical systems or virtual machines.
Besides those two environments, dynamic analysis has been performed in a sandbox
environment to detect the malware. Tools used for analyzing malware dynamically are
Process Monitor, Process Explorer, Regshot, NetCat, Wireshark, etc. The dynamic
analysis gives better results in identifying the zero-day malware when compared with
the static analysis approach. The major limitation of performing the dynamic anal-
ysis is providing a safe environment, because erroneous setting of the environment
will infect the system and damage it badly.
Hybrid analysis includes the combination of both static and dynamic analysis to
perform malware detection. Information obtained from both approaches has been
efficiently used to perform the malware analysis.

1.2 Machine Learning in Malware Detection

A few years ago, anti-malware engines have been used to perform malware detection.
These engines used signature and heuristic-based techniques to detect the malware. In
the signature-based method, known malwares are identified by matching the signa-
tures present in the anti-malware database. Unknown malware is not detected by
the engines which limits its capability. The heuristic-based method is the process
of investigating the source code for apprehensive action. Polymorphic malware and
zero-day malware have been detected using the heuristic methods. But there is a
problem of getting false positives and false negatives with this method. False posi-
tive, which identifies benign files as malware and doesn’t affect costly, whereas false
negatives identify malware files as benign which destructs badly to the system.
Nowadays with the increase of digital data, it is difficult for humans to continu-
ously monitor and analyze to take decisions. To overcome this issue, machine learning
has been used to train the model on viable data and apply the model to similar data
246 H. Varikuti and V. K. Vatsavayi

items for taking decisions. Attackers also apply various technologies in generating
malicious code to perform illegal operations throughout the world. Digitalization
increased the opportunities for attackers with more volume availability. On the other
side, the difficulty increases in detecting malicious structures manually for huge
volumes of data. Here machine learning techniques are used to perform malware
analysis and detect malicious files. Along with the detection of malware files, classi-
fying the malware class is also important to know the functionality of the malware file.
Machine learning approaches initially perform feature engineering steps to extract
the features. Features are extracted based on the user domain knowledge. Here the
features identified are used in building the model using trained dataset to perform the
required operation. Based on the trained model, test data is used to find the perfor-
mance of the model. Various malware detection models built using machine learning
are discussed in the related work section.

1.3 Artificial Neural Networks

Traditional machine learning methods believe in user-extracted features and

their analysis. But these hand-crafted feature models sometimes lead to inefficient
models due to inappropriate feature selection. To overcome the problem that occurred
by the manual feature selection, deep learning models perform automatic deep feature
extraction using neural networks. Here artificial neurons are created and connected
to form artificial neural networks. Every neuron is associated with weight when it has
been given as input to the next layer. An artificial neural network includes an input
layer that takes source data, one or more hidden layers which analyze and perform
computation on the data to extract features, and an output layer that predicts the output
of the source data provided. Applications, where neural networks play an impor-
tant role, include chatbots, pattern recognition, machine vision, speech recognition,
medical, telecommunication, transport, and time series prediction, etc. unlike hand-
crafted features used in machine learning models, neural networks extract mean-
ingful patterns from the data to give better performance when compared to traditional
machine learning models. Neural networks have been performing exceptionally well
in the field of malware detection and classification. Detection of malware samples
relies on the behavior and structure of the malware. To detect the malware samples,
patterns are identified as feature vectors by the neural network.

2 Related Work

In this section, various research works published in literature on malware detec-

tion and classification are discussed. Using machine learning and neural network
models, efficient identification of malware is possible. Traditional machine learning
approaches extract the hand-crafted features in the feature generation process. With
Evaluating Different Malware Detection Neural Network Architectures 247

limited domain expert knowledge, feature extraction is difficult to perform. By

observing the immense growth in the machine learning field, malware analysis
has advanced toward deep learning. Using deep learning techniques, an end-to-end
model takes input of malware sample with little preprocessing and directly detects
its related malware family as output. Malware detection also is efficiently performed
by the neural network models by differentiating the behavior patterns of malware
and benign files. To deal with malware bytes files, various approaches [10–16] use
grayscale images for malware classification. Narayanan et al. [11] used principal
component analysis (PCA) for getting reduced features from the images, then various
machine learning algorithms such as K-nearest neighbor, support vector machines,
and artificial neural network (ANN) were applied. K-nearest neighbor gives the best
performance with the PCA features. In the paper [12], the authors proposed deep
learning model which uses autoencoders [16] in the multi-layer architecture.
As per the studies, Convolutional Neural Network (CNN) models performed well
on the image dataset. Gibert et al. [14] presented a CNN architecture for the clas-
sification of malware images, which performs better with minimal prediction time.
Gibert et al. [13] represented the malware executable as an entropy stream [17], which
gives randomness of the system. Haar wavelet transform was used on the entropy
times series to reduce the noise and size, and CNN model was applied for malware
classification. The author concluded that wavelet features performed strongly against
obfuscation techniques. Singh et al. [15] performed malware analysis on Android
malware images using feature fusion methods. Malware APK files have been used to
generate grayscale images. Hand-crafted feature extraction methods such as Gray-
Level Co-occurrence Matrix (GLCM), Global Image descriptors (GIST), and Local
Binary Pattern (LBP) uproot features. Traditional machine learning approaches such
as support vector machine, K-nearest neighbor, and random forest have been applied
on the above hand-crafted features. Feature fusion applied with CNN and hand-
crafted features to perform malware classification. Feature fusion method gives better
results when compared with traditional methods. Xiao et al. [16] used structural
entropy graphs for extracting features using CNN model and then support vector
machine method was applied to perform malware classification. Yamashita et al.
[10] provided an overview of CNN model, also gave how to overcome overfitting
using various techniques such as data augmentation, batch normalization, dropout,
etc. Khan et al. [18] classified the malware using deep neural networks using binary
files, later a CNN model was built using grayscale images as input. CNN model
gives better results emphasizing the fact that convolution models are more efficient
in classifying the image data. M. Alam et al. [19] used standard neural network models
such as VGG-16, ResNet-50, and Inception v3 and proposed convolutional neural
network for classifying the malware data which gives a significant performance.
Depuru et al. [20] performed a comparative analysis of malware classification using
neural network models and machine learning models. Shinde et al. [21] constructed
four neural networks with malware data image sizes of 224 × 224, 240 × 240, 260
× 260, and 300 × 300, respectively. Using a large dataset, the neural network which
uses high image pixel size gives better results when compared with low image pixel
size.
248 H. Varikuti and V. K. Vatsavayi

Yamashita et al. [10] used neural network models to perform malware detection on
the EMBER dataset and observed better performance. Akthar et al. [22] performed
malware detection using machine learning models and CNN model. Among the
machine learning models used, decision tree gives good results along with neural
network system. Pavithra et al. [23] applied malware analysis using various machine
learning classifiers in detecting the adware malware files. In their work, random
forest, SVM, and Naïve Bayes algorithms have been used, in which random forest
gives better results with lesser time complexity. Aslan et al. [24] proposed a hybrid
model which uses pre-trained models such as AlexNet and ResNet-50 for feature
engineering and combined features have been applied to fully connected layers and
final output layer. Xing et al. [25] used autoencoders in the detection of Android
malwares. Kumari VV et al. [26] proposed an efficient model for malware detection
and also gives impact of neural networks in malware detection.
From the above-discussed papers, it is observed that mostly deep learning models
give good results when compared with the machine learning models in classifying
and detecting the malware samples for both balanced and unbalanced data. This
paper provides implementation of machine learning and neural network models for
the task of malware detection and classification and also gives better results when
compared to the state-of-art models discussed above.

3 System Overview

3.1 Datasets

In this work, three benchmark datasets are used. They are (1) the MALIMG dataset for
malware classification, (2) the Microsoft malware dataset for malware classification,
and (3) the BODMAS dataset for malware detection and classification.
Malimg Dataset: The Malimg [7] dataset contains 9339 malware samples of
grayscale images which belong to 25 malware families. The malware family distri-
bution of the Malimg dataset is shown in Table 1. Allaple.A has a greater number of
samples with a count of 2949 and Skintrim.N has the least number of samples with
a count of 80.
Microsoft Malware Dataset: Microsoft [8] provided the malware dataset with
10,868 samples belonging to 9 malware families. It provides both bytes file and
assembly code file for every malware sample. The data has been launched for the
challenge hosted on Kaggle. The malware family distribution of Microsoft malware
data has been shown in Table 2. Kelihos_ver3 has a greater number of samples with
a count of 2942 and Simda has the least number of samples with a count of 42.
It is considered as the benchmark dataset to evaluate the performance of machine
learning algorithms in malware classification.
BODMAS Dataset: BODMAS [6] stands for Blue Hexagon Open Dataset for
Malware Analysis. The researchers performed joint work with Blue Hexagon to
Evaluating Different Malware Detection Neural Network Architectures 249

Table 1 Class distribution of

Malware family Number of samples
the malimg dataset
Allaple.A 2949
Allaple.L 1591
Yuner.A 800
Instantaccess 431
VB.AT 408
Fakerean 381
Lolyda.AA1 213
C2LOP.gen!g 200
Alueron.gen!J 198
Lolyda.AA2 184
Dialplatform.B 177
Dontovo.A 162
Lolyda.AT 159
Rbot!gen 158
C2LOP.P 146
Obfuscator.AD 142
Malex.gen!J 136
Swizzor.gen!I 132
Swizzor.gen!E 128
Lolyda.AA3 123
Adialer.C 122
Agent.FYI 116
Autorun.K 106
Wintrim.BX 97
Skintrim.N 80

Table 2 Class distribution of

Malware family Number of samples
the microsoft malware dataset
Kelihos_ver3 2942
Lollipop 2478
Ramnit 1541
Obfuscator.ACY 1228
Gatak 1013
Vundo 475
Tracur 751
Kelihos_ver1 398
Simda 42
250 H. Varikuti and V. K. Vatsavayi

Table 3 Malware category

Malware category Number of samples
distribution of BODMAS
dataset Trojan worm 29,972
Worm 16,697
Backdoor 7331
Downloader 1031
Ransomware 821
Dropper 715
Information stealer 448
Virus 192
Pua 29
Cryptominer 20
P2p-worm 16
Exploit 12
Trojan-game thief 6
Rootkit 3

produce timestamped malware data and related malware family information. The
dataset includes 57,293 malware samples and 77,142 benign samples with a total of
134,435 samples. The total malware samples constitute 14 malware categories such
as virus, worm, Trojan, downloader, dropper, ransomware, backdoor, information
stealer, rootkit, p2p-worm, cryptominer, trojan-game thief, pua, and exploit. The
malware samples cover 581 malware families. Here the data samples include the
SHA-256 hash of the malware file and pre-extracted features. The total malware
samples category distribution is shown in Table 3.

3.2 Malware Representation and Features

Malware files can be represented in different ways such as hex view (bytes files) and
assembly view (assembly language files). In the bytes file, all the machine code which
represents opcodes, functions, jump statements, control statements, texts, images,
etc., are represented in hexadecimal format shown in Fig. 1a. The left-most value
(e.g., 00401000) represents the starting address of the machine code in the memory
and consecutive 16-byte values represent opcode or data. Malware samples are also
represented as grayscale images shown in Fig. .b from the Malimg dataset. In the
assembly language files, all the malware sample data has been represented in the form
of mnemonic codes to represent machine codes, function calls, register information,
etc., as shown in Fig. 2.
Evaluating Different Malware Detection Neural Network Architectures 251

Fig. 1 a Hexadecimal representation of sample malware file from the Microsoft malware dataset,
b Grayscale image of sample malware file from the Malimg dataset

Fig. 2 Assembly language code of sample malware file

To perform the malware classification or detection, features have been extracted

from the malware binary files. Feature extraction plays a significant role in the perfor-
mance of the model. Various techniques like N-Gram analysis and image-based
analysis have been used to extract the features.
N-Gram Analysis
N-gram analysis is the standard technique to extract the features for various problems
along with malware detection and classification. N-grams are the sequence of n data
items considered as a feature and count the frequency of the sequence in the entire
document. In malware bytes file, bytes N-grams are considered. Opcode, registers,
keywords, prefixes n-grams are considered from the assembly language source files.
252 H. Varikuti and V. K. Vatsavayi

1-g, 2-g, 3-g, etc., are examples of n-grams. Features extracted using 1-g are discussed
below.
Byte_histogram. It counts the frequency of each byte value in the file. Byte values
range from 00 to ff which is 256 values. Same family malware files have nearly
matched the frequency of byte values.
Prefixes. These features are extracted from assembly files. In assembly language
files, various sections are present such as data section, bss section, text section, etc.
Features extracted include HEADER, text, pav, idata, data, bss, rdata, edata, rsrc, tls,
reloc, BSS, and CODE.
Opcodes. Assembly code contains various operation codes which give behavioral
aspect of the file. Opcode features considered are rol, jnb, jz, rtn, lea, movzx, push,
pop, dec, add, imul, xchg, or, xor, retn, nop, sub, inc, shr, cmp, call, shl, ror, jmp,
mov, and retf.
Keywords. It includes std, dword, and dll.
Registers. It stores data values used for processing in assembly files. Registers used
in assembly code are eax, edx, esi, eip, ecx, ebx, ebp, esp, and edi.
The BODMAS dataset used for malware detection contains 2351 pre-extracted
1-g features of malware and benign files shown in Table 4.
Image-Based Analysis
In image-based analysis, malware binaries are converted into image files such as
grayscale images of size 128 × 128, 256 × 256, etc. Every hexadecimal view of byte
values is represented as one pixel in the image. By using neural networks, features
are extracted automatically from malware images. To extract these types of features,
no additional knowledge is required in contrast to the handcrafted features which
use n-gram analysis. Malware files belonging same family have similar grayscale
image structures. Bytes data has been converted into 2D array, thereby array values
are represented as pixels. Neural networks take a large computation time to extract

Table 4 Feature groups

Feature groups No of features
present in the BODMAS
dataset Header information 62
General file information 10
Byte histogram 256
Byte entropy histogram 256
Imported functions 1280
Exported functions 128
String information 104
Section information 255
Total 2351
Evaluating Different Malware Detection Neural Network Architectures 253

the features when compared with handcrafted features. We considered two datasets
named Malimg and Microsoft malware datasets to perform image-based analysis for
feature extraction.

4 Architectures

4.1 Machine Learning Models

Machine learning models learn knowledge from either labeled data or unlabeled data
and perform predictions or decisions. In this work, multiclass malware classification
and malware detection have been performed. Both these problems come under the
classification method of supervised learning.
In this chapter, machine learning models such as KNN, Random Forest, Support
Vector Machine, Decision Tree, and Extra Tree classifiers are used in the detection
and classification of malware data.
K-Nearest Neighbor
K-nearest neighbor (KNN) [9] is one of the most commonly used classification
algorithms in detecting malware samples. KNN is a supervised machine learning
algorithm used for both classification and regression problems. It is an instance-
based learning technique which doesn’t construct the generalized function. New
instances are classified by mapping with already stored trained data points using
distance measures.
Random Forest
Random forest is a supervised machine learning model which uses an ensemble tech-
nique that assumes predictions from multiple decision trees to perform classification.
It is the most used algorithm due to its effective handling of high-dimensional and
imbalanced datasets. Hyperparameter tuning has been performed in the model to get
the optimal results. Hyperparameters include a number of decision trees, depth of
the tree, number of samples, minimum sample split, number of features, etc. Grid
search and random search have been used to identify the best hyperparameter values.
Support Vector Machines (SVMs)
Support vector machines are classification models which use the hyperplane to sepa-
rate the data points of different classes in the high-dimensional space. The hyper-
plane represents the boundary line which differentiates the data points in feature
space. Support vectors are used to identify the margin, which is the distance from
the hyperplane to the closest data points. Kernel function has been used to convert
the data points into feature vectors. Some of the kernel functions used are linear,
radial bias function, sigmoid, and polynomial. Based on the characteristics of the
254 H. Varikuti and V. K. Vatsavayi

data points, one of the kernel functions is used. SVM performs classification on
linearly separable as well as non-linear separable data points.
Decision Trees
Decision trees are tree-like structures which are used to perform classification. Each
internal node defines the selected feature which either splits into different branches
or it represents the target class based on the outcome of the feature for all the data
points. Each leaf node represents the target class label. Features are selected based
on the entropy and information gain parameter values. Tree pruning technique has
been used if an overfitting situation occurs in the decision tree. Ensemble methods
such as AdaBoost and gradient boosting have been used to increase the performance
of the decision tree systems.
Extra Trees Classifier
Extra trees classifier is an ensemble machine learning model which uses the decision
trees. It considers entire data points to construct the trees which increases the variance.
Extra tree classifier trains multiple number of decision trees and aggregates all the
tree results to predict the output. It randomly selects the feature for splitting the node
into child nodes. It is robust against overfitting and gives a more generalized model.
It is similar to random forest algorithm but computationally faster than it.

4.2 Deep Learning Models

Deep learning models extract high-level abstract features and give improved perfor-
mance when compared with traditional machine learning models. It automatically
extracts the standard features from data samples. Unlike deep learning models,
machine learning models need more domain knowledge and computational time for
feature extraction. As per the study from related work, convolutional neural networks
perform better with image data. Deep learning models use large datasets and neural
networks for training. Classification has been performed by directly training the
model using images, text, etc.
Grayscale Image-Based Convolutional Neural Network for Malware Classifi-
cation
Convolutional neural network (CNN) is a widely used architecture to perform image
classification and detection. In this work, CNN architecture is implemented as shown
in Fig. 3. to extract features from grayscale images for the malware classification.
Initially, grayscale images of all malware files have been resized to 256 × 256. These
resized images are given to convolution layer. The first two blocks of the architecture
are the convolution layer followed by the max-pooling layer. The third block contains
the dropout layer immediately after the convolution layer which cuts some features
to reduce the computational time, and then the max-pooling layer. Now the low
Evaluating Different Malware Detection Neural Network Architectures 255

dimensional space has been flattened and given as input to a fully connected layer,
which is followed by the output layer.
At output layer, malware image datasets such as Malimg and Microsoft malware
gave 25 and 9 units, respectively, as per the count of malware families present.
Malware samples have been detected based on the extracted features. The above
CNN model is a shallow model with three convolution layers, one fully connected
layer, and one output layer.

Fig. 3 Convolutional neural

network architecture for
malware classification using
grayscale images
256 H. Varikuti and V. K. Vatsavayi

4.3 Transfer Learning

Transfer learning is the one type of machine learning method which uses the knowl-
edge of pre-trained model and applies on new tasks. Construction of deep learning
models from scratch takes more amount of time in training the data and also adjusting
the weights to extract the features. Pre-trained models use large datasets for training
the model so that smaller datasets also give good performance with it. Meanwhile,
usage of pre-trained model knowledge reduces the computational time and resources.
A large number of pre-trained transfer learning models, which are trained on large-
scale datasets, have been available. The pre-trained models such as VGG-16, Resnet-
50, and InceptionV3 for initial layer training as shown in Fig. 4. Firstly, malware
images are provided to pre-trained models. These pre-trained models extracted
the features based on the kernel filters provided. Output of pre-trained models is
connected to one fully connected layer (dense layer) with 1024 units and one output
layer (softmax layer) which classifies the data. Inner layer training depends on the
pre-trained model, VGG16 has 13 convolution layers and 3 fully connected layers.
ResNet-50 contains 50 layers which include convolution layer, pooling layer, normal-
ization layer, etc., InceptionV3 contains factorized convolutions, regularization, and
parallelized computations.

5 Evaluation

This section provides the experimental evaluation of various machine learning

models and deep learning models discussed in Sect. 4. We considered bench-
mark malware datasets such as Malimg, Microsoft malware data, and BODMAS
as discussed in Sect. 3 to perform malware classification and detection. To build
the models, all the datasets are divided into train set and test set with 75 and 25%,
respectively. Accuracy and logloss are the two metrics used for analyzing the model
performance.

number of correct predictions

Accuracy =
total number of predictions

Accuracy is the performance metric used to get the correctly predicted samples
among all the samples. Accuracy is not enough to assess the performance of the
models, because of the class imbalance that exists in the datasets. Logloss gives the
closeness of the predicted state and target state.

1 ∑∑
N M
Logloss = − yi, j log( pi, j )
N i=1 j=1
Evaluating Different Malware Detection Neural Network Architectures 257

Fig. 4 Transfer learning

model to perform malware
classification

where N represents the number of samples, M represents the number of class labels,
yi,j is 1 if the sample i belongs to class j and 0 otherwise, and pi,j is the prediction
probability that observation i is in class j.

5.1 Feature Performance Analysis

In this work, 256 features from bytes files and 48 features from assembly code files
of 10,868 malware samples of the Microsoft malware dataset are extracted. From
the BODMAS dataset, 2351 pre-extracted features have been provided as shown in
258 H. Varikuti and V. K. Vatsavayi

Table 4. Machine learning models (ML models) such as KNN, Random Forest, SVM,
Decision Trees, and Extra trees classifiers have been applied to the above datasets
to perform malware classification and detection. Table 5 shows the performance of
malware classification on the Microsoft malware dataset. Byte features and Assembly
features (ASM) individually are applied on the machine learning models, among them
ASM features attain better performance when compared with byte features. Feature
fusion has been performed by combining byte features and ASM features and applied
to the machine learning models.
Table 6 shows the performance of malware detection on the BODMAS dataset.
Here each feature category labeled as G1, G2, … given below.
• G1 represents the byte histogram features.
• G2 represents the byte entropy histogram features.
• G3 represents the imported functions and library features.
• G4 represents the string information features.
• G5 represents the general file information features.
• G6 represents the header information features.
• G7 represents the combined features of all the groups specified in Table 4.

Tree-based classifiers such as Random Forest, Decision tree, and Extra Trees
classifiers give good performance with accuracy above 98% in malware detection
as well as classification, when compared with KNN and SVM (linear). SVM gives

Table 5 Malware classification performance based on feature category of the Microsoft dataset
Machine learning models Accuracy
Byte features (256) ASM features (48) Combined Byte + ASM
features (304)
KNN 92.5 99.08 97.6
Random Forest (RF) 98.11 99.63 99.07
SVM (Linear) 79.6 88.5 89.73
Decision Tree 95.6 98.75 98.71
Extra Trees Classifier 98.7 99.3 99.1

Table 6 Malware detection performance based on feature category of the BODMAS dataset
Machine learning models Accuracy
G1 G2 G3 G4 G5 G6 G7
KNN 96.5 98.03 97.7 86.4 97.4 98.1 98.1
Random Forest (RF) 98.2 98.5 98 98.4 99.1 99 99.5
SVM (Linear) 77.2 83 97 86 75.7 87.9 70.3
Decision Tree 95.5 96.6 97 96.36 98.6 98.8 98.9
Extra Trees Classifier 98.3 98.6 97.9 98.4 99.09 98.9 99.5
Evaluating Different Malware Detection Neural Network Architectures 259

Fig. 5 a Training and validation accuracy of malimg data and b training and validation logloss of
malimg dataset

better performance with imported library functions when compared with other feature
groups.

5.2 Image Based Model Analysis

Two datasets Malimg and Microsoft malware (discussed in Sect. 3) are taken, with
gray scale images to perform malware classification. In Sect. 4.2, a shallow CNN
model has been proposed to classify the malware samples. The CNN model gives
us the training accuracy of 1.00 and validation accuracy of 0.988 on malimg dataset
as shown in Fig. 5a. The logloss values of the model on malimg data is shown in
Fig. 5b.

5.3 Transfer Learning Model Analysis

In the malware classification, pre-trained models such as VGG16, ResNet-50 and

InceptionV3 are used. Pre-trained models reduce the training time, as they are already
trained on large Image dataset. Features obtained from pre-trained model are given
to two customized dense layers to classify the malware data. In Sect. 4.3, transfer
learning model has been proposed and applied on the malimg and Microsoft data,
which gives accuracy as shown in Figs. 6 and 7. As per the observation, from the bar
plotting, among the three pre-trained models, VGG16 performs betters and gives
260 H. Varikuti and V. K. Vatsavayi

Fig. 6 Accuracy scores of models using pre-trained architectures such as VGG16, ResNet-50, and
InceptionV3 in the Malimg data

good training accuracy and validation accuracy for both malimg and Microsoft
malware datasets.

5.4 Comparison of Architectures

This section presents a performance analysis of all the models discussed in the
previous sections of machine learning models and neural network models.
Feature Engineering
Machine learning models are highly dependent on the hand-crafted features. Feature
engineering has been performed by the domain expert with much-needed knowledge
to get the best feature set. On the other side, neural networks perform automatic
extraction of features from raw data. For malware analysis, attackers always tend to
use new techniques to perform malicious activities. So, if we apply similar hand-
crafted features to the latest malwares, it may not give good performance.
Handling Complex and Non-linear Data
Smaller datasets with less complexity in feature extraction can be easily handled
by the machine learning models, whereas with high dimensionality datasets, non-
linearity in the data and unstructured data such as images can be handled by the neural
Evaluating Different Malware Detection Neural Network Architectures 261

Fig. 7 Accuracy scores of models using pre-trained architectures such as VGG16, ResNet-50, and
InceptionV3 in the Microsoft malware data

network models and give good performance. Complex patterns or relationships can
be easily extracted by the neural network layers.
Time Complexity and Resource Consumption
Training machine learning models takes small amount of time when compared with
deep learning models, because of a smaller number of parameters. Neural networks
extract deep features in the training stage of large datasets. Sometimes neural
networks need GPUs to run the large and complex datasets. In deep learning training
time is reduced by using pre-trained neural network models. These pre-trained models
transfer the knowledge.
Table 7 represents the comparison of theproposed neural network model and
transfer learning models which use various pre-trained models such as VGG16,
ResNet-50, and InceptionV3 with previous bytes file implementations. Except model
pre-trained with ResNet-50, all the proposed models give better performance with
99% accuracy in classifying and detecting the malware data.

5.5 Concept Drift in Malware Analysis

Concept drift is one of the significant challenges to come across when we are
performing malware detection or classification. Models trained on historic malware
data may/may not detect current or future malware variants, due to differences in the
262 H. Varikuti and V. K. Vatsavayi

Table 7 Bytes file approach comparison of various neural networks in the Microsoft malware data
Method Accuracy Macro F1-score
PCA + K-NN [11] 0.9660 0.9102
CNN IMG [14] 0.975 0.940
CNN ENTROPY [13] 0.9828 0.9636
Deep neural network using autoencoders 0.9915 –
Proposed CNN model 0.993 0.98
Neural network model with pre-trained VGG16 0.99 –
Neural network model with pre-trained ResNet-50 0.93 –
Neural network model with pre-trained InceptionV3 0.99 –

characteristics and behavior of the malware samples. Malware creators continuously

change the techniques used in new malware variant creation to escape detection.
Sometimes, adding new malware variant data to an already existing dataset may
change the overall data distribution. To overcome the concept drift challenge, contin-
uous updation of models has been taking place. Ensemble methods have been used
to combine the different time series models to get more generalized models which
reduces the concept drift issue. Sustained monitoring of updated malware signatures
helps in detecting the trends in the new malware variants.

6 Conclusion and Future Work

In this paper, malware detection and classification are performed on various datasets
using machine learning and neural network models. As per the observation, feature
fusion is effective in detecting and classifying the malware data. A novel neural
network architecture has been used to perform malware classification and get
better results than the state-of-the-art model’s performance. Obfuscated malware
files, which use encryption and code masking, are detected efficiently using neural
networks. To reduce the training time of the neural networks, a transfer learning model
is applied and gives good performance. Results show that the proposed shallow neural
network model outperforms the machine learning models and performs as better as
state-of-the-art models in the task of malware classification. Transfer learning models
also give a good way of performing malware classification and detection by reducing
model training time. A future direction of research work is to perform malware anal-
ysis on the assembly language files by getting graph features such as control flow
graphs, which gives a technical perspective of different patterns used in malware
logic and also provides the broad domain knowledge in depth. Concept drift is a
significant issue in malware detection because models trained on historical data are
less efficient when new malware variants emerge. Generally, as times change, new
types of malwares also emerge rapidly with different structures and patterns.
Evaluating Different Malware Detection Neural Network Architectures 263

References

1. https://cybersecurityventures.com/cybercrime-to-cost-the-world-8-trillion-annually-in-2023/
2. https://www.bleepingcomputer.com/news/security/t-mobile-hacked-to-steal-data-of-37-mil
lion-accounts-in-api-data-breach/
3. https://heimdalsecurity.com/blog/companies-affected-by-ransomware/
4. https://www.akamai.com/blog/security/defeating-triple-extortion-ransomware#:~:text=
This%20combination%20of%20encryption%20and,as%20an%20additional%20extortion%
20technique.
5. https://www.alliedmarketresearch.com/malware-analysis-market-A05963
6. Yang L, Ciptadi A, Laziuk I, Ahmadzadeh A, Wang G (2021) BODMAS: an open dataset for
learning based temporal analysis of PE Malware. In: 2021 IEEE security and privacy workshops
(SPW), San Francisco, CA, USA, pp 78–84. https://doi.org/10.1109/SPW53761.2021.00020
7. Nataraj L et al (2011) Malware images: visualization and automatic classification. In:
Proceedings of the 8th international symposium on visualization for cyber security
8. Ronen R, Radu M, Feuerstein C, Yom-Tov E, Ahmadi M (2018) Microsoft Malware
classification challenge
9. Choi S (2020) Combined kNN classification and hierarchical similarity hash for fast Malware
detection. Appl Sci 10:5173. https://doi.org/10.3390/app10155173
10. Yamashita R, Nishio M, Do RKG et al (2018) Convolutional neural networks: an overview
and application in radiology. Insights Imaging 9:611–629. https://doi.org/10.1007/s13244-018-
0639-9
11. Narayanan BN, Djaneye-Boundjou O, Kebede TM (2016) Performance analysis of machine
learning and pattern recognition algorithms for Malware classification. In: 2016 IEEE national
aerospace and electronics conference (NAECON) and Ohio innovation summit (OIS), pp 338–
342, July 2016
12. Kebede TM, Djaneye-Boundjou O, Narayanan BN, Ralescu A, Kapp D (2017) Classification
of Malware programs using autoencoders based deep learning architecture and its application
to the microsoft Malware Classification challenge (BIG 2015) dataset. In: 2017 IEEE national
aerospace and electronics conference (NAECON), Dayton, OH, USA, 2017, pp 70–75. https://
doi.org/10.1109/NAECON.2017.8268747
13. Gibert D, Mateu C, Planes J, Vicens R (2018) Classification of malware by using struc-
tural entropy on convolutional neural networks. In: Proceedings of the thirty-second AAAI
conference on artificial intelligence, (AAAI-18), the 30th innovative applications of artificial
intelligence (IAAI-18), and the 8th AAAI symposium on educational advances in artificial
intelligence (EAAI-18), New Orleans, Louisiana, USA, February 2–7, 2018, pp 7759–7764
14. Gibert D, Mateu C, Planes J, Vicens R (2018) Using convolutional neural networks for
classification of malware represented as images. J Comput Virol Hacking Tech, August 2018
15. Singh J, Thakur D, Gera T, Shah B, Abuhmed T, Ali F (2021) Classification and analysis of
android malware images using feature fusion technique. IEEE Access 9:90102–90117. https://
doi.org/10.1109/ACCESS.2021.3090998
16. Xiao G, Li J, Chen Y, Li K (2020) Malfcs: an effective malware classification framework with
automated feature extraction based on deep convolutional neural networks. J Parallel Distrib
Comput 141:49–58
17. Lyda R, Hamrock J (2007) Using entropy analysis to find encrypted and packed Malware. In:
IEEE security & privacy, vol 5, no 2, pp 40–45, March–April 2007. https://doi.org/10.1109/
MSP.2007.48
18. Khan M, Baig D, Khan US, Karim A (2020) Malware classification framework using convo-
lutional neural network. In: 2020 international conference on cyber warfare and security
(ICCWS), Islamabad, Pakistan, pp 1–7. https://doi.org/10.1109/ICCWS48432.2020.9292384
19. Alam M, Akram A, Saeed T, Arshad S (2021) DeepMalware: a deep learning based
malware images classification. In: 2021 international conference on cyber warfare and secu-
rity (ICCWS), Islamabad, Pakistan, pp 93–99. https://doi.org/10.1109/ICCWS53234.2021.970
3021
264 H. Varikuti and V. K. Vatsavayi

20. Depuru S, Santhi K, Amala K, Sakthivel M, Sivanantham S, Akshaya V (2023) Deep learning-
based malware classification methodology of comprehensive study. In: 2023 international
conference on sustainable computing and data communication systems (ICSCDS), Erode,
India, pp 322–328. https://doi.org/10.1109/ICSCDS56580.2023.10105027
21. Shinde S, Dhotarkar A, Pajankar D, Dhone K, Babar S (2023) Malware detection using effi-
cientnet. In: 2023 international conference on emerging smart computing and informatics
(ESCI), Pune, India, pp 1–6. https://doi.org/10.1109/ESCI56872.2023.10099693
22. Akhtar MS, Feng T (2022) Malware analysis and detection using machine learning algorithms.
Symmetry 14(11):2304. https://doi.org/10.3390/sym14112304
23. Pavithra J, Selvakumara Samy S (2022) A comparative study on detection of Malware and
benign on the internet using machine learning classifiers. Math Prob Eng 2022, Article ID
4893390, 8 p. https://doi.org/10.1155/2022/4893390
24. Aslan Ö, Yilmaz AA (2021) A new malware classification framework based on deep learning
algorithms. IEEE Access 9:87936–87951. https://doi.org/10.1109/ACCESS.2021.3089586
25. Xing X, Jin X, Elahi H, Jiang H, Wang G (2022) A Malware detection approach using autoen-
coder in deep learning. IEEE Access 10:25696–25706. https://doi.org/10.1109/ACCESS.2022.
3155695
26. Kumari VV, Jani S (2023) An effective model for Malware detection. In: Rao BNK, Balasubra-
manian R, Wang SJ, Nayak R (eds) Intelligent computing and applications. Smart innovation,
systems and technologies, vol 315. Springer, Singapore. https://doi.org/10.1007/978-981-19-
4162-7_35
Protecting Your Assets: Effective Use
of Cybersecurity Measures in Banking
Industries

Sulagna Das and Devraj Ganguly

Abstract The Indian banking system, originating from the first Bank of Hindustan,
has evolved through nationalization, liberalization, and mergers, enhancing service-
ability, automation, and operational cost reduction. Banks are focusing on digiti-
zation, launching digital payment systems like UPI and BHIM, and promoting a
cashless society through NEFT, ECS, RTGS, and Prepaid Cards. The digital transfor-
mation aims to enhance operational efficiency, client experiences, and competitive-
ness. Key areas include digital channels, process automation, AI, machine learning,
robotics, personalized customer experiences, sustainable banking, agile workforce,
and remote employment. Technological advancements are increasing cyber-attacks
on banks and financial organizations, posing privacy and phishing issues. Cyberse-
curity technology is crucial in protecting financial information, preventing fraud, and
ensuring system reliability. Key factors include advanced threat detection, collabo-
ration, quantum-safe encryption, AI, machine learning, a cybersecurity workforce, a
zero-trust framework, and continuous monitoring for incident response and security.

Keywords Indian banking sector · Cybersecurity · Digitalization · Blockchain ·

Cybercrime

1 Evolution of the Indian Banking System

The traditional banking system in India has evolved since the 1770s when the first
Indian bank (Bank of Hindustan) was established. Over the last five centuries, the
banks have been nationalized, then liberalized, and recently have gone through bank
mergers. Before this, the banks were performing manually, the account-related infor-
mation was maintained in hard copy, through ledger books, transactions were limited,
with limited transactional accuracy, and customers had to visit branches physically
to conduct transactions. The Indian banking industry understood the importance

S. Das (B) · D. Ganguly

Department of Management Studies, JIS University, Kolkata, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 265
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_12
266 S. Das and D. Ganguly

Fig. 1 Evolution of banking. Source Author

of computerization in the 1980s to improve serviceability and the recording and

accounting of financial transactions.
Banking automation began with the computerization of key departmental func-
tions, such as Account-related banking operations [9]. Banks improved customer
service by automating branches and providing “Single Window Service” facili-
ties, which reduced transaction processing times and enabled interbank commu-
nication. This revolution was welcomed by the banking industry [2]. The multi-
channel banking strategy prioritized the creation of a centralized data infrastructure
capable of consolidating databases from various banks into a single large database.
ATMs fundamentally changed the customer–bank relationship, allowing customers
to conduct basic financial transactions without entering a bank. The implementation
of a new customer service interface “Core Banking Applications” benefited banks by
significantly lowering operational costs. (Fig. 1, gives a clear idea of the Evolution
of Banking, from branch banking to Block Chain).
The Institute for Development and Research in Banking Technology (IDRBT)
was founded in 1996 to advance banking technology in India. The Indian Financial
Network (INFINET), Structured Financial Messaging System (SFMS), and National
Financial Switch (NFS) were established by the organization. It deployed high-
end PKI-based services and solutions to protect transactions across INFINET in
India’s banking and financial sectors. PayPal, Google Pay, and Amazon Pay have
significantly impacted the banking industry by automating processes and enabling
computerized financial transfers through the widespread availability of the Internet.

2 Present Scenario of the Indian Banking System

Banks seek to give their consumers a banking experience that is rapid, accurate, and
of high quality. For Indian banks, digitization is a primary priority, and the NPCI’s
(National Payments Corporation of India) launch of United Payments Interface (UPI)
and Bharat Interface for Money (BHIM) are two major steps for innovation. Digital
payment innovation and popularity are driving up digital banking transactions. Indian
banks are increasingly adopting electronic payment systems like National Electronic
Fund Transfer (NEFT), Electronic Clearing Service (ECS), Real Time Gross Settle-
ment (RTGS), Cheque Truncation System, Mobile Banking, Debit Cards, Credit
Cards, and Prepaid Cards. Online banking has transformed the banking industry,
with low-cost methods like UPI and BHIM. NEFT is India’s most common elec-
tronic payment mechanism, with half-hourly batches and 23 settlements. Real-Time
Gross Settlement (RTGS) is used for high-value transactions with a minimum sum
Protecting Your Assets: Effective Use of Cybersecurity Measures … 267

Fig. 2 Trends in digital

banking. Source Author

of Rs. 2 lakhs. Prepaid payment instruments like gift cards, travel cards, corporate
cards, and mobile wallets have become increasingly popular, leading to a cashless
society. Prepaid payment instruments like gift cards, travel cards, corporate cards,
and mobile wallets have become increasingly popular, leading to a cashless society.
Indian banks are offering innovative products and features to attract more clients,
benefiting both banks and making banking easier for customers. a few examples are
listed below (Fig. 2, illustrates the modern trends in digital banking):
• Online and Mobile Banking: With the rise of the internet and mobile technology,
users may now access their accounts, conduct transactions, and manage their
money from any location at any time. The use of physical branch visits is decreased
by the convenience, real-time updates, and self-service capabilities offered by
online and mobile banking systems.
• Digital Payments and Wallets: Technology has enabled the development of
digital payment systems such as mobile wallets, contactless payments, and peer-
to-peer payment platforms. The use of these solutions encourages financial inclu-
sion and lessens dependency on the physical currency by providing quick, safe,
and practical substitutes for conventional cash and card transactions.
• Blockchain and Cryptocurrencies: The financial industry has taken notice of
blockchain technology, which enables safe and open transactions. It facilitates
the growth of cryptocurrencies like Bitcoin, allows for quicker and more secure
cross-border payments, and streamlines procedures like trade finance and supply
chain management.
268 S. Das and D. Ganguly

• Open Banking and APIs: Application Programming Interfaces (APIs) have facil-
itated cooperation and innovation in the banking industry. Open Banking efforts
have done the same. Financial technology (fintech) firms can safely get client data
from banks, enabling the creation of new services and fostering competition.
• Enhanced Security and Fraud Prevention: As the number of digital transactions
has increased, banks’ key priorities now include cybersecurity. Security measures
have been strengthened by the use of cutting-edge technology like multi-factor
authentication and biometric authentication (such as fingerprints and face recog-
nition). Systems for detecting fraud that is driven by AI pore over enormous
volumes of data to spot suspicious activity and stop fraudulent transactions.
• Disruption and Innovation: Technology has paved the road for novel finan-
cial solutions and industry disruptions. Blockchain technology provides smart
contracts, safe and transparent transactions, and facilitates cross-border payments.
Cryptocurrencies have developed as new forms of digital currency, posing a threat
to existing financial institutions. Fintech firms have provided new financial goods
and services, increased competition, and forced banks to adapt and innovate.

Banks are undergoing digital transformation to improve operational effective-

ness, improve client experiences, and maintain competitiveness. Digital channels,
process automation, data analytics, and cutting-edge technology such as AI, machine
learning, and robotic process automation are essential for success.
• Personalization and Customer Experience: Banks are putting a lot of effort
into providing consumers with personalized experiences. Insights into client
preferences, behavior, and requirements may be gained by banks by utilizing
data analytics and AI, enabling the creation of specialized product offerings,
specialized suggestions, and proactive assistance.
• Sustainable Banking: Environmental, social, and governance (ESG) issues are
becoming increasingly important in the banking industry. Banks are making more
and more decisions about their investments and operations based on sustainable
practices. This involves using responsible lending and risk assessment techniques,
delivering green financial products, and financing renewable energy projects.
• Agile Workforce and Remote Employment: The epidemic has caused a move-
ment in the banking sector toward remote employment. To increase efficiency,
recruit talent, and swiftly adjust to shifting market conditions, banks are embracing
flexible work arrangements and using agile approaches.
• Application Programming Interfaces (APIs) are used by banks to securely
exchange client data with third-party suppliers as part of the open banking initia-
tives that are gaining traction globally. This pattern supports collaboration between
banks and fintech firms, fosters innovation, and permits individualized financial
services.
Artificial intelligence is helping the banking sector to offer a higher degree of
value, reduce risks, and provide better chances. It aids in the development of new ideas
and the transformation of customer service. The banking and financial industries’
key aims are to improve performance, increase profitability, and reduce risk. Data
Protecting Your Assets: Effective Use of Cybersecurity Measures … 269

Analytics is essential for lowering costs, developing new products, and expanding
client base, it has also enabled banks to lend at a reduced interest rate to key industries
such as agriculture, housing, and education.
Cyber-attacks are increasing as technology advances, and criminals are actively
seeking victims for catastrophic cyber-attacks on sensitive data stored by banks
and financial organizations. Most banks have been pushed to go online, leading
to privacy issues and phishing efforts. Cybercriminals use customer and employee
information to steal bank data and money. It is important to understand the importance
of cybersecurity in the banking sector before delving into cybersecurity concerns.

3 Status of Cybersecurity in the Indian Banking System

Cybersecurity is the act of preventing unauthorized access to, damage to, and theft
of computer systems, networks, and data. Systems and procedures must be imple-
mented to protect information from cyber threats. Cyber threats such as hacking,
malware, phishing, and ransomware attack people, businesses, and even entire coun-
tries, posing a threat to sensitive information and financial and reputational harm.
Banks need to protect their users’ assets, such as debit and credit cards, from cyber-
attacks. Banks must implement security protocols to protect data from cyber-attacks
[17]. Banks must prioritize cybersecurity to protect their data from cyber-attacks, and
they must implement cybersecurity measures to protect their customers’ data from
cyber-attacks, which can have a negative impact on their reputation and assets. Card
fraud is usually recoverable, but data infringement can take time to recover funds.
This results in the loss of customers. Data infringement is a critical issue for banks
[6], as it results in the loss of user data and makes it difficult for customers to trust
the bank. Banks must have cybersecurity requirements to evaluate current security
measures and protect critical data. Cyber-attackers have found various ways to attack
and steal data, as everything turns digital these days, and the country goes for “Digital
India”. Banks are vulnerable to organized criminals and hackers. In the recent past,
A cyber-attacker attacked and vandalized Canara Bank’s website by inserting a mali-
cious page and blocking e-payments. Union Bank of India also suffered a significant
loss due to a cybersecurity attack in India’s banking sector (Fig. 3, illustrates the Key
Cyber Considerations for Banks to Maintain Cybersecurity).
Hackers gained access by impersonating an RBI employee, and one of the
bank’s employees clicked on a malicious link, allowing the malware to manipu-
late the system. Regarding the report of Money Control [19] there were around 248
successful cybercrime cases of data breaches by hackers and criminals in the year
2022. According to the data shared by Statista, in India, there were 1,343 cybercrime
cases related to online banking in the year 2016 and 2,095, 968, 2,093, 4,047, and
4,823, in 2017, 2018, 2019, 2020, and 2021, respectively (Fig. 4, Graph illustrates
the above discussion on cybercrime cases in Indian Banking).
Worldwide around four out of ten internet users have experienced cybercrime
in the year 2022, and India was the most likely country to experience cybercrime,
270 S. Das and D. Ganguly

Fig. 3 Key cyber considerations for banks to maintain cybersecurity. Source Author

Cybercrime Cases in Indian Banking System

6,000
4,823
5,000 4,047
4,000
3,000 2,095 2,093
2,000 1,343 968
1,000
0
2016 2017 2018 2019 2020 2021
No of Cybercrimes

Fig. 4 Cybercrime cases related to online banking across India. Source Statista [28]

with nearly 70% of internet users reporting having experienced it. The United States
came in second, with 49% reporting having been victims. Australia is third with
40%, followed by New Zealand with 38%, the United Kingdom, France with 33%,
Germany with 30%, and Japan with only 21%. Cybercrime has become a major threat
to the banking industry, as the technology and expertise used by hackers are becoming
increasingly advanced, making it impossible to prevent attacks consistently. Banks
face a variety of threats to their cybersecurity, which have been elaborated on below
(Fig. 5, illustrates the country specific experience of cybercrime).

4 Cybersecurity Threats Faced by the Banking System

Cybercrime is the use of digital instruments to commit illegal acts, such as fraud,
invasion of privacy, and identity theft, which can be carried out from afar, increasing
the risk for both the bank and the consumer [23], with financial institutions and
banks being the most common targets due to their sensitive client data and potential
financial gain. Banks and financial organizations have prioritized cybersecurity by
Protecting Your Assets: Effective Use of Cybersecurity Measures … 271

Country-Specific Cybercrime Experienced By Internet

Users In 2022

Japan 21%
Germany 30%
United Kingdom 33%
France 33%
New Zealand 38%
Global 39%
Australia 40%
United States 49%
India 68%
0% 10% 20% 30% 40% 50% 60% 70% 80%
Share of Respondents

Fig. 5 Cybercrime experienced by internet user’s country wise. Source Statista [31]

investing in security infrastructure and educating clients. The Reserve Bank of India
has conducted financial awareness campaigns, but cybercriminals are constantly
changing their strategies [8]. Some of the most common strategies they use are listed
below:
• Reverse Engineering of Mobile Apps: A scammer may reverse engineer an
app to examine its source code and parts to create malware or tamper with it.
For example, attackers could create their malicious app to exploit vulnerabilities
discovered while reverse engineering the banking app [21]. If a user has both
applications installed on their device, the malicious app can redirect banking
deposits to the account without the user being aware of the breach (Fig. 6, gives
a clear image of the list of cybercrimes).
• Screen Overlay Attack: Overlay attacks are used to hijack data entry and trick
users into installing additional malware or performing unsafe tasks on their mobile
devices. They consist of an attacker-generated screen that appears on top of the
legitimate application UI and is programmed to send any information entered
into it directly to the attacker. Overlay strikes are also used to trick users into
installing other malware or performing insecure tasks on their mobile devices,
such as giving a malware app complete control of the user’s phone [33].

Fig. 6 List of cybercrimes. Source Author

272 S. Das and D. Ganguly

• Fraudulent Screen Sharing: Screen sharing fraud is a recent scam that involves
a fraudster posing as an employee of an online gaming company or a bank and
asking for remote access to the victim’s phone under a false guise. The victim is
then prompted to share the screen-sharing app’s code and complete a transaction
for a small fee. The catch is that once the victim gives the scammer the code, they
can see what they are typing on their screen, their bank account number, and all
of the information they are seeing in real time. To avoid this, one should always
check the authenticity of the official website of the company from which they get
a call.
• Keylogging/Screen Reading: Keyboard apps are available in the app store to
replace the original keyboards on mobile devices. Users usually download these
apps in an innocent attempt to personalize their gadgets. People may enjoy the
new keyboard’s hue and functionality, but some programs are rogue and contain
code that can steal personal information or do other dangerous operations.
• Message App Banking Fraud: Cheaters contact bank customers on behalf of
the bank with which the customer has an account and request that the customer
download an app in exchange for increased security or rewards. When the app is
installed, the customer is prompted to enter sensitive information, which is then
relayed back to the fraudster on the other end of the line. To avoid this, one must
be cautious when answering calls from unknown numbers and immediately notify
his branch if he has installed such an app.
• Malicious Applications Fraud: Malicious application fraud is a scam in which
the victim receives a call offering a freelance/Work-from-home job in exchange
for installing an app suggested by them. To avoid becoming a victim of this scam,
one should contact the company directly on whose behalf they were offered the
job [35].
• Sim Swap Fraud: The idea of a “sim swap” involves exchanging a defective SIM
for a new one, but the effects of this fraud can be extremely detrimental. The
fraudster can make multiple money transfers from the victim’s account without
his knowledge when they switch his SIM. To prevent this, one should maintain a
two-step verification process and set a withdrawal amount cap. Additionally, one
should contact their bank and block all transactions if their SIM suddenly stops
working.
• QR Code Scams: This fraud mostly affects people who are attempting to sell their
goods on internet marketplaces, typically valuable items such as automobiles or
cell phones. The con artist contacts the consumer and offers to buy the item for
sale. Once accepted, the fraudster will request account information through which
the payment can be paid. The buyer will claim that they are unable to send the
payment and will request that the seller scan a QR code and enter their UPI
PIN. The seller then scans the code and enters the pin, allowing the fraudster to
withdraw any amount of money from the seller while remaining anonymous. Only
scan QR codes from reliable sources to avoid this. Furthermore, the UPI PIN is
never used to receive payments.
Protecting Your Assets: Effective Use of Cybersecurity Measures … 273

• Phishing/Smishing/Vishing/Spoofing Attacks: Phishing is a scam where private

information is stolen through emails from an authentic source, such as bank offi-
cials. Hoaxers create fake links to direct customers to a fake page, which is used for
fraudulent transactions. SMS phishing is called Smishing, while voice phishing
is a social engineering attack where the attacker calls the victim to obtain infor-
mation, pretending to be representative of financial institutions. They often ask
the victim to install screen-sharing apps to gain access to their device.
• Spurt in Spy Loan Apps: Spy loan apps are instant loans with high interest rates
and fees. They collect personal information before approval and use it to harass the
borrower during loan recovery. This has caused harm to victims, such as sending
scandalous messages and blackmailing them [25].
• Crime-as-a-Service: Cybercriminals are marketing their services in the form of
Crime-as-a-Service (CaaS) and Malware-as-a-Service (MaaS), allowing novice
hackers to launch targeted attacks.
The Indian banking system has faced many real-life challenges in the recent
past, which have devastated the financial sector, and impacted a lot. These types of
incidents have been discussed in the below section.

4.1 Cyber-Attack Cases in the Indian Banking Sector

• Cosmos Bank Cyber-Attack in Pune: The Cosmos Cooperative Bank in Pune

was the target of a cyber-attack in 2018 that stole Rs. 94.42 crores. Hackers
gained access to the bank’s ATM server and stole personal information from
debit cardholders and visa holders. Hacker gangs from 28 countries withdrew
funds as soon as they were notified. To prevent this type of fraud, toughening
surveillance measures and assisting approved individuals are needed [27].
• ATM System Hacked: Canara Bank’s ATM servers were targeted for a cyber-
attack in 2018, resulting in the transfer of 20 lakh rupees from several bank
accounts. According to sources, cybercriminals gained access to ATM information
for over 300 individuals, resulting in the deaths of 50 people. Hackers acquired
information from debit cardholders using skimming devices, with transactions
utilizing stolen information ranging in value from Rs. 10,000 to Rs. 40,000. To
prevent data misuse, ATM security methods must be strengthened [32].
• RBI Phishing Scam: Criminals attacked the Reserve Bank of India in an auda-
cious phishing attempt, promising Rs.10 lakhs in prize money within 48 h if the
recipient clicked on a link to a website that looked like the RBI’s official website.
The user was asked to give personal information such as their password, I-pin,
and savings account number. The RBI had issued a warning about its bogus email
address and official website [35].
• The Bank NSP Case: A bank management trainee was engaged to be married
and used company computers to exchange emails. After a while, the young lady
made up false email addresses and sent them to the boy’s international clientele,
274 S. Das and D. Ganguly

leading to the boy’s company losing many customers and going to court against
the bank. The court determined that the bank was accountable for the emails sent
through the bank’s system [27].
• UIDAI Aadhaar Software Hacked: In 2018, the UIDAI released an official
notification on a data breach that exposed 1.1 billion Indian Aadhaar card info.
Anonymous merchants on WhatsApp were selling Aadhaar information for Rs.
500 and Aadhaar card copies for Rs. 300. The UIDAI received notification of the
hacking of 210 Indian government websites [27].
• SIM Swap Scam: In August 2018, two Navi Mumbai hackers deceptively
obtained the SIM card information of the people and illegally transferred Rs. 4
crores from their bank accounts, through online banking. In this respect, organiza-
tions must implement cybersecurity measures and adhere to the security guidelines
outlined below, and financial sectors should be aware of the risks of cyber threats
and take steps to protect themselves. One should be alert and restrict himself from
sharing his personal information with any unknown domains, as this can help
in reducing the risk of malicious content reaching people. The cyber-attacks in
India should serve as a warning to all vulnerable individuals and businesses to
implement cybersecurity measures and follow security guidelines.
Cybersecurity is very important for banks, as Digital India has resulted in a rise in
the use of digital currency and cashless transactions, making it critical to implement
all security measures to preserve data and privacy. Data breaches are a major issue in
the banking industry, as a faulty cybersecurity system could expose their consumer
database to outsiders, creating cybersecurity threats. The data breach is sensitive and
could be used against someone and cause severe harm. Banks must be vigilant 24 ×
7, or else the customers’ data with the bank may be compromised, and recovering
the data may be time-taking and annoying. Banking security must be improved to
ensure the safety of customers.

5 Steps to Reduce Cybersecurity Threats in Banking

• Access and Monitor Cloud Security: Cloud infrastructure should be examined

regularly to ensure it is up-to-date and secure. Multifactor authentication (using a
mix of tokens, biometrics, passwords, and other safe authentication techniques)
and a vulnerability management solution can be used to automate threat detection
and protect against potential threats before they become an issue [14]. Continual
monitoring and intrusion detection systems should be put in place to quickly iden-
tify and address security problems. Network traffic can be monitored, abnormal-
ities may be found, and possible threats can be quickly addressed with the use of
intrusion detection systems (IDS) and security information and event management
(SIEM) technologies.
Protecting Your Assets: Effective Use of Cybersecurity Measures … 275

• Management Policies Regarding Strict Access: Establishing strict access

management policies is essential for a successful organization, as it limits access
to personnel information to keep it secure from within the firm, and grants access
permissions to those authorized employees/users who need it. Strong authentica-
tion and access controls should be used to guarantee the authorized use of the data.
Access restrictions should be reviewed and updated often to reduce the danger of
unauthorized access.
• Increase Employee Awareness through Training: Banks must implement a
complete training program for their employees, to educate them, so that they can
deal with the cyber-attacks. Cybersecurity training for employees and profes-
sionals is essential to enhance their knowledge and skill set, related to data
protection and cyber-awareness, and keep them up to date [36].
• Data Encryption: The cryptography method is used for data encryption, which
ensures the protection of the most sensitive digital assets and data security.
• Data Recovery: Having a backup plan in place can help reduce downtime after a
disruption and prevent data loss, but only if a backup is done on a routine basis.
• Vendor Risk Management: It involves assessing and controlling how securely
third-party suppliers and service providers behave when they have access to private
information or systems. To guarantee that suppliers adhere to the necessary secu-
rity requirements, implement suitable due diligence, contractual agreements, and
security assessments.

6 Application of Cybersecurity Tools in Banking

To safeguard confidential financial information, avoid fraud, and guarantee the reli-
ability of banking systems, technology in cybersecurity must be implemented. The
following are some essential technologies that are frequently used in the banking
industry’s cybersecurity environment.
– Firewalls are used to create a barrier between internal and external networks by
regulating incoming and outgoing network traffic. They are also used in conjunc-
tion with intrusion detection systems (IDS) and intrusion prevention systems
(IPS). IDS/IPS technologies can be used to stop or prevent possible attacks by
scanning network traffic for suspicious activity.
– Secure communication channels over the internet are provided by the Transport
Layer Security (TLS) and Secure Socket Layer (SSL) protocols, respectively.
They protect the security and integrity of sensitive data, including login passwords
and financial transactions, as it is sent between servers and clients.
– MFA (multi-factor authentication) and two-factor authentication (2FA) Beyond
the usual username and password combinations, some authentication techniques
are used. By requesting additional authentication elements from users, such
as biometrics (fingerprint, facial recognition), hardware tokens, SMS codes, or
276 S. Das and D. Ganguly

mobile apps, they increase security by lowering the possibility of unauthorized

access.
Blockchain technology offers a distributed, unchangeable ledger for securely
documenting financial transactions. It can improve banking operations’ security and
transparency, especially in areas like identity verification, international payments,
and smart contracts.

7 Adoption of Blockchain Technology in Banking

Blockchain is a distributed ledger that is immutable and cryptographically secure,

making it a vital tool for checking authenticity and assuring security while performing
financial transactions online, it is used to transfer money between parties without the
involvement of a third party or middlemen. It functions by constructing a network of
computers [20] linked over the internet that records, stores, preserves, and synchro-
nizes copies of transaction data that cannot be changed or destroyed. It is also hard to
fabricate, making it an essential instrument for ensuring the legitimacy and security
of financial transactions conducted online. The banking industry was among the first
to see the potential of Distributed Ledger Technology, which is a technology that
allows for the secure functioning of a decentralized digital database. It is a tech-
nology that underpins cryptocurrencies such as Bitcoin, but its applications extend
beyond virtual money [1].
Global Banking Institutions have made significant strides toward digitalization,
but their efforts have been mostly neglected when it comes to blockchain in banking,
in contrast with the widespread interest in blockchain technology of other businesses.
But soon, it is expected to rise from 4.9 billion dollars in 2021 to more than 67.4
billion dollars by 2026. The Fig. 7 graph, illustrated by Statista gives a clear idea of
the market size of Blockchain Technology Worldwide:

Blockchain Technology Market Size Worldwide from 2017 -

2027
200
162.84
150
94.89
100
55.54
50 32.69
0.98 1.57 2.55 4.19 6.92 11.54 19.36
0
2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027

Market in billion U.S. Dollars

Fig. 7 Blockchain technology market size worldwide. Source Statista [3, 4]

Protecting Your Assets: Effective Use of Cybersecurity Measures … 277

The Global Market for Blockchain Technology in Banking

from 2018 - 2026
40
22.46
20 13.27
4.39 7.64
0.28 0.48 0.84 1.46 2.53
0
2018 2019 2020 2021 2022 2023 2024 2025 2026

Market in billion U.S. Dollars

Fig. 8 Blockchain technology in banking market size worldwide. Source Statista [5]

The reticence of banks to use blockchain technology in banking and finance is

due to the lack of large-scale applications and regulatory barriers. However, banking
institutions have begun to employ the technology on a limited scale. Blockchain tech-
nology is becoming increasingly important in banking and real-world applications,
and it has many uses, that has been discussed below, but the global banking scenario
is still too slow to innovate due to paperwork, security concerns, and time-consuming
and costly processes. For example, banks still require a significant amount of paper-
work, have security concerns, and have several time-consuming and costly processes
in place. A report published by Statista Research Department, 2022 also gives an idea
of the market size for blockchain solutions for banking and financial institutions in
US Billion dollars (Fig. 8, illustrates the Global Market for Blockchain Technology
in Banking):
• Payment Transfer: Cryptocurrencies like Ether and Bitcoin are built on public
blockchains, allowing anyone to send and receive money in real-time with no
transaction fees and no authentication required, making money transfers faster
and cheaper
• Settlement and Clearance: A typical bank transfer can take up to 3 days to
complete, which is very problematic both for consumers and banks. Bank’s cross-
border transfers are now simpler and faster than ever, eliminating the need for
middlemen and custodial services, with the advent of blockchain technologies.
Blockchain is a decentralized ledger that records transactions transparently and
publicly, allowing transactions to be resolved in the public blockchain and real-
time settlement, hence speeding up and simplifying transactions. Banks utilize
Ripple Net as an illustration of a blockchain-based payment network for inter-
national transfers. Example: Al Rajhi Bank’s cross-border payment services are
based on Ripple’s blockchain capabilities.
• Securities and Debts: Blockchain technology generates a decentralized database
of digital and unique assets, making tokenized asset transfers quicker and
allowing for the elimination of intermediaries. Tokenization makes previously
illiquid assets more accessible and allows for fractional ownership. The issuing,
trading, and settlement of digital securities are made possible by blockchain-based
systems, possibly revolutionizing the financial markets.
278 S. Das and D. Ganguly

• Loans & Credits: Blockchain in consumer banking expands peer-to-peer lending,

making it one of the most investable fintech areas. Banks assess the risk of nonpay-
ment of interest on loans based on factors such as credit score, ownership status,
and debt-to-income ratio. Blockchain in banking provides an efficient, low-cost,
and secure way to obtain personal loans, made easier with a decentralized register
of payment history.
• Digital Identity and Customer Onboarding (KYC): Blockchain can revo-
lutionize the way banks handle customer identity verification and onboarding
processes. By leveraging blockchain-based identity systems, banks can enhance
security, streamline customer due diligence, and reduce the need for redundant
verification processes. This can improve the customer experience and reduce costs
associated with compliance.
• Innovation and Competitive Advantage: Banks can maintain their compet-
itiveness in the quickly changing financial market by using blockchain tech-
nology. Banks may use cutting-edge solutions, investigate novel business models,
and provide their clients with improved services by implementing blockchain
technologies [10]. Additionally, it permits them to work with fintech compa-
nies and investigate joint ventures to promote digital innovations. Example: JP
Morgan Link encourages and enables businesses to securely and quickly exchange
payment-related information and The HSBC bank is using R3 blockchain tech-
nology to enable Digital Vault, a custodial blockchain platform for keeping digital
assets. The technology decreases the cost of their custodial service dramatically
(Fig. 9, graph gives an idea of the uses of blockchain technology)
• Trade Finance: By digitizing and automating procedures like letters of credit,
invoice financing, and supply chain finance, blockchain can increase the efficiency
and transparency of trade finance. By automating the execution of commercial
agreements and enabling real-time tracking of commodities, smart contracts on
the blockchain can cut down on paperwork and streamline the entire trading
process. Example: UBS launched the first publicly traded digital bond that could
be resolved on both traditional and blockchain-based platforms [24].
• Fundraising Through Digital Currencies Issued by Central Banks (CBDCs):
Initial coin offerings (ICO) is a new concept for financing, which unbundles access
to finance capital from traditional fund-raising services. Several central banks are
looking at using blockchain technology to create their digital currencies, this
increases financial inclusion, and payment efficiency, and gives central banks
more influence over monetary policy. Example: Swedish Central Bank (SCB) is
experimenting with the idea of introducing its digital currency e-krona in part-
nership with R3. This will become a country-wide usable cryptocurrency. On the
other hand, Goldman Sachs is assisting payment company Circle in its goal of
becoming a blockchain-enabled issuer of USD Coin, a digital stablecoin linked
to US dollars [26].
In India, Blockchain technology is gaining popularity among Indian banks due
to its ability to promote transparency and increase traceability of money movement.
Money laundering frauds are the most prominent ones, that involve rubbing out
Protecting Your Assets: Effective Use of Cybersecurity Measures … 279

Uses of Blockchain
70%
60%
60%

50%

40%

30% 23%
20% 19% 19%
20% 16% 15%
12% 11%
10%

0%

Fig. 9 Uses of blockchain technology. Source Delloite & EFMA [7]

crores and crores of rupees of the taxpayers, putting the money at stake. Banks are
exploring blockchain-based trade finance initiatives to promote a safe and secure
banking environment. The RBI is trying to explore the use of blockchain technology
in the banking industry, by developing a proof-of-concept blockchain project, to
handle trade finance [11]. State Bank of India has collaborated with other commer-
cial banks to prototype a blockchain-based application. Private sector banks like Yes
Bank, HDFC Bank, ICICI Bank, and Axis Bank are also adopting blockchain tech-
nologies into their banking operations. The Securities and Exchange Board of India
directed its depositories to use blockchain technologies to encourage transparency
in keeping records and monitoring the creation of securities [30].

8 Fraud Prevention Using Blockchain Technology

Blockchain technology has the potential to be a valuable tool in the prevention of

fraud. While blockchain has many useful features, few, in particular, are crucial for
fraud detection and prevention. They are as follows.
280 S. Das and D. Ganguly

• Immutable Ledger: The distributed ledger used by blockchain technology

assures that once a transaction is recorded, it cannot be changed or removed.
Due to its immutability, it is challenging for fraudsters to alter or modify trans-
action records. A significant disincentive against fraudulent actions is created by
the rapid identification of any efforts to change data on the blockchain [26].
• Consensus: Blockchain networks are dependent on consensus processes to
authenticate and verify transactions [34]. The network guarantees that transac-
tions are authorized by a majority of members, making it impossible for a single
entity to influence or engage in fraudulent activity, depending on the consensus
method employed, such as Proof of Work (PoW) or Proof of Stake (PoS) [15].
• Collaboration and Shared Data: Blockchain enables safe data exchange across
many organizations, facilitating cooperation between banks and other ecosystem
players. Shared data may include banned addresses, questionable transaction
patterns, and fraud indications. By utilizing blockchain, banks may work together
in real-time to jointly identify and stop fraud, fortifying their defense against such
actions [12].

9 Barriers to Adoption of Blockchain Technology

Banks and traditional financial institutions are embracing blockchain technology,

which enables new services and aids in internal back-end procedures. However, adop-
tion remains limited due to some typical obstacles banks encounter when merging
blockchain technology and digital assets. Banks viewed blockchain as a high-risk,
short-term business, similar to bitcoin, but technology has continued to alter the
financial sector in recent years. Banks are learning how to use blockchain to increase
efficiency and automate procedures, and it can generate new revenue streams and
attract customers. However, it is important to remember that all growth occurs when
banks face challenges. Below are a few challenges faced by the banks during the
adoption of Blockchain Technology (Fig. 10, illustrates the barriers of blockchain
technology).
• Budget: When it comes to blockchain deployment, many organizations prioritize
the budget. Because the technology is new, the banks may be uninformed of the
expense of implementation.
• Lack of Understanding and Skillset: Blockchain is a complex concept that
requires a full understanding to maximize its potential benefits. Banking IT teams
are unlikely to have enough blockchain professionals to assist in the creation of
blockchain projects and the dissemination of blockchain knowledge and concepts.
Blockchain expertise is on the rise, but it can be difficult to understand which skills
a bank is looking for [19].
• Privacy and Security Concerns: Blockchain and privacy cannot get along
because the system is powered by decentralized ledger technology. Businesses
that deal with privacy must follow tight criteria to minimize data access. As
Protecting Your Assets: Effective Use of Cybersecurity Measures … 281

Fig. 10 Barriers of blockchain technology. Source Author

hackers may take over and exploit the network, blockchain technology poses a
few security concerns. To address this, the protocol layer requires increased secu-
rity, yet only a few scenarios have effective protocols. Nobody knows if they are
safe to use for a lengthy period.
• Slow and Cumbersome: The blockchain is a complicated technology that takes
longer to execute transactions, and its encryption considerably slows it down. It
is best suited for huge transactions when speed is not a factor, but it is risky and
was not designed to remove the “insecure” nature of blockchains. It works best
for large transactions where time is not an issue.
• Public Perception: Blockchain technology has a lack of knowledgebase, making
it unpopular among the general public. To be successful, it must gain acceptability
and gain adequate promotion to attract more customers [13]. Without adequate
promotion, blockchain technology will remain unpopular. Therefore, if someone
is not involved in this segment, he will be unaware that it exists. Majority of the
people are unaware of the fact that not only Bitcoin but other digital currencies
come under the blockchain network.
• Scalability: When implementing blockchain, scalability is an issue that must be
addressed. As the network’s user base expands, transitions take longer to execute,
increasing transaction costs and limiting the number of users on the network.
As a result, blockchain adoption has been challenging, making the technology
282 S. Das and D. Ganguly

less profitable. Few blockchain technologies produced faster results, but they also
slowed as more users registered onto the system [29].
• Inefficient Technological Design: Blockchain technology is currently lagging in
key technological areas due to code flaws and loopholes. Bitcoin was the cutting
edge in this regard, but Ethereum attempted to cover up its weaknesses, but it was
insufficient. The majority of these issues are caused by improper programming
and loopholes, which users can quickly exploit and gain access to the system.
Security jargon is not working here.
• Regulation: Blockchain has advantages, but it also poses challenges of regulation
and compliance. Financial institutions are used to tight regulations and laws,
and blockchain is pseudonymous, so a strong regulatory framework is needed
to prevent illicit activity. However, due to a lack of clear rules, integration is
problematic [18].
• Energy Consumption: Blockchain technology is based on Bitcoin and uses Proof
of Work as a consensus procedure. However, mining requires the use of a computer
to solve complex equations, which requires more electricity. To counteract this,
the PC will use more electricity during mining.
• The Criminal Connection: Blockchain technology has attracted both profes-
sionals and criminals, leading to the use of Bitcoin as the principal currency in the
illegal market and dark web. Criminals are using bitcoins to purchase restricted
equipment and payment methods, and are demanding cryptocurrency as a ransom.
To combat this, it is important to break the criminal link and improve blockchain
implementation [7].

The potential for incorporating and embracing blockchain technology in India

is immense, as the Indian Government is trying to initiate a national blockchain
framework that will transform as many as 44 sectors. According to projections,
blockchain technology will be worth 176 billion dollars by 2050, leading to the
adoption of blockchain technology in the future. In the banking sector, blockchain has
the potential to be a disruptive technology that banks may use to automate processes,
improve services, and provide a greater digital experience to their customers. Service
providers will collaborate with banks and other financial institutions to bridge the
trust and knowledge gap, as well as provide technology infrastructure support and
the appropriate services. Making the right investment at the right time is key to riding
this growing tide. The below graph is a study by Statista, where it is seen the growth
rate of different cybersecurity products market between the years 2019 and 2022.
The growth rate for data security products is the maximum (22.20%), in comparison
to endpoint security (19.10%), Identity and Access Management (16.30%), Network
Security (15.30%), and security IDR (14.80%) (Fig. 11, gives an idea of the growth
rate of India’s Cyber Security Product).
Protecting Your Assets: Effective Use of Cybersecurity Measures … 283

India's Cyber Security Product Growth Rate

between 2019-2022, by Product Type
25.00% 22.20%
19.10%
20.00%
16.30% 15.30% 14.80%
15.00%

10.00%

5.00%

0.00%
Data Security Endpoint Identity and Network Security IDR
Security Access Security
Management

Growth Rate %

Fig.11 Growth rate of cybersecurity products in India. Source Statista [16]

10 Future of Cybersecurity in the Banking Industry

Cybersecurity in the banking industry is anticipated to evolve as a result of growing

cyber threats, regulatory changes, and continuous technological improvements.
The following significant factors are probably going to have an impact on how
cybersecurity in the banking industry develops:
• Advanced Threat Detection and Response: Banks will need to use advanced
threat detection and response techniques as cyber-attacks become more complex.
This involves the use of behavior-based analytics, real-time monitoring, artificial
intelligence (AI) and machine learning (ML) algorithms to detect and quickly
respond to possible cyberattacks.
• Collaboration and Information Sharing: are essential to combating the
constantly developing cybersecurity threats that banks cannot handle alone. It will
be essential for banks, industry peers, governmental organizations, and cybersecu-
rity organizations to collaborate and share information more often. Sharing threat
intelligence, best practices, and lessons gained can assist banks in proactively
detecting and countering new cyber threats.
• Prepare for Quantum Computing: Banks must embrace quantum-safe encryp-
tion to get ready for the post-quantum era as a result of the probable advent of
quantum computing. To safeguard confidential information and guarantee long-
term security in the face of risks posed by quantum computing, quantum-resistant
encryption methods will become crucial.
284 S. Das and D. Ganguly

• Artificial Intelligence and Machine Learning: These two types of technology

will remain crucial to cybersecurity. They may support incident response, auto-
mate threat detection, and search through massive amounts of data for patterns
and abnormalities. AI-driven systems can also strengthen proactive defenses and
boost overall cybersecurity posture by learning from previous assaults [22].
• Cybersecurity Workforce and Skills: There is a growing need for qualified
cybersecurity experts. Human error continues to be a major cybersecurity risk. To
inform staff about new dangers, phishing tricks, and security best practices, banks
will need to invest in thorough cybersecurity training programs. Campaigns to
raise awareness of cybersecurity will be necessary to develop a security-conscious
organizational culture. Artificial intelligence and automation will also play a part
in enhancing cybersecurity capabilities, streamlining procedures, and addressing
the skills problem.
• Zero Trust Framework: The financial sector is embracing the idea of a zero-trust
framework. It places a strong emphasis on stringent access restrictions, ongoing
supervision, and user verification at each point of the user experience. Banks may
reduce the danger of lateral movement across their networks and guarantee that
each transaction or access request is properly verified and authorized by putting
Zero Trust principles into practice [6].
• Continuous Monitoring and Incident Response To quickly identify and miti-
gate cyber threats, real-time monitoring, threat hunting, and fast incident response
skills will be essential. To improve their capacity for responding to incidents,
banks will invest in enhanced security operation centers (SOCs) with advanced
analytics, automation, and threat intelligence capabilities.

References

1. How Blockchain Could Disrupt Banking (2022). https://www.cbinsights.com/research/blockc

hain-disrupting-banking/#:~:text=Further%2C%20blockchain%20technology%20allows%
20for,transaction%20days%20after%20a%20payment
2. Alam SMS, Chowdhury MAM, Razak DBA (2021) Research evolution in banking perfor-
mance: a bibliometric analysis. Future Bus J 7:66. https://doi.org/10.1186/s43093-021-001
11-7
3. Blockchain—statistics & facts (nd). https://www.statista.com/topics/5122/blockchain/#topicO
verview
4. Blockchain technology market size worldwide in 2017 and 2018, with forecasts from 2019
to 2027 (2023). https://www.statista.com/statistics/1015362/worldwide-blockchain-techno
logy-market-size/#:~:text=This%20statistic%20shows%20the%20size,billion%20U.S.%20d
ollars%20by%202027
5. Blockchain use in banking and financial services market size worldwide in 2018 and 2019 with
a forecast to 2026 (2022). https://www.statista.com/statistics/1229290/blockchain-in-banking-
and-financial-services-market-size/
6. Collins J, Singh S (2020) Cybersecurity risks and safeguards in the banking sector: a systematic
review. Int J Inf Manage 50:327–339
Protecting Your Assets: Effective Use of Cybersecurity Measures … 285

7. Dalton D, Gianturco P (2016) Out of the blocks: Blockchain—from hype to prototype.https://

www2.deloitte.com/content/dam/Deloitte/nl/Documents/financial-services/deloitte-nl-fsi-blo
ckchain-from-hype-to-prototype-out-of-the-blocks.pdf. Accessed 31 May 2023
8. Dave P (2022) Financial cyber frauds in India 2022 and learnings for 2023. https://timeso
findia.indiatimes.com/blogs/voices/financial-cyber-frauds-in-india-2022-and-learnings-for-
2023/. Accessed 31 May 2023
9. Dhameja NL, Arora S (2020) Banking in India: evolution, performance, growth, and future.
Indian J Pub Adminis 66(3):312, 326. https://doi.org/10.1177/0019556120953711
10. Dhar S, Bose I (2016) Smarter banking: blockchain technology in the indian banking system.
Asian Manag Insights (Singapore Manag Univ) 3(2):46–53. Asian Management Insights.
https://ink.library.smu.edu.sg/ami/3
11. Dubey V (2022) The adoption of blockchain technology in India and its future. 2022 The
adoption of blockchain technology in India and its future. BusinessToday. Accessed 8 June
2023
12. Balagolla EMSW, Fernando WPC, Rathnayake RMNS, Wijesekera MJMRP, Senarathne AN,
Abeywardhana KY (2021) Credit card fraud prevention using blockchain. In: 6th International
conference for convergence in technology (I2CT), Maharashtra, India, pp 1–8. https://doi.org/
10.1109/I2CT51068.2021.9418192
13. Garg P, Gupta B, Kapil KN, Sivarajah U, Gupta S (2023) Examining the relationship between
blockchain capabilities and organizational performance in the Indian banking sector. Ann Oper
Res. https://doi.org/10.1007/s10479-023-05254-0, http://hdl.handle.net/10454/19373
14. Glaeser S, Osgood ND, Bamber MR (2018) Cybersecurity in banking: an organizational
perspective. J Inf Syst 32(2):69–88
15. Gregory A (2021) Digital vision: digital banking financial services and insurance global opinion
paper. https://atos.net/wp-content/uploads/2021/10/Digital-Vision-Digital-Banking-Opinion-
Paper-Report.pdf
16. Growth rate of cyber security products market in India between 2019 and 2022, by product type
(2022). https://www.statista.com/statistics/1099965/india-growth-rate-cyber-security-market-
by-product-type/
17. Gupta S, Gollmann D (2017) Cybersecurity in banking: perspectives on the role of national
culture. Inform Manag 54(2):227–237
18. India and the Blockchain technology, 2022 India and the Blockchain technology. J India.
Accessed 8 June 2023
19. Indian banks reported 248 data breaches in last four years, says government (2022) Indian banks
reported 248 data breaches in last four years, says government. moneycontrol.com. Accessed
8 June 2023
20. Javaid M, Haleem A, Singh RP, Suman R, Khan S (2022) A review of blockchain technology
applications for financial services. In: Bench council transactions on benchmarks, standards
and evaluations, vol 2, no 3. https://doi.org/10.1016/j.tbench.2022.100073
21. Jun Y, Park SY, Kim D (2019) Cybersecurity investments in the banking industry: an empirical
analysis. Int J Inf Manage 44:173–185
22. Khanna P, Haldar A (2023) Will adoption of blockchain technology be challenging: evidence
from Indian banking industry. Qual Res Finan Markets 15(2):361–384. https://doi.org/10.1108/
QRFM-01-2022-0003
23. Kim J, Solomon MG (2019) Assessing cybersecurity risks in the banking sector: a Bayesian
network approach. J Bus Res 98:401–410
24. Kujur A (2018) DATA STORY: India houses the second largest pool of blockchain developers
in the world. moneycontrol.com
25. Kurmanath KV (2023) Cybersecurity trends 2023: from Crime-as-a-Service to banking
trojans, these are the threats to watch out for. https://www.thehindubusinessline.com/info-
tech/cybersecurity-trends-2023-from-crime-as-a-service-to-banking-trojans-these-are-the-thr
eats-to-what-out-for/article66508450.ece. Accessed 31 May 2023
26. Li C (2022) A fraud detection system for reducing blockchain transaction risks using explain-
able graph neural networks. The George Washington University ProQuest Dissertations
Publishing, p 29325150
286 S. Das and D. Ganguly

27. Lodha H, Mehta D (2022) An overview of cyber crimes in banking sector. https://www.leg
alserviceindia.com/legal/article-7694-an-overview-of-cyber-crimes-in-banking-sector.html.
Accessed 31 May 2023
28. Number of cybercrimes related to online banking across India from 2016 to 2021 (2022) https://
www.statista.com/statistics/875887/india-number-of-cyber-crimes-related-to-online-banking/
29. Pandey AA, Fernandez TF, Bansal R, Tyagi AK (2021) Maintaining scalability in blockchain.
In: Intelligent systems design and applications: 21st international conference on intelligent
systems design and applications (ISDA 2021) held during December 13–15, pp 34–45. https://
link.springer.com/chapter/10.1007/978-3-030-96308-8_4
30. Patel E (2022) How RBI and Indian Banks are piloting blockchain trade financing. inc42.com.
Accessed 8 June 2023
31. Percentage of internet users in selected countries who have ever experienced any cybercrime
in 2022 (2023). https://www.statista.com/statistics/194133/cybercrime-rate-in-selected-cou
ntries/#:~:text=In%202022%2C%20around%20four%20in,to%20have%20ever%20experie
nced%20cybercrime
32. Sarmah A, Sarmah R, Baruah AJ (2017) A brief study on cyber crime and cyber laws of India.
IJRET 4(6)
33. The Rise of Cybercrime in Indian Banks (nd) https://www.clari5.com/the-rise-of-cybercrime-
in-indian-banks/. Accessed 31 May 2023
34. Ullah N, Al-Rahmi WM, Alfarraj O, Alalwan N, Alzahrani AI, Ramayah T, Kumar V (2022)
Hybridizing cost saving with trust for blockchain technology adoption by financial institutions.
Telematics Inform Rep 6. https://doi.org/10.1016/j.teler.2022.100008
35. Verma VK (2014) Phishing (Indian Case Law, 16 July 2014). https://indiancaselaw.in/phishing/.
Accessed 1 June 2023
36. Vijayalakshmi P, Priyadarshini V, Umamaheswari K (2021) Impacts of cyber crime on internet
banking. Int J Eng Technol Manag Sci 5(2). https://doi.org/10.2139/ssrn.3939579
Revolutionizing Banking
with Blockchain: Opportunities
and Challenges Ahead

Shilpa Mahajan and Meghna Nanda

Abstract This chapter explores the opportunities and challenges that blockchain
presents for the banking industry. Secure transactions are possible due to blockchain’s
distributed and inflexible nature, which results in efficient and modest cross-border
payments, streamlined remittance processes, and reduced settlement times. It also
explores the potential impact of integrating cloud computing with blockchain and
critically examines the current state of blockchain endorsement in this sector, high-
lighting successful use cases and early implementations. It delves into the poten-
tial benefits that blockchain can bring to different aspects of banking and assesses
the challenges that need to be overcome for full-scale integration. It also empha-
sizes blockchain’s disruptive influence on the financial sector and the significance
of handling the opportunities and difficulties posed by this technology. This chapter
also discusses the traditional approach used by banking earlier and how blockchain
has changed the outlook of the present banking industry.

Keywords Blockchain · Architecture · Cloud computing · Case studies ·

Opportunities

1 Introduction

Today Banking is one of the most used sectors of the world, from UPI payment
to stock market everything moves around banking and its transactions only. Banks
are enhancing their customer services day by day to increase the rate of security,
transparency, and the speed of transactions. With the publication of blockchain tech-
nology around the world in the past years, it shows the power to revolutionize the
banking sector to create a secure and fast transaction which also helps in creating an
automatic world of banking.
Blockchain is a distributed ledger system which make transactions recorded and
verified without the need for intermediaries. It is a decentralized system which allow

S. Mahajan (B) · M. Nanda

Department of Computer Science, The NorthCap University, Gurugram, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 287
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_13
288 S. Mahajan and M. Nanda

safe, tamper-proof, and transparent transactions. Blockchain technology can address

many challenges by giving a protected and translucent system for noting transactions,
decreasing the use of intermediaries, and increasing the speed and productivity of
transactions.
Cloud computing allows on-demand access to a shared pool of resources over
the internet, and has gained significant attraction across various industries. It offers
scalability, flexibility, cost-efficiency, and the ability to access computing power and
data storage resources as needed. By leveraging cloud computing, banks can optimize
their operations, streamline processes, and effectively manage vast amounts of data.

2 Characteristics and Architecture of Blockchain

Blockchain is a digital ledger system which sanction secure and transparent recording
and verification of monetary exchange. It is a decentralized system that is intended
to function independently of a central authority and is preserved and proven by a
network of computers, called nodes. The copy of the ledger is kept by every node,
and new transactions which are checked and supplementary to the ledger through
a consensus mechanism, like Proof of Stake (PoS) or Proof of Work (PoW). In a
blockchain system, each block contains a different identifier, a timestamp, and a
digital signature that verifies the authenticity for the transaction. The online proof
is created using cryptographic algorithms, which ensure that the data is secure
and cannot be altered or manipulated. One of the most important attributes of the
blockchain model is its immutability, which implies a transaction which is saved once
in the blockchain will not be updated. This creates a transparent and tamper-proof
system that can be used for a huge variety of usage like financial transactions and
identity verification. In a blockchain system, each transaction is saved in a block and
added to the chain of the last blocks, producing a stable and tamper-proof history of
every change in data as shown in Fig. 1.
The architecture of blockchain consists of the following key components (see
Fig. 2).
Distributed network: In a blockchain, each node is connected to another node in
a distributed manner, where each node stores a copy of the blockchain ledger. This
distributed network enables blockchain to operate in a decentralized and transparent
manner, where no single entity controls the network, or the data stored on it.
Blocks: Blockchain stores data in blocks that has a set of transactions and a unique
digital signature called a hash. The integrity of the block is assured by hash makes it
tamper-proof. Each block is attached to the last block with the hash, making a chain
of blocks called a blockchain.
Consensus mechanism: Blockchain operates on consensus mechanism which
makes all nodes in the network accord with the state of ledger. There are several
consensus mechanisms, like Proof of Work and Proof of Stake. These mechanisms
ensure that the blockchain is secure and resilient to attacks.
Revolutionizing Banking with Blockchain: Opportunities … 289

Fig. 1 Working of blockchain

Fig. 2 Components of
blockchain

Smart contracts: Smart contracts mean blockchain can safely secure and run
contracts on their own which means automatically contracts can be executed in
blockchain to increase the performance by providing the contracts with certain
conditions.
Cryptography: Blockchain operates by using cryptographic techniques to make
sure that the protection and confidentiality of the data stored on the blockchain. For
290 S. Mahajan and M. Nanda

example, blockchain uses cryptography to ensure that only the owner of a private
key can access and modify the work saved in the blockchain.
Overall, the structure of blockchain is made to provide a secure, transparent, and
decentralized platform for storing and sharing data. The architecture of blockchain
has the energy to convert many distinct industries, like finance and banking, supply
chain, and healthcare, through permitting secure and efficient records sharing and
collaboration.

3 Blockchain and Cloud Computing Integration

The integration of blockchain generation with cloud computing provides a unique

opportunity for the banking area. By combining the security and transparency of
blockchain with the scalability and flexibility of cloud computing, banks can liberate
new tiers of innovation and efficiency.
Blockchain generation, with its decentralized and tamper-proof nature, addresses
the challenges of conventional banking systems, together with sluggish transaction
speeds, excessive charges, and shortage of transparency. It allows stable and trans-
parent recording and verification of transactions, eliminating the want for interme-
diaries. On the other hand, cloud computing offers the infrastructure and computing
assets required to assist blockchain applications at scale. Figure 3 shows blockchain
and cloud computing integration.
By utilizing cloud computing, banks can leverage the elastic computing capabili-
ties to process blockchain transactions quickly and efficiently. A number of services
are provided by cloud service providers, including virtual machines, storage, and
networking, which can be utilized to build and deploy blockchain networks. Addi-
tionally, the cloud provides a reliable and scalable environment for hosting blockchain
nodes, ensuring continuous availability and redundancy.

Fig. 3 Blockchain cloud computing integration

Revolutionizing Banking with Blockchain: Opportunities … 291

4 Challenges with the Traditional Approach

A conventional system, used by banks in the financial industry, is the foundation of

international trade in finance and is thought to be responsible for the daily transfer
of more than a trillion dollars. Credit card systems, the Internet, and mobile appli-
cations are just a few of the significant advancements and technology that have been
integrated into the banking system over time. By enhancing efficiency, convenience,
and transaction speed as well as minimizing the distance between parties involved,
emerging technologies like blockchain have greatly aided banks in running their
businesses.
The integration of blockchain technology and cloud computing opens a myriad
of opportunities for the banking sector. It enables banks to develop and deploy
blockchain-based applications, streamline processes such as cross-border payments,
trade finance, and identity verification, and enhance customer experiences by
providing secure and transparent financial services.
The major challenge includes the insufficient comprehension of the technology
and its potential applications in the banking sector. Most of the banks hesitate to
adopt blockchain due to the perceived risks and uncertainties associated with it.
Another challenge is the regulatory environment, which varies across the country,
not only banking sector but also hinders the adoption of blockchain among others.
Banks have to comply with laws on privacy, data security, and financial crime
prevention. There may also be barriers to blockchain adoption.
However, the combination of blockchain and cloud computing also presents some
challenges. These include data privacy concerns, compliance, interoperability, busi-
ness improvements, and the need for strong governance structures. Banks need to
carefully consider these challenges and implement appropriate solutions to ensure
the successful integration and adoption of blockchain and cloud computing in their
operations.
Finally, blockchain’s scalability and performance limits are also a concern, espe-
cially as the number of transactions and data volumes in the banking sector continue
to grow. Banks need to ensure their blockchain systems can handle more transactions
and data while maintaining high levels of security and reliability.
This chapter addresses the challenges and opportunities of enforcing blockchain
technology in the banking sector and the solutions that can be used to overpower
these difficulties and foster the widespread adoption of blockchain in banking.

5 Literature Review

Blockchain technology is getting popular day by day in the last few years due to
its unrealized power to transform several industries including finance. Blockchain
provides a decentralized and transparent platform for storing and sharing data, which
292 S. Mahajan and M. Nanda

can increase the efficiency, security, and transparency of banking operations. Signif-
icant research has been done in this area to measure the effectiveness of the amal-
gamation of blockchain in banking sectors. Few of the works have been presented
below.
The meaning of blockchain and the way it works which means how data is trans-
ferred in the form of blocks during any transaction have been well explained in the
research paper of Mahajan [1]. The research paper mainly focuses on the application
of using blockchain in healthcare sector which also discusses the advantages and
limitations of using blockchain.
To assist students in understanding and conducting research effectively, Adams
et al. [2] published a book in the year 2014 providing a valuable way to evaluate
the key features, strengths, and limitations of research and helped to find a way of
how research must be done following what pattern that helped in finding the best
methodologies for research. To understand the digital currencies and central bank
policies, Energet and Fung [3] published their thoughts in a discussion paper in the
year 2017 which aims to highlight the evolving landscape of digital currencies and
central bank policies helping in finding the issues faced by central banks in secure
transaction leading to motivation for development.
To provide an insight into using blockchain technique in finance, Shorman et al.
[4] published a research in the year 2020 providing a review of using blockchain in
finance which helped in finding the potential application of banking using blockchain
which includes the sectors which all can be transformed using blockchain which helps
in working smooth running of blockchain.
The usage of blockchain technology in banking area is explained in a Deloitte
Insights [5] report as one of the most potential and powerful weapons to be used in
banking by elaborating about the meaning of blockchain, benefits of using blockchain
in banking industries about transparency, security, and many other benefits. They have
also discussed the difficulties of applying blockchain in banking the funds required,
proper knowledge of using it, and many more.
A review of applying blockchain in banking has been discussed by Sridharan and
Shankar [6] in their article which includes the basics of blockchain that is what is
blockchain, how transformation occurs in blocks, advantages of using blockchain in
banking, areas in which blockchain could be used like transaction. They have also
discussed the challenges of using blockchain in banking as applying blockchain in
banking is not easy and it requires a lot of funds that have to be raised which can be
used for smooth running of banks using blockchain.
An article written by Iqbal and Wasiq [7] was released providing a comprehensive
review of the recent literature about research on applying blockchain in banking and
finance sector which also includes the potential challenges and prospects of using it.
To examine the capability of using blockchain in banks, Vali and Madani [8]
published an article about the capabilities of using blockchain not only in banking but
also in other sectors like health sectors and others. It provides information about how
secure, efficient, transparent, and smooth banks can run their day-to-day operations
including the same for other sectors too.
Revolutionizing Banking with Blockchain: Opportunities … 293

To inform the world about the importance of addressing some issues such as
accessibility, usability, and trust in using blockchain for remittances, microfinance,
and online identity, Yeoh and Choi [9] published their research paper which consisted
of all the above issues along with its solutions. The research paper also includes
various applications of blockchain which also consist of banking and finance. The
challenges and opportunities are discussed in detail in the research paper.
The applications of blockchain and features like efficiency, security, and trans-
parency have been published in a research paper by Heshmati and Gupta [10]
which consists of the challenges required to adopt, regulate, and interoperate before
applying blockchain in every sector. The advantages which will help in smooth
running of the sector after applying blockchain will lead to advancement of the
world.

5.1 Technology Acceptance Model (TAM) Perspective

TAM is one of the most widely used theories which explains users’ adoption of new
technology based on their discern value and discern use. TAM proposed that the way
of using a technology is determined by the discern value of technology and discern
use that is to make it work-friendly with everyone as shown in Fig. 4.
In the context of blockchain in banking, TAM can be used to explain banks’
adoption of blockchain technology. The the usefulness of blockchain in the banks can
include reducing fraud, improving cross-border payments, and enhancing regulatory
compliance. It considers the ease of use, which can make the technical complexity
of integrating blockchain with existing banking systems, the cost of implementation,
and the level of regulatory compliance required.

Fig. 4 Original TAM [11]

294 S. Mahajan and M. Nanda

6 Diffusion of Innovation Theory

This theory provides an explanation, for how new innovations are adopted and spread
throughout the society. It suggests that various factors play a role in determining the
acceptance of technology such as the advantages it offers compared to existing tech-
nology, its compatibility with systems, and its level of complexity. When considering
the use of blockchain in the banking sector, we can apply the diffusion of inno-
vation theory to understand why banks are increasingly adopting blockchain. The
advantages of blockchain over banking systems can include improved efficiency,
transparency, and security. The compatibility of blockchain with existing systems
depends on factors like integration complexity, regulatory compliance requirements,
and the bank’s cultural and organizational norms. Lastly, the complexity of imple-
menting blockchain can vary based on the expertise needed, implementation costs,
and potential risks associated with adopting technology.
Focusing on the core benefits of blockchain, this study examines the key factors
that compel banks to adopt blockchain solutions and offers effective strategies to
accelerate their adoption. By implementing distributed ledger technology, banks can
enhance security, streamline operations, reduce costs, and increase transparency. In
addition, the chapter highlights the importance of smart contracts and how decen-
tralized finance applications are changing traditional banking practices. Suggested
strategies include building industry-wide collaborations, establishing regulatory
frameworks, fostering collaboration between blockchain platforms, and investing
in blockchain-focused R&D. Overall, this chapter seeks to provide a comprehensive
understanding of the importance, blockchain provides to the banking landscape.

7 Blockchain and Cloud Computing Applications

in Banking Sector

By leveraging the decentralized and immutable nature of blockchain, combined with

the agility and flexibility of cloud computing, banks can unlock new possibilities for
optimizing operations, improving customer experiences, and reducing operational
costs. This chapter explores how combining these two transformative technologies
can revolutionize traditional banking practices and address existing challenges in the
industry. Some of the use cases are highlighted below.
Cross-border payments: Blockchain offers cost-effective, secure, and faster ways
of exploring cross-border payments by eliminating intermediaries. For example,
Ripple, which is one of the blockchain-based payment networks, uses instant cross-
border payments between banks and financial institutions.
Interoperability and Collaboration: Cloud computing facilitates seamless inte-
gration and interoperability between different blockchain networks and banking
Revolutionizing Banking with Blockchain: Opportunities … 295

systems. Banks can leverage cloud-based platforms to connect with other finan-
cial institutions [12], enable cross-network transactions, and foster collaborative
initiatives within the industry.
Know Your Customer (KYC) compliance: Blockchain can streamline the KYC
compliance, a process in which a platform is given to provide a secure and immutable
ledger of customer data. Banks may also use blockchain to verify customer identities
and share customer data securely and efficiently across different institutions.
Trade finance: Blockchain can enhance trade and finance on a transparent and
secured platform for tracking trade transactions and reducing fraud. For example,
Marco Polo is a blockchain-based trade finance platform which tracks and identifies
trade transactions in real time.
Identity verification: Blockchain can enable secure and decentralized identity
verification, reducing the risk of identity theft and fraud. For example, uPort is a
platform designed on the blockchain that gives leverage to the users to control their
digital identities and share their identity data securely with different institutions.
Improved Data Management: Cloud computing offers robust data storage and
management capabilities, allowing banks to securely store and access large volumes
of blockchain data. This integration enables efficient data synchronization and repli-
cation across multiple nodes, ensuring the integrity and availability of blockchain
data.
Cost Optimization: The integration of cloud computing with blockchain allows
banks to optimize their infrastructure costs by leveraging the on-demand nature of
cloud resources. Banks can dynamically allocate computing resources based on their
needs, reducing operational expenses and improving cost-effectiveness.
Smart contracts: Self-executing smart contracts, which automate contract execu-
tion and eliminate the need for middlemen, can be created using blockchain
technology. For example, the blockchain-based platform Ethereum allows for the
development and execution of smart contracts.
For the most part, blockchain technology can change the banking sector by
improving efficiency, transparency, and security and by integrating cloud computing,
banks can unlock the full potential of this transformative technology, enabling
improved scalability, data management, productivity, cost optimization and parallel
performance for banking applications, through widespread deployment, and pre-
banking adoption.

8 Challenges of Blockchain Adoption in Banking Sector

The adoption of blockchain technology in the banking sector presents several chal-
lenges that hinder its widespread implementation. Some of the key challenges
include:
Regulatory compliance: Banks operate within a highly regulated environment, and
compliance requirements may vary across jurisdictions. Integrating cloud computing
with blockchain requires careful consideration of regulatory frameworks to ensure
296 S. Mahajan and M. Nanda

compliance with data protection, privacy, and financial regulations. Banks need
to navigate the complexities of compliance while leveraging the benefits of this
integration.
Interoperability: Achieving interoperability between different cloud platforms and
blockchain networks can be challenging. Banks may encounter compatibility issues
and difficulties in integrating diverse technologies and protocols. Standardization
efforts are crucial to establishing uniformity and seamless interaction between cloud-
based blockchain systems.
Security and Privacy: The combination of cloud computing and blockchain intro-
duces new security and privacy concerns. Banks need to ensure the confidentiality and
integrity of sensitive data stored in the cloud and transmitted within the blockchain
network. Robust encryption, access control mechanisms, and secure protocols must
be implemented to avoid risks associated with unauthorized access, data breaches,
and insider threats.
Technical complexity: Combining blockchain technology and cloud computing
with the present banking systems may be technically complex and challenging. Banks
need to ensure that their blockchain systems can integrate seamlessly with their
existing systems, databases, and applications.
Cost: Implementing it can be expensive, particularly for smaller banks and finan-
cial institutions. Banks need to make sure that the advantages of adopting blockchain
technology outweigh the costs, and that they have the required resources and expertise
to implement and maintain their blockchain systems.
Security: Blockchain technology is often considered to be more secure than tradi-
tional banking systems, but it is not immune to security threats. Banks need to ensure
that their blockchain systems are secure and that they have the right measures at time
to prevent hacking, data breaches, and other security threats.
Governance and Consensus Mechanisms: Cloud-based blockchain networks may
require modifications to the governance and consensus mechanisms to accommodate
the distributed nature of cloud computing. Consensus protocols and decision-making
processes need to be designed to ensure the integrity and trustworthiness of the
network while considering the involvement of multiple cloud nodes.
Overall, these challenges need to be resolved before using it widely and adopting
it in the banking sector. Banks need to work together with regulators, technology
providers, and other stakeholders to resolve these difficulties and realize the power
of using blockchain technology with cloud computing in the banking sector.

9 Methodology

This chapter presents an overview of the usage of the blockchain in the banking sector.
The aim of this work is to enhance understanding and knowledge of blockchain
technology in the banking industry with cloud computing. This will be achieved
by analyzing the existing research literature and comparing it with insights gained
from the people who were having their practical knowledge and experience in using
Revolutionizing Banking with Blockchain: Opportunities … 297

blockchain in banking. Furthermore, review papers, experiments, and case studies

conducted by other researchers have also been considered. Since blockchain tech-
nology has been founded recently for a decade, this study has a contemporary time
dimension. Nevertheless, as the study’s findings include information on the potential
challenges of utilizing blockchain technology, it also has a future-oriented dimension.
The data gathered for this work has been taken from multiple sources. Firstly,
data was collected from Wikipedia, which provided a foundational understanding of
the subject matter. Additionally, research conducted by other scholars was consulted
to gain insight into previous findings and knowledge on the topic. Furthermore,
interviews were conducted with individuals who were having hands-on knowledge
in both the banking and non-banking sectors to provide first-hand insights into the
use of blockchain in banking. The collection of data from various sources enhances
the comprehensiveness of the study’s findings and adds depth to the review analysis.

10 Usage of Blockchain in Different Sectors

This work provides an outline of how blockchain technology can provide signifi-
cant benefits to the banking industry. The use of technology in the financial sector
has the potential to provide several benefits while also posing a number of prob-
lems. According to the findings of this survey, one of the most extensively used
applications of blockchain technology is the ability to increase efficiency and lower
costs connected with traditional banking operations. As seen in Fig. 5, blockchain
technology has shown potential in a variety of areas because of its decentralized,
transparent, and secure nature. Blockchain technology can help banks run more effec-
tively and profitably by streamlining operations and enabling speedier transactions
at reduced prices.

Fig. 5 Usage of Blockchain in different Sectors

298 S. Mahajan and M. Nanda

Improved security is another important advantage of blockchain technology in

banking. Blockchain creates a transparent, decentralized record that is resistant to
cyber-attacks, making it an important tool for protecting financial transactions [13].
This technology can also increase transaction speed by lowering the time necessary
for verification and approval processes, resulting in faster and more smooth financial
transactions.
The study did, however, highlight several difficulties that have to be resolved
before this public ledger technology can be broadly implemented in the banking
industry. Among the most major issues are regulatory and legal problems, such
as compliance with data privacy regulations [14]. Furthermore, the necessity for
blockchain system standardization and interoperability is a substantial barrier to
widespread implementation [15, 16]. To enable smooth transactions and collabo-
ration among diverse entities, blockchain systems must be interoperable with one
another.
Finally, there is the issue of automation and its potential impact on employment in
the banking industry. As blockchain technology is adopted, some jobs may become
automated, which could lead to job losses for workers in certain areas of the industry.
Therefore, the application of blockchain technology in banking has the capacity to
transform the industry by augmenting productivity, enhancing security, and speeding
up financial transactions. However, careful consideration and planning are required
to overpower the obstacles while adopting it, including regulatory and legal concerns,
standardization, interoperability, and automation. With proper planning and execu-
tion, however, blockchain technique has the capacity to revolutionize the banking
industry, boosting productivity, integrity, and customer experience.
Blockchain technology offers several promising use cases for Indian banks,
addressing various challenges and improving efficiency and transparency as shown
in Table 1. Some of the notable use cases include.
Trade Finance and Supply Chain Management: Indian banks can leverage
blockchain to streamline trade finance processes and enhance supply chain manage-
ment [17]. By creating a distributed ledger, banks can facilitate secure and transparent
transactions between exporters, importers, and other stakeholders. This can reduce

Table 1 Indian banks using blockchain

Bank name Year (usage of Use case of blockchain
blockchain)
SBI Bank 2016 KYC compliance, supply chain financing, digital identity
verification
ICICI 2016 Domestic and International remittances
BANK
AXIS 2017 Cross-border remittances, trade finance platform
BANK
YES BANK 2018 Supply chain financing
Revolutionizing Banking with Blockchain: Opportunities … 299

paperwork, eliminate delays, and mitigate the risk of fraud, thereby improving the
overall efficiency of cross-border trade.
KYC and Identity Verification: Know Your Customer (KYC) processes can be
time-consuming and costly for banks. Blockchain-based identity solutions can enable
customers to maintain control over their personal data and provide verified informa-
tion only when required. This can accelerate onboarding processes, enhance data
privacy, and reduce the duplication of efforts for both banks and customers.
Remittances and Cross-Border Payments: Cross-border remittances are made
quicker and more affordable by blockchain technology, which lowers transaction
costs and does away with middlemen. Indian banks can use blockchain networks to
establish direct connections with international financial institutions, enhancing the
speed and affordability of remittance services for customers.
Digital Lending and Smart Contracts: Blockchain-based smart contracts can
automate and enforce lending agreements, making loan disbursals and repayments
more secure and efficient. Indian banks can use smart contracts to execute lending
contracts transparently and automatically trigger loan repayments based on prede-
fined conditions, reducing the need for manual intervention and minimizing default
risks.
Trade Settlements: Blockchain can be applied to simplify and expedite the settle-
ment process for securities and commodities trading. By using distributed ledger
technology, banks can achieve real-time settlement, reduce counterparty risk, and
improve liquidity management.
Fraud Detection and Prevention: Blockchain’s transparent nature and
immutability make it a valuable tool for fraud detection and prevention. Banks can
track and monitor transactions on the blockchain to identify suspicious activities and
enhance their security measures.
Loan Syndication and Debt Issuance: Blockchain can facilitate loan syndication
and debt issuance by creating a secure and auditable platform where multiple banks
can collaborate, verify loan data, and participate in loan syndication efficiently.
Loyalty Programs and Rewards: Indian banks can implement blockchain-based
loyalty programs, allowing customers to earn and redeem rewards across various
partners, providing a seamless and more valuable customer experience.
Central Bank Digital Currency (CBDC): The Reserve Bank of India (RBI) has
been exploring the possibility of issuing a central bank digital currency. Blockchain
technology can serve as the underlying infrastructure for a secure and efficient CBDC
system, enabling faster payments and reducing operational costs.
These use cases demonstrate how blockchain technology can revolutionize various
aspects of Indian banking, fostering financial inclusion, improving operational effi-
ciency, and enhancing customer experiences. However, successful implementation
requires collaboration among banks, regulators, and technology providers, along
with addressing challenges related to scalability, interoperability, and regulatory
compliance.
300 S. Mahajan and M. Nanda

10.1 Real-World Case Studies of Blockchain Implementation

in Banking

Ripple and Santander’s One Pay FX—Changes to Border Payments

Santander, one of the world’s largest banks, has embarked on a groundbreaking
journey to use blockchain technology to transform cross-border payments In part-
nership with Ripple, a blockchain technology company, Santander launched One Pay
FX, a blockchain-based platform aimed at increasing efficiency and speed.
Santander recognized the challenges its customers face when moving money
across borders, such as high transaction costs, long processing times, and lack of visi-
bility in the payment process and the bank sought to address these pains by finding
blockchain solutions, giving birth to One Pay FX. Ripple’s blockchain technology
provided the foundation for One Pay FX. After implementing Ripple’s XCurrent on
the platform, XRP, the native cryptocurrency of the Ripple network, enabled secure
and real-time cross-border transactions to be settled using blockchain for instant
liquidity and efficient exchange, eliminating the need for multiple intermediary
banks.
The implementation of One Pay FX has brought several key benefits to Santander
and its customers.
Faster transactions: Cross-border payments that previously took days can now be
completed in minutes, significantly reducing payment time [18].
Increased transparency: Customers gained real-time insight into the status and
progress of their services, increasing trust and confidence.
Lower costs: Bypassing traditional media banking, Santander reduced transaction
fees for its customers.
Improved customer experience: One Pay FX offers a simple and user-friendly
interface, which simplifies the process of initiating and tracking payments.
Following a successful One Pay FX trial, Santander expanded the service to other
countries and regions, providing improved cross-border payment capabilities to a
broader customer base that widely adopted the platform, attracting other financial
institutions and paving the way for consolidation and innovation in the banking
sector. There are various challenges when using One Pay FX. Considerable effort
and cooperation was needed to ensure compliance, safety standards, and greater
acceptance from other financial institutions.
The success of One Pay FX sent waves throughout the banking industry, prompting
other financial institutions to seek out blockchain solutions for their cross-border
payment systems The adoption of blockchain technology in banking grew as more
organizations realized the potential benefits. As blockchain technology continues to
evolve, Santander’s One Pay FX is a prime example of its potential to revolutionize
cross-border payments. The expansion and flexibility of the platform serve as a
benchmark for future blockchain applications in the banking industry [19].
Revolutionizing Banking with Blockchain: Opportunities … 301

Santander’s collaboration with Ripple and the subsequent launch of One Pay FX
demonstrates the real-world potential of blockchain technology to transform banking
systems, particularly in cross-border payments.
Komgo—Transforming Trade Finance with Blockchain
Komgo is a blockchain-based totally change finance platform that ambitions to
streamline and revolutionize the commodity trading industry [20]. Launched in
2018, it’s far a joint assignment between primary global banks, trading corporations,
and electricity agencies. Komgo makes use of blockchain generation to beautify
transparency, performance, and safety in trade finance operations. The commodity
trading enterprise has lengthy struggled with complicated, paper-extensive, and time-
consuming strategies. Komgo was born out of the need to cope with these challenges
by way of leveraging blockchain generation to create a greater efficient and trusted
atmosphere for exchange finance.
Komgo’s platform is constructed on a personal permissioned blockchain, ensuring
that the simplest legal individuals can get admission to and validate transactions. It
makes use of smart contracts to automate and streamline numerous exchange finance
methods, such as letters of credit score, virtual record sharing, and commodity trade
financing. There are several key benefits to the adoption of Komgo’s platform.
Enhanced Efficiency: By digitizing and automating approaches, Komgo reduced
the time wished for exchange finance transactions, resulting in quicker settlements
and reduced paperwork.
Improved Transparency: All stakeholders involved in the trade, including banks,
traders, and regulators, can access real-time data and verify transactions, increasing
trust and reducing disputes.
Fraud Prevention: The immutable nature of blockchain ensures that trade docu-
ments and contracts cannot be altered or tampered with, mitigating the risk of
fraud.
Cost Reduction: The elimination of manual processes and intermediaries led to
cost savings for participants in the trade finance ecosystem.
Komgo’s success can be attributed to cooperation with leading global players in
the commodity trading industry. Banks such as ABN AMRO, BNP Paribas, and ING
Group partnered with trading companies such as Shell and Mercuria to create a trusted
massive network in a platform where Komgo users faced compliance challenges, data
privacy, and integration with existing systems worked well with Mackay. The intro-
duction of Komgo generated excitement and excitement among various stakeholders
in the commercial economy. This demonstrated the potential of blockchain tech-
nology to transform traditional methods and paved the way for further adoption of
blockchain in the region.
As Komgo continues to evolve, it has the potential to become the industry stan-
dard for commercial finance. The success of the platform has encouraged financial
institutions and other commercial entities to explore similar blockchain solutions
to improve efficiency and transparency in their operations. Komgo serves as a real-
world case study of how blockchain-based trading and financial systems can trans-
form the commodity trading industry. Through long-term challenges and resulting
302 S. Mahajan and M. Nanda

insights and greater efficiencies, Komgo has demonstrated the transformative power
of blockchain technology in trade finance [21]. As the platform continues to evolve
and gain traction, it sets a precedent for other companies to explore blockchain
solutions for their specific trade finance needs.
Beyond these, there are a number of real-world banking case studies that have
adopted blockchain for improved efficiency and greater security. Some of them are
described below.
JPMorgan Chase—Interbank Information Network (IIN)
JPMorgan Chase, one of the world’s largest banks, launched the Interbank Informa-
tion Network (IIN) to address global payments challenges IIN on Quorum, a private
blockchain platform developed by JPMorgan, enabling faster resolution of compli-
ance and regulatory issues. As a result, international payments can be made faster,
benefiting both the bank and its customers.
BBVA—Blockchain-Based Syndicated Loan
BBVA, a Spanish multinational bank developed a pilot project to provide blockchain
technology for integrated lending. The bank partnered with various parties involved
in the syndication process, including BNP Paribas and MUFG, to use blockchain to
record and manage the loan agreement. The blockchain platform facilitated seam-
less communication and data sharing between all participants, reducing paperwork
and manual errors. The successful study demonstrated blockchain’s potential to
revolutionize the mortgage loan market by streamlining processes and improving
transparency [22].
Abu Dhabi Commercial Bank (ADCB)—ADCB collaborated with IBM to
broaden a blockchain-based trade finance platform [23]. The platform digitizes
and automates trade finance techniques, inclusive of letter of credit score issuance,
alternate document control, and agreement. By the use of blockchain generation,
the bank aimed to lessen processing time, decorate security, and enhance visibility
throughout the change finance environment. The a success implementation demon-
strated the capacity for blockchain to revolutionize trade finance operations and
growth performance for each financial institution and its customers.
Standard Chartered—Cross-Border Remittances
Standard Chartered, an outstanding global bank, partnered with Ripple, a blockchain-
based payments community, to broaden a move-border remittance solution. The
blockchain generation allowed for actual-time and value-effective fund transfers
among the bank’s branches in diverse international locations, significantly lowering
transaction instances and prices [24].
Emirates NBD—“Cheque Chain”: Emirates NBD, a leading financial institution
within the United Arab Emirates, carried out a blockchain-primarily based plat-
form referred to as “Cheque Chain” to enhance the safety and authenticity of its
issued cheques. The platform allows real-time verification of the cheque’s popularity,
lowering the danger of fraud and ensuring a clean clearing procedure [25].
Revolutionizing Banking with Blockchain: Opportunities … 303

These real-world case researches show the various programs of blockchain tech-
nology in the banking quarter, starting from global bills and change finance to
debt issuance and syndicated loans. As more monetary establishments discover and
embrace blockchain solutions, the industry maintains to witness the transformative
capacity of this era in improving efficiency, security, and customer revel.

11 Conclusion

This chapter delves into the ideas of blockchain generation, its layout, and the under-
lying theories that underpin its functionality, supplying a detailed evaluation of the
position of blockchain inside the economic zone. Generation additionally empha-
sizes using blockchain in the financial enterprise, demonstrating the massive potential
generation has to revolutionize setup banking practices.
This bankruptcy’s consciousness in the Indian banking quarter is one of its signifi-
cant contributions, with particular use of instances and illustrations of the way Indian
banks would possibly use the blockchain era to handle their unique problems and
necessities. The chapter emphasizes the importance and feasibility of blockchain
implementation for Indian banks via focusing on these software instances.
Furthermore, the subject of merging cloud computing with blockchain highlights
a vital feature of the chapter. This integration affords banks with strategic advantages
with the aid of helping them to conquer scalability and power intake constraints. By
leveraging cloud computing sources, banks can further decorate the overall perfor-
mance, flexibility, and computational capabilities of their blockchain-based systems.
As the Indian banking sector continues to evolve, the findings and recommenda-
tions presented here serve as a valuable guide for driving future developments and
innovations using blockchain technology.

References

1. Mahajan S (2022) Blockchain in smart healthcare systems: hope or despair? In: Borah M, Zhang
P, Deka G (eds) Prospects of blockchain technology for accelerating scientific advancement in
healthcare, pp 1–20
2. Adams J, Khan HTA, Raeside (2014) Research methods for business and social science
students, 2nd edn. Sage, New Delhi
3. Engert W, Fung BS (2017) Central bank digital currency: motivations and implications (No.
2017-16). Bank of Canada Staff Discussion
4. Shorman A, Sabri KE, Abushariah M, Qaimari (2020) Blockchain for banking systems:
opportunities and challenges. J Theor Appl Inf Technol 98(23)
5. Deloitte (2017) Blockchain in banking: a measured approach. 1–34
6. Sridharan R, Shankar R (2021) Blockchain technology in banking: a comprehensive review.
Int J Manag Technol Soc Sci (IJMTS) 6(2):27–39
7. Iqbal MA, Wasiq M (2021) Blockchain adoption in the banking industry: a review of challenges
and opportunities. J Risk Financ Manag 14(4):156
304 S. Mahajan and M. Nanda

8. Vali M, Madani SHH (2021) Blockchain technology in banking industry: a systematic review
of recent studies. J Bus Res 135:747–761
9. Yeoh W, Choi SY (2021) Understanding the potential of blockchain technology for financial
inclusion: a review of key opportunities and challenges. Int J Inf Manag 58
10. Heshmati M, Gupta RK (2021) Blockchain in banking sector: a systematic review of recent
studies. Int J Bank Market 39(3):516–531
11. Davis FD (1985) A technology acceptance model for empirically testing new end-user
information systems: theory results. Massachusetts Institute of Technology, Cambridge, pp
233–250
12. Chang V, Baudier P, Zhang H, Xu Q, Zhang J, Arami (2020) How blockchain can impact
financial services—the overview, challenges, and recommendations from expert interviewees.
Technol Forecast Soc Change 158
13. Garg P, Gupta B, Chauhan AK, Sivarajah U, Gupta S, Modgil S (2021) Measuring the perceived
benefits of implementing blockchain technology in the banking sector. Technol Forecast Soc
Change 163
14. Garg L, Choudhary P, Sharma P, Kaur H (2020) Exploring the advantages of using blockchain
technology in the banking sector. J Adv Manag Res 17(2):151–168
15. Cernian A, Paul AM (2020) Review and perspectives on how blockchain can disrupt education.
eLearn Softw Educ 3:45–50
16. Kalfoglou Y (2021) Blockchain for business: a practical guide for the next frontier. Routledge
17. Pichardo-López P, Alcaraz-Quintero OE (2020) Blockchain technique in the banking area. J
Ind Eng Manag 13(5):840–855
18. Chamria R (2023) Blockchain in cross-border payments: a game changer? Blockchain
Deployment and Management Platform, Zeeve
19. Brett C (2018) Santander One Pay FX, a blockchain-based international money transfer service.
Enterprise Times
20. Weerawarna R, Miah SJ, Shao X (2023) Emerging advances of blockchain technology in
finance: a content analysis. Pers Ubiquit Comput
21. komgo | Consensys, Consensys (2022). https://consensys.net/blockchain-use-cases/finance/
komgo
22. Geroni D (2023) A guide on quorum blockchain and their use cases. 101 Blockchains
23. Communications (2018) BBVA to test syndicated loans over blockchain. NEWS
BBVA. https://www.bbva.com/en/innovation/bbva-ready-negotiate-and-contract-syndicated-
loans-blockchain/
24. Jagtap J (2023) How ripple is shaping cross-border transactions in banking. The Crypto Times.
https://www.cryptotimes.io/how-ripple-is-shaping-cross-border-transactions
25. Admin (2023) Emirates NBD announces historic blockchain pilot project. Naseba. https://nas
eba.com/content-hub/article/emirates-nbd-announces-historic-blockchain-pilot-project/
Leveraging AI and Blockchain
for Enhanced IoT Cybersecurity

Iraq Ahmad Reshi and Sahil Sholla

Abstract This book chapter presents an overview of the cybersecurity concerns

of the Internet of Things (IoT). It investigates how artificial intelligence (AI) and
blockchain technologies address these challenges. This chapter describes the growing
number of accessories in IoT and the increasing sophistication of cyberattacks
targeting these devices. Each of these factors presents its own set of unique security
challenges. Furthermore, we are investigating the potential benefits of incorporating
AI and blockchain into IoT cybersecurity. These advantages include improved threat
detection and response, increased data privacy and integrity, and increased attack
resistance. Moreover, we present a review of particular novel approaches in the
field. This chapter presents brief case studies of AI and blockchain-based Internet of
Things cybersecurity solutions. These case studies show the practical applications
and benefits of these technologies in safeguarding Internet of Things environments.
The chapter provides insights into the changing landscape of cybersecurity for the
Internet of Things (IoT) and AI and blockchain’s role in mitigating cyber threats in
this sector.

Keywords Blockchain · Smart contracts · Internet of Things · Machine learning ·

Threat detection

1 Introduction

The “Internet of Things” (IoT) is a comprehensive framework encompassing a

sophisticated network of physical entities, such as diverse devices, autos, and house-
hold appliances. Integrating electronics, software, sensors, and networking capa-
bilities establishes this network. Illustrations of such entities encompass a diverse
array of contrivances, transportation means, and domestic apparatuses. The Internet
of Things (IoT) possesses significant potential to transform societal operations,

I. A. Reshi (B) · S. Sholla

Department of Computer Science and Engineering, Islamic University of Science and
Technology, Awantipora, Kashmir, J&K, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 305
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_14
306 I. A. Reshi and S. Sholla

including enhanced efficiency, streamlined automation, and unparalleled connec-

tivity. However, the Internet of Things (IoT) ecosystem faces much skepticism over
its capacity for revolutionary change. Academic researchers face the challenging
endeavor of establishing coherent regulations, standardised frameworks, and prac-
tical management approaches to guide the progression of the Internet of Things (IoT)
in a manner that fosters innovation and overcomes any barriers that may impede its
inherently dynamic characteristics. The proliferation of networked devices has led
to a significant surge in their overall quantity, presenting a diverse range of security
concerns. It is crucial to recognise that the aforementioned challenges profoundly
affect not only the security of our community but also the privileges we enjoy [11].

1.1 IoT Cybersecurity and Its Challenges

The enormous quantity and diverse range of Internet of Things (IoT) devices pose a
significant challenge in safeguarding their security. The absence of well-defined secu-
rity mechanisms, such as encryption, authentication, and authorisation, in Internet
of Things (IoT) devices renders them vulnerable to cyberattacks. Moreover, many
Internet of Things (IoT) devices primarily focus on cost-efficiency, resulting in poten-
tial limitations regarding essential resources such as memory and computational
power. Consequently, these constraints may hinder the implementation of advanced
security measures. The presence of diverse Internet of Things (IoT) devices and
protocols poses a significant challenge in establishing a cohesive security architec-
ture encompassing the entire IoT ecosystem. Wi-Fi, Bluetooth, and Zigbee represent
a subset of the communication protocols utilised by Internet of Things (IoT) devices.
Additionally, numerous devices incorporate proprietary protocols that still need to be
more explicit regarding widespread recognition. Consequently, security specialists
need help in developing universally effective security solutions.
Moreover, Internet of Things (IoT) devices present an enticing opportunity for
malicious actors because of their capability to collect and transmit sensitive personal
data, such as individual names, addresses, and credit card details. Hackers can facil-
itate the unauthorised infiltration of additional devices and data within a network by
utilising Internet of Things (IoT) devices as potential entry points [15].
In general, the issue of cybersecurity in the Internet of Things (IoT) presents a
complex and formidable challenge that necessitates a comprehensive and holistic
approach. It is imperative to prioritise the safeguarding of individual devices as
well as the establishment of comprehensive security protocols. Manufacturers must
prioritise security and incorporate safety considerations into the design of their goods
from the outset.
Leveraging AI and Blockchain for Enhanced IoT Cybersecurity 307

1.2 Importance of AI and Blockchain in IoT Cybersecurity

Both artificial intelligence (AI) and blockchain possess the capacity to significantly
enhance the security of the Internet of Things (IoT). The critical nature of real-
time identification and mitigation of cybersecurity vulnerabilities arises from the
continuous data collecting and transmission facilitated by Internet of Things (IoT)
devices. Algorithms grounded in machine learning can acquire the ability to identify
discernible patterns of behaviour that signify an imminent cyberattack, thus enabling
them to proactively or reactively implement precautionary or remedial measures. An
exemplary demonstration of the utilisation of artificial intelligence (AI) lies in the
automated activation of alarms or shutdowns as a response to dubious activities about
Internet of Things (IoT) devices. In contrast, blockchain technology is a decentralised
and immutable ledger that has the potential to enhance the security of Internet of
Things devices. Utilising blockchain technology enables the decentralised storage
of data, ensuring its immutability, and facilitates the transparent transfer of data
among Internet of Things (IoT) devices. Utilising blockchain technology enhances
the security of centralised systems or databases by rendering it more challenging for
hackers to tamper with or modify the stored data. As a result of the implemented
security measures, the likelihood of cyberattacks is reduced.
Furthermore, implementing blockchain technology can enhance the authentica-
tion and permission mechanisms for Internet of Things (IoT) devices. Information
and resources are restricted to authorised devices for confidentiality and security.
Any unauthorised device is restricted from accessing confidential information or
resources. Implementing blockchain-based identity management systems can effec-
tively establish secure authentication and authorisation protocols for IoT devices,
preventing unauthorised access.
Furthermore, using artificial intelligence (AI) and blockchain technology has
proven to be highly efficient in enhancing the security of the Internet of Things (IoT)
infrastructure. The utilisation of artificial intelligence for real-time detection and
response to cybersecurity threats, along with the implementation of blockchain tech-
nology for safe and decentralised record-keeping and identity management systems,
can significantly boost the security and reliability of IoT devices.

2 IoT Cybersecurity Challenges

The “Internet of Things” (IoT) refers to a network of interconnected physical objects

with built-in sensors, software, and other technologies, enabling them to collect and
exchange data. The Internet of Things (IoT) encompasses numerous benefits but
also introduces a range of cybersecurity vulnerabilities. Figure 1 illustrates the main
obstacles.
The susceptibility of numerous IoT devices to assaults is attributed to their
constrained processing and storage capacities. The susceptibility of these gadgets
308 I. A. Reshi and S. Sholla

Fig. 1 IoT cybersecurity challenges

arises from inadequate security measures, including the utilisation of weak pass-
words, the absence of protected communication channels, and the usage of outdated
software. The absence of a standardised framework for Internet of Things (IoT)
devices results in software and hardware configuration variations. The need for
standardisation poses challenges in implementing uniform security measures across
diverse devices. The collection and transmission of data by Internet of Things (IoT)
devices give rise to significant apprehensions over privacy and security. The potential
compromise of this information could have severe consequences. The governance
and monitoring of IoT networks pose significant challenges due to their dispersed
and decentralised character. This poses a challenge in promptly identifying and
addressing security concerns. The complex interconnectivity of the various devices,
networks, and platforms comprising an Internet of Things (IoT) ecosystem presents
a significant obstacle in ensuring their security. Malicious actors have the potential
to exploit any vulnerability within the ecosystem. A prevalent deficiency observed
in IoT devices is their incapability to receive firmware upgrades, rendering them
vulnerable to post-publication problems. This leaves the gadgets open to attacks for
as long as they stay in use [10]. Collaborative efforts among hardware and soft-
ware manufacturers, network and infrastructure service providers, and end users are
necessary to identify effective resolutions for these challenges. The deployment of
security measures, establishment of standards, and promotion of best practices will
Leveraging AI and Blockchain for Enhanced IoT Cybersecurity 309

be crucial in ensuring the security and privacy of IoT devices and the data they collect
[15].

2.1 Types of Cyber Threats in IoT

An increasing number of individuals use IoT (Internet of Things) devices in their

residential, professional, and critical public infrastructure settings. However, the
current landscape provides fraudsters with an increased scope to exploit vulnera-
bilities in Internet of Things (IoT) devices, enabling them to initiate a diverse range
of cyberattacks. The increased vulnerability is linked to the extensive distribution
of these devices. The following enumeration presents a limited selection of the
possible manifestations of cyberattacks against Internet of Things (IoT) devices.
The rapid expansion of Internet of Things (IoT) devices necessitates implementing
strong cybersecurity measures.
• Botnets, refer to networks of compromised devices under the control of an attacker
or a collective entity. Distributed denial-of-service (DDoS) attacks, initiated from
one of these devices, can disrupt large networks of computers that host critical
websites.
• Ransomware, A malicious software program designed to encrypt a user’s files
after that, demanding a monetary ransom in exchange for the decryption key
required to regain access to the contents. The infiltration of Internet of Things
(IoT) devices by ransomware can cause significant disruptions to commercial
operations and result in the theft of personal data.
• MITM, Man-in-the-middle attacks, or MITM attacks, manifest when an unau-
thorised individual interposes within the communication channel, connecting
two parties to manipulate or pilfer data. Man-in-the-middle (MITM) attacks can
compromise Internet of Things (IoT) devices in instances where these devices
engage in communication via networks that lack adequate security measures or
encryption protocols.
• Remote code execution, Remote code execution (RCE) attacks use vulnerabilities
in software to facilitate the execution of malicious code on a device. This technique
enables the perpetration of harmful activities on the targeted system. Internet of
Things (IoT) devices may be susceptible to Remote Code Execution (RCE) attacks
when they utilise outdated or unpatched software.
• Theft of private information, Internet of Things (IoT) devices can collect and
store sensitive personal information such as passwords, bank account numbers,
and health records. Cybercriminals can exploit vulnerabilities in these electronic
devices to illicitly get sensitive information.
• Physical attacks, In order to gain control over or extract data from an Internet of
Things (IoT) device, an attacker must first achieve physical proximity to the device
and subsequently implement modifications. In order to carry out this sort of attack,
the perpetrator must possess physical proximity to the targeted device. However,
310 I. A. Reshi and S. Sholla

it is noteworthy that detecting and mitigating such attacks poses considerable

challenges [4].

3 AI-Based Solutions for IoT Cybersecurity

IoT technology has revolutionised how everyday objects communicate, leading to a

seamless interconnectivity as never before. One potential drawback associated with
a more prominent online population is the increased vulnerability of individuals to
cybercriminal activities. The utilisation of artificial intelligence (AI) has the promise
of substantially mitigating vulnerabilities inside the Internet of Things (IoT).
Several AI-based precautions for the Internet of Things (IoT) are as follows:
• Anomaly Detection, In scenarios where an Internet of Things (IoT) device
exhibits signs of compromise or deviates from expected behaviour, artificial intel-
ligence (AI) can identify these anomalies and promptly notify the user through
an alarm system. Taking this step can help prevent potential cybersecurity threats
in advance.
• Predictive Analytics, Artificial intelligence (AI) can analyse substantial volumes
of data derived from Internet of Things (IoT) devices and detect apparent trends
that could signify a breach in security. The utilisation of predictive analytics has
the potential to facilitate the timely identification of potential hazards.
• Dynamic Risk Assessment, through the surveillance of device activity and iden-
tifying potential vulnerabilities, artificial intelligence (AI) can aid in analysing
security threats associated with Internet of Things (IoT) devices in a real-time
manner. To ensure safety, it is important to promptly identify and manage poten-
tial hazards. In Dynamic Risk Assessment, Artificial intelligence can facilitate
intelligence acquisition on potential cyber-attacks targeted towards Internet of
Things (IoT) devices. This can aid enterprises in maintaining a competitive edge
in the dynamic threat landscape and safeguarding their Internet of Things (IoT)
infrastructure.
• Automatic Reaction, Artificial intelligence (AI) can potentially optimise the
management of security events related to Internet of Things (IoT) devices. Imple-
menting this approach can enhance response times, mitigate the impact of attacks,
and bolster the overall defensive capabilities of the organisation.
When implemented on a broad scale, artificial intelligence (AI) holds the potential
to provide a comprehensive set of tools for addressing cybersecurity challenges in the
Internet of Things (IoT) domain. It is important to note that artificial intelligence (AI)
should not be considered a universal solution but rather a complementary tool that
should be used with other protective measures to mitigate cyberattacks successfully.
The following section provides a concise overview of specific machine learning
methods employed in Internet of Things (IoT) security.
Leveraging AI and Blockchain for Enhanced IoT Cybersecurity 311

3.1 Machine Learning Algorithms for Threat Detection

and Response

Machine learning algorithms have demonstrated remarkable efficacy in threat iden-

tification and response, owing to their ability to swiftly and accurately assess vast
amounts of data. Various types of machine learning algorithms, including supervised
learning, unsupervised learning, and reinforcement learning, are commonly utilised
in this domain.
Supervised learning methods can be employed to train models in identifying
distinct kinds of risks, leveraging annotated data. An example of a supervised learning
algorithm involves its training on a dataset consisting of verified instances of phishing
emails. Subsequently, this algorithm can detect and classify new phishing attempts
based on their resemblance to the previously known examples. Unsupervised learning
methods do not require prior knowledge, making them advantageous for identifying
patterns and anomalies within extensive datasets. These algorithms possess signifi-
cant use in threat detection due to their capacity to find novel and unidentified dangers
that may have eluded identification using alternative methodologies.
Reinforcement learning methods are employed in various threats and response
domains, including intrusion detection. These algorithms can undergo training to
respond to distinct categories of threats effectively, leveraging the feedback obtained
from the system over some time. An example of an application of reinforcement
learning involves training an algorithm to autonomously identify and isolate infected
machines, followed by the deployment of security patches as a response to a particular
type of malware. Furthermore, several specialised techniques and models, including
neural networks, decision trees, and support vector machines, are utilised in the threat
detection and response field in addition to the algorithms mentioned earlier. The
selection of an algorithm or technique is contingent upon several factors, including
the characteristics of the identified threat, the data accessible for analysis, and the
specific goals of the implementing organisation. The following are many instances
of frequently employed machine learning techniques utilised in the domain of threat
identification and response:
• Random forest, is a classification-centric supervised learning algorithm. Intru-
sion detection systems commonly identify malicious network traffic through this
technique. The technique creates numerous decision trees and aggregates their
outputs to ascertain a conclusive classification [7].
• K-means clustering, Comparable data points are grouped using this unsupervised
learning approach. Identifying data points that significantly deviate from the rest
of the dataset allows for detecting anomalies. The utilisation of this approach
holds potential in the identification and detection of previously unidentified or
novel threats [9].
• Deep learning, is a neural network commonly employed for intricate tasks like
image and speech recognition. Techniques like Deep packet inspection utilised in
threat detection, namely in analysing network data to identify malicious behaviour
originated from Deep Learning [13].
312 I. A. Reshi and S. Sholla

• Support Vector Machines (SVMs), This algorithm for supervised learning is

used for classification tasks. It operates by locating a hyperplane that divides
various classes of data points. SVMs have been utilised for malware detection by
classifying samples as benign or malicious based on their characteristics [18].
• Reinforcement learning, This algorithm acquires knowledge by reinforcing posi-
tive behaviour and penalising poor behaviour. The application of this approach
has been observed in the context of threat response, which entails the automated
deployment of security patches or the isolation of infected workstations as a
reactive measure to detect threats [20].
These are just a handful of the countless machine learning algorithms and
approaches used in threat identification and response. The selection of the algo-
rithm or technique to be employed will be contingent upon the characteristics of the
danger and the data at hand. Additionally, some intrusion prevention and detection
solutions are based on artificial intelligence.

3.2 AI-Based Intrusion Detection and Prevention

The use of machine learning algorithms for the detection and prevention of intru-
sions into Internet of Things (IoT) devices and networks is a relatively new area
of study. Because there are so many different types of devices on an IoT network,
it’s very difficult for humans to keep track of all of the activities on the network
and see any signs of intrusion. The use of AI for spotting and stopping intrusions is
therefore crucial. Anomaly detection, signature-based detection, behavioural anal-
ysis, and hybrid implementations are just some of the AI-based intrusion detection
and prevention strategies that can be used to the Internet of Things. In order to detect
potential attacks, these methods use machine learning algorithms to examine typical
network traffic, spot outliers, and compare incoming traffic to a library of known
attack signatures. Integrating AI-based intrusion detection and prevention systems
with preexisting security systems and protocols can add a new degree of defence for
the Internet of Things (IoT). As a whole, AI-based intrusion detection and preven-
tion is an encouraging step in protecting IoT systems from vulnerabilities like data
breaches, illegal access, and so on.

4 Blockchain-Based Solutions for IoT Cybersecurity

While the IoT has greatly improved our experience with technological devices, it
has also introduced novel cybersecurity threats. The limited processing power and
storage capacity inherent in many IoT device designs make them ideal targets for
cyber-attacks, which is a major problem. As a decentralised, immutable, and secure
architecture for IoT devices, blockchain technology has emerged as another possible
Leveraging AI and Blockchain for Enhanced IoT Cybersecurity 313

option to address these difficulties. Using blockchain in the Internet of Things is

significant since it ensures confidential data exchange between gadgets. To prevent
unauthorised changes or deletions, blockchain technology uses a distributed ledger to
record all financial transactions. So, it is more challenging for attackers to eavesdrop
on data transmissions or alter the data in any way. And because it eliminates the need
for third parties or centrally controlled agencies, blockchain technology makes data
sharing among IoT devices much safer [1].
Smart contracts, which are self-executing contracts where the conditions of the
agreements between the two parties are encoded directly into coding lines, are another
major advantage of blockchain in IoT. Smart contracts can be used to remove the
need for user intercession in transactions and exchanges by executing them automat-
ically and ensuring that only authorised devices can communicate with each other.
Also blockchain technology may be utilised to manage Internet of Things devices
in a safe and decentralised manner. Devices can be registered, authenticated, and
granted permission to communicate with other devices by establishing a blockchain-
based network. In this way, the network is protected from intruders and only autho-
rised devices are allowed access. In addition, blockchain technology offers a safe,
decentralised method of handling software patches and upgrades, which lessens the
likelihood of security flaws and exploits.
When applied to the security of Internet of Things (IoT) devices, blockchain tech-
nology’s decentralised, immutable, and secure architecture for communication, data
exchange, and device management might significantly change the current approach.
Blockchain-based solutions for IoT cybersecurity are anticipated to play a crucial
role in guaranteeing the security and privacy of our connected devices in the years
to come, despite the fact that there are still some problems to be addressed, such as
scalability and interoperability.
The authors [8] provide a thorough examination of significant security concerns
in the realm of the Internet of Things (IoT), specifically focusing on the layered
architecture. This survey involves the classification and examination of these diffi-
culties, with a particular emphasis on networking, communication, and management
protocols. This study aims to provide a comprehensive analysis of the fundamental
security requirements for the Internet of Things (IoT), with a particular focus on
examining existing risks, attacks, and the latest responses available. The research
effectively compares and contrasts the recognised security problems associated with
the Internet of Things (IoT) with the solutions that have been previously described
in academic literature. The examination of blockchain, the fundamental technology
behind Bitcoin, holds great importance in addressing many security challenges that
afflict the Internet of Things (IoT). The authors provide valuable perspectives on
emerging areas of research and significant obstacles that need to be addressed in
order to enhance the security of the Internet of Things (IoT).
A new and efficient blockchain (BC)-based architecture specifically built for the
Internet of Things (IoT) is proposed by Dorri et al. [5]. The framework aims to
reduce the conventional overheads associated with blockchain technology, while
yet preserving most of its security and privacy benefits. The suggested architec-
tural design presents a centralised, yet confidential, immutable ledger for Internet of
314 I. A. Reshi and S. Sholla

Things (IoT) devices. This ledger is inspired by blockchain ideas and is specifically
optimised to enhance energy efficiency. In order to ensure total end-to-end secu-
rity and privacy, a decentralised blockchain is deployed using an overlay network
on devices that possess sufficient power. By utilizing distributed trust methods, the
processing time for block validation is significantly decreased. The efficacy of this
methodology is demonstrated within the framework of a smart home environment,
acting as a sample exemplification for wider Internet of Things (IoT) implemen-
tations. The effectiveness of the architecture in providing security and privacy for
Internet of Things (IoT) use cases is emphasised through qualitative evaluations
conducted against existing threat models. In addition, simulations are used to verify
the efficacy of the suggested methodology, demonstrating significant decreases in
packet transmission and processing overhead compared to the blockchain structure
utilised in Bitcoin.
In this particular context, a resilient Proposed Application (PA) driven by
blockchain technology is envisioned with the objective of creating, maintaining, and
verifying healthcare certificates [17]. The PA functions as an intermediary conduit,
facilitating smooth communication between the foundational blockchain infrastruc-
ture and key entities within the application ecosystem, including hospitals, patients,
doctors, and Internet of Things (IoT) devices. The primary focus of its basic func-
tionality lies in the generation and verification of medical certifications. In addition,
the Public Administration (PA) demonstrates proficiency in implementing a variety
of essential security measures, including confidentiality, authentication, and access
control. These measures are effectively enforced through the integration of smart
contracts. The effectiveness of the suggested framework is shown through a rigorous
comparative and performance study, showcasing its superiority in comparison to
existing alternatives.
Another paper presents a novel architecture for sharing IoT data, known as TEE-
and-Blockchain-supported IoT Data Sharing (TEBDS) [21]. TEBDS combines on-
chain and off-chain methods to effectively fulfil the security requirements of the IoT
data sharing framework. The TEBDS framework utilises a consortium blockchain
to ensure the security of on-chain Internet of Things (IoT) data and manage access
controls for IoT users. In addition to this, an introduction is made to a Distributed
Storage System (SDSS) that utilises Intel SGX technology in order to enhance the
security of off-chain data. In addition, a meticulously designed incentive mecha-
nism has been formulated to promote the smooth functioning of the entire system. A
comprehensive security analysis confirms that TEBDS effectively meets the require-
ments for ensuring both data security and identity security. Empirical assessments
provide evidence supporting the effectiveness of TEBDS, demonstrating its improved
performance compared to the centralised SPDS strategy. Table 1, summarises the
works of blockchain for IoT security.
Leveraging AI and Blockchain for Enhanced IoT Cybersecurity 315

Table 1 Blockchain solutions for IoT

S. No. Author Technique
1 Ali et al. Proposed a blockchain-based architecture for IoT data security,
[1] enhancing confidentiality and integrity of data exchange, as well as
device management
2 Khan and Blockchain technology’s advantage in enabling automated and secure
Salah [8] smart contracts for IoT devices is highlighted, along with its potential in
managing device registration, authentication, and secure data sharing
3 Dorri et al. Introduce a novel and energy-efficient blockchain architecture for IoT
[5] devices, utilising both centralised and decentralised approaches to
enhance security and privacy
4 Sharma Present a proposed application (PA) utilising blockchain to generate and
et al. [17] verify healthcare certificates, incorporating robust security measures
through smart contracts
5 Xie et al. Introduced TEBDS, a hybrid on-chain and off-chain architecture utilising
[21] consortium blockchain and Intel SGX technology to secure IoT data
sharing

5 AI and Blockchain Integration for IoT Cybersecurity

The Internet of Things (IoT) could benefit from improved security measures if arti-
ficial intelligence (AI) and blockchain were coupled. Data collected by Internet of
Things (IoT) devices can be analysed by AI to reveal vulnerabilities. While tradi-
tional methods of data storage and dissemination have their limitations, blockchain
technology offers a safe, distributed alternative. Anomaly detection is one method in
which AI can strengthen the security of the Internet of Things. Algorithms powered
by AI may be taught to analyse data for anomalies that can indicate a breach in
security. This can be especially helpful in preventing and responding to attacks that
leverage Internet of Things (IoT) devices as vectors into bigger networks. Blockchain
technology has the potential to serve as a trustworthy, decentralised database for IoT
information. Blockchain’s use of a distributed ledger system makes data immutable
and resistant to hacking. This can safeguard information collected by IoT devices
and lessen the likelihood of data breaches.
Together, AI and blockchain can strengthen the IoT ecosystem’s defences and
make it more resistant to disruption. Artificial intelligence (AI) can be used to detect
security risks, setting off automated responses that employ blockchain technology to
protect and verify the integrity of the relevant data. This can aid in protecting the IoT
from threats and guaranteeing that only authorised parties have access to its data.
Artificial intelligence (AI) and blockchain technology (blockchain) could also be
used for identity management purposes in IoT cybersecurity. Artificial intelligence
can examine patterns of use and flag outliers that may suggest intrusion. To further
ensure that only authorised users have access to IoT devices and data, blockchain can
be used to securely store user IDs and authentication data. Several other applications
exist for combining AI and blockchain technology in IoT security.
316 I. A. Reshi and S. Sholla

• Fraud detection, For the purpose of detecting fraudulent behaviours, such as data
manipulation or illegal access, artificial intelligence algorithms can be applied to
IoT data in real time. The immutable recordings of these actions can then be stored
in blockchain, providing a safe and verifiable audit trail.
• Threat intelligence sharing, To better understand cyberattack patterns and trends,
threat intelligence data from many sources can be analysed by AI. Secure data
sharing via a blockchain-based platform makes it possible for businesses to work
together to combat security concerns.
• Device authentication, To verify the legitimacy of a device, we can utilise artifi-
cial intelligence to monitor its activity for any deviations that would point to the
presence of a malicious piece of hardware. The device’s authenticity can then be
verified via blockchain, allowing only approved gadgets access to the network.
• Safety of smart contract, Smart contracts are blockchain-stored, automatically-
executing contracts, commonly used to interface IoT in Blockchain. Artificial
intelligence (AI) can inspect these agreements for loopholes and other security
problems. This can protect smart contracts from attacks that make use of security
flaws.
• Smart contract security, With AI and blockchain technology, supply chains may
be made more secure and transparent by keeping track of items as they flow from
supplier to customer. This has the potential to deter forgery, tampering, and other
fraudulent activities.
When combined, AI and blockchain have the potential to greatly strengthen the
reliability and safety of IoT infrastructure. The Internet of Things (IoT) ecosystem
can be made safer from cyberattacks by integrating these two technologies.

5.1 State-of-Art Survey of Combination of AI

and Blockchain for IoT Security

Integrating Blockchain and AI facilitates the resolution of issues about accuracy,

latency, centralisation, and, most importantly, security in IoT architectures. The
database within blockchain technology is responsible for storing transactions accom-
panied by a digitally signed hash value. Hence, it is imperative to address accuracy,
latency, security, privacy, and centralisation concerns to mitigate potential challenges.
Artificial intelligence (AI) algorithms are also employed to address these challenges.
Utilising blockchain networks offers the advantage of decentralisation, which facil-
itates automatic and efficient data validation. This feature effectively addresses the
issue of single points of failure that commonly arise in cloud servers utilised for
extensive data analysis.
Furthermore, integrating blockchain technology with the Internet of Things (IoT)
and artificial intelligence (AI) provides further benefits in enhancing the capabilities
of IoT systems. Rathore and Park [14] present a security architecture for Internet
of Things (IoT) networks. The architecture aims to ensure the secure and scalable
Leveraging AI and Blockchain for Enhanced IoT Cybersecurity 317

transmission of IoT data from decentralised IoT applications at the fog layer. Arti-
ficial intelligence (AI) is employed in diverse domains of advanced technologies,
including blockchain thinking, decentralised AI, the intelligence of things, and intel-
ligent robots, among others, in the daily lives of individuals [15]. The convergence
between artificial intelligence (AI) Internet of Things (IoT) enables the collection
of a vast amount of data and facilitates its analysis. Machine learning is utilised in
various domains, including healthcare, smart home technology, smart farming, and
intelligent vehicles, among others, to facilitate effective learning processes. Rathore
and Park [14] introduced a novel approach that utilises blockchain technology to
enhance the security of deep learning in the context of Internet of Things (IoT)
applications. By integrating Blockchain and artificial intelligence (AI) at the device
layer, their suggested method aims to ensure the integrity and reliability of data in IoT
systems. The system demonstrates a significant level of precision and a notable delay
in processing time for Internet of Things (IoT) data. In their study, Gil et al. [6] exam-
ined the role of intelligent machines in several domains, including medical science,
automatic sensing devices, automated vehicle driving, and cooking, to reduce human
labor. Intelligence can be defined as the cognitive capacity to use acquired knowledge
to address intricate challenges. In contrast, artificial intelligence (AI) refers to the
learning approach that facilitates the development of innovative procedures and the
dissemination of collected initial insights. A report by McKinsey [3] projected that
the AI market will experience significant growth, reaching a value of 13 trillion US
dollars by 2030. The decentralised AI approach is a fusion of artificial intelligence
(AI) and blockchain technology. Its purpose is to facilitate secure and trustworthy
information sharing without relying on intermediaries. This is achieved through the
utilisation of cryptographic signatures and robust security measures.
Furthermore, it can autonomously make decisions in Internet of Things (IoT)
applications. In recent years, the rapid evolution of technologies, devices, and Internet
of Things (IoT) devices has resulted in Blockchain, Artificial Intelligence (AI),
and IoT emerging as the most influential technologies, driving the acceleration of
innovative ideas across several domains.
Researchers [19] examine privacy, accuracy, latency, and centralisation concerns
in integrating Blockchain and AI technologies within Internet of Things (IoT) appli-
cations. Blockchain and AI are integrated to propose an Intelligent IoT Architecture
incorporating Blockchain technology, named as block intelligence. This architecture
consists of a decentralised cloud infrastructure enabled by Blockchain at the cloud
layer, distributed fog networks based on Blockchain at the fog layer, distributed
edge networks based on Blockchain at the edge layer, and the convergence of peer-
to-peer Blockchain networks at the device layer. Study by authors [16] introduces a
novel security model called the Artificial Intelligence-based Lightweight Blockchain
Security Model (AILBSM) to improve privacy and security in Industrial Internet of
Things (IIoT) systems based on cloud computing. The framework utilises a combi-
nation of lightweight blockchain technology and a Convivial Optimized Sprinter
Neural Network (COSNN) based AI mechanism. This integration enables the frame-
work to deploy an Authentic Intrinsic Analysis (AIA) model, effectively converting
features into encoded data. As a result, the framework mitigates the potential impact
318 I. A. Reshi and S. Sholla

of attacks. By conducting extensive experiments and assessments, the AILBSM

framework demonstrates notable outcomes, such as a decreased execution time of
0.6 s, an improved classification accuracy of 99.8\%, and an enhanced detection
performance of 99.7%. The superior anomaly detection capability of the model can
be attributed to its new features, such as auto-encoder-based transformation and
blockchain authentication. This represents a noteworthy improvement in Industrial
Internet of Things (IIoT) security, distinguishing it from other existing methodolo-
gies. Another study [22] presents a novel framework for cloud-edge-end architecture
that utilises blockchain technology as a foundation for a trust mechanism based on the
principles of blockchain consensus. The system utilises a consensus approach known
as BLS-based proof of replication (PoRep) to establish confidence among devices.
The data from devices is securely maintained within servers, utilising mechanisms
like proof generation and consensus establishment by broadcasting information by
nodes. This methodology improves the level of network data security and fosters
the establishment of trust across a wide range of devices. In addition, the solu-
tion integrates verifiable delay functions (VDF) based on secret sharing to mitigate
the need for servers to perform dynamic data replication calculations in Proof of
Replication (PoRep), enhancing consensus efficiency among nodes. Implementing
this complete strategy represents a notable progression in establishing secure and
reliable interactions inside the network.
The authors [12] introduced a highly advanced intrusion detection system
designed for smart cities. The system consists of three essential modules:
• A trust module incorporating a reputation system based on blockchain technology
and addresses
• A privacy module with two tiers employing an improved Proof of Work technique
based on blockchain
• An intrusion detection module
This study presents an integrated infrastructure called Cloud-Block, Fog-Block,
and Edge-Block, explicitly designed for smart city environments. This infrastructure
aims to optimise the Edge-Fog-Cloud architecture by capitalising on its advantages
and mitigating its constraints. The system’s effectiveness is demonstrated through
the evaluation conducted on the sBoT-IoT and TON-IoT datasets. A comparative
comparison with other state-of-the-art systems shows that the suggested system
surpasses its equivalents, emphasising its notable progress in the field of smart city
cybersecurity.
Furthermore, the study in [2] explores the intricacies of reconciling various tech-
nologies and suggests practical solutions to address these challenges. It provides a
complete analysis of the prospects of this amalgamation, emphasising its potential to
bring about significant changes in various areas, including banking, healthcare, and
transportation. The study also investigates existing models’ pre-integration limita-
tions, including problems such as lack of transparency, concerns over data privacy and
security, and deficiencies in automation. The comprehensive examination of the inte-
gration of blockchain, artificial intelligence, and IoT not only highlights their promise
Leveraging AI and Blockchain for Enhanced IoT Cybersecurity 319

Table 2 Summary of state-of-art proposed solutions

S. No. Year Author Main theme of paper
1 2016 Gil et al. [6] Role of intelligent machines across domains, defining
intelligence and AI’s role in developing innovative
procedures
2 2018 Rathore and Security architecture for IoT networks, utilising AI and
Park [14] blockchain to ensure secure and scalable IoT data
transmission
3 2020 Singh et al. Privacy, accuracy, latency concerns in integrating Blockchain
[19] and AI, proposing an Intelligent IoT Architecture
incorporating Blockchain
4 2023 Selvarajan Introduction of an AI-based Lightweight Blockchain
et al. [16] Security Model (AILBSM) for IIoT security enhancement,
utilising blockchain and AI mechanisms
5 2023 Zhang et al. Novel cloud-edge-end architecture using blockchain-based
[23] trust mechanism, integrating BLS-based PoRep consensus,
enhancing network security and trust
6 2023 Moawad et al. Intrusion detection system for smart cities using
[12] blockchain-based trust, privacy, and intrusion modules,
optimising smart city cybersecurity
7 2023 Aruna et al. Exploration of integrating blockchain, AI, and IoT,
[2] discussing potential changes and challenges in various
domains

but also recognises the significant challenges that prompted this convergence. Table 2
summarises the literature.

5.2 Challenges and Limitations of AI and Blockchain

Integration for IoT Security

Integrating AI with blockchain for IoT security offers the ability to address some
of the most pressing security issues with the IoT, including data breaches, device
tampering, and unauthorised access. There are, however, a number of problems and
restrictions with this integration, as well as some research areas that should be covered
up.
• Scalability: One of the challenges encountered in the implementation of IoT
security through the use of AI and blockchain is the issue of scalability. The volume
of data generated by Internet of Things (IoT) devices is substantial, necessitating
significant computational resources for analysis. Integrating artificial intelligence
(AI) and blockchain technology may result in a substantial increase in processing
requirements, thus impeding the system’s scalability.
320 I. A. Reshi and S. Sholla

• Cost: The cost of the integration of artificial intelligence (AI) and blockchain tech-
nology has the potential to enhance the security of the Internet of Things (IoT).
However, this advancement has its associated costs. Both blockchain technology
and AI algorithms require significant computational resources. Consequently,
the cost associated with implementing this system may increase, rendering it
financially unattainable for many smaller enterprises.
• Complexity: Integrating artificial intelligence (AI) and blockchain technology
to enhance security measures for Internet of Things (IoT) devices may present
specific challenges. Integrating both systems necessitates proficiently under-
standing their potential applications in safeguarding Internet of Things (IoT)
devices. Implementing this system may pose challenges for certain businesses
due to its inherent complexity.
• Interoperability: Interoperability is the ability of different systems or components
to exchange and use information. Interoperability poses an additional challenge
in utilising artificial intelligence (AI) and blockchain technologies to enhance
security in the Internet of Things (IoT). Integrating diverse IoT devices into a
unified system can present difficulties when these devices operate on disparate
protocols and standards.

5.3 Research Gaps

In order to improve interoperability and compatibility across devices and systems,

standardisation is necessary for the integration of AI and blockchain for IoT security.
To further ensure the security of personal information submitted by users retained
on IoT devices, privacy-preserving algorithms must be developed. In addition to
integrity models, trust models are crucial to the security of the system. In addition,
ensuring the IoT system and its data are secure from cyberattackers relies heavily on
the governance of AI and blockchain integration. Despite the possible advantages,
there are also restrictions and difficulties to overcome, such as scalability, complexity,
and expense. Improving the efficacy and safety of AI and blockchain integration for
IoT security requires addressing research gaps in areas like standardisation, privacy,
trust, and regulations.

6 Case Studies of AI and Blockchain-Based IoT

Cybersecurity Systems

In this section, we will brief about few use cases in the field of AI, IoT and blockchain,
and how one technology can benefit the other.
Leveraging AI and Blockchain for Enhanced IoT Cybersecurity 321

Case study 1

AI-based threat detection in industrial IoT

A manufacturing facility could be one use case for AI-based threat detection in the
industrial Internet of Things. Various instruments and devices in a typical manufac-
turing facility are network-connected and communicate data with one another and
the central control system. This network of devices, known as the Industrial Internet
of Things (IIoT), is responsible for maintaining the efficiency of the manufacturing
process. Nonetheless, the IIoT is a primary target for cyber attacks, and any network
disruption can result in production delays, outages, and financial losses. Here, AI-
based threat detection can be helpful. Using machine learning algorithms to analyse
network traffic, AI can identify peculiar patterns and anomalies that may be indicative
of a cyber attack. AI can detect, for instance, if a machine sends and receives data at
an abnormal rate or if it communicates with an unauthorised device. Once a threat has
been identified, the AI system can mitigate it by alerting the security team, blocking
suspicious traffic, or closing down the afflicted machine. In addition, AI-based threat
detection can assist with countermeasures, such as detecting network vulnerabilities
and identifying potential weak spots before they are exploited by cyber attackers.
Overall, AI-based threat detection in industrial IoT can aid in ensuring the safety
and security of the manufacturing facility, preventing cyber attacks and production
process disruptions, and saving the organisation time and money.
Case study 2

Blockchain-based secure data sharing in healthcare IoT

Another use case can be, Blockchain-based secure data sharing in healthcare IoT
wishing to exchange patient information with other facilities in a safe and reliable
manner. Sharing patient data across healthcare providers has historically been a
challenging and tedious procedure due to the need to safeguard the data’s accuracy,
security, and compliance with standards like HIPAA and GDPR. To remedy this, we
may utilise blockchain technology to safely share information. Healthcare providers
may build a trustworthy, distributed network for exchanging patient information
using blockchain technology. The blockchain stores an immutable and comprehen-
sive record of patient information, and each healthcare practitioner may keep their
own copy. The blockchain network may be queried by healthcare providers whenever
they need to access patient data, at which point the network will confirm the provider’s
identity and provide them access. Since the data is maintained in a decentralised and
tamper-proof way, it protects the privacy of patients and allows for the free flow
of information among healthcare providers without compromising any of the data’s
integrity. Data interoperability can be aided by blockchain-based secure data sharing
since it lets healthcare professionals exchange patient information between disparate
systems and platforms without jeopardising patient privacy or security. This has the
potential to enhance patient outcomes, cut down on wasted resources, and lower
healthcare costs.
322 I. A. Reshi and S. Sholla

Overall, blockchain-based safe data sharing in healthcare IoT can aid in protecting
patients’ privacy, keeping their data private, and facilitating the secure transfer of data
across various medical facilities.
Case study 3

AI and blockchain-based identity management in smart homes

A smart apartment complex is another potential use of artificial intelligence and
blockchain-based identity management in the context of smart housing. Thermostats,
cameras, and house assistants are just some of the internet-connected equipment that
may be remotely managed in a typical “smart home.” This, however, makes them
susceptible to cyber assaults and unauthorised access, which in turn threatens the
inhabitants’ privacy and safety. Artificial intelligence and identity management using
blockchain technology can assist here. AI may monitor the locals’ actions for signs
of suspicious or out-of-the-ordinary behaviour using machine learning methods. A
cyber assault may be in progress if, for instance, a smart thermostat is repeatedly
toggled on and off in a short amount of time. A decentralised and secure network
for managing identification and access control may be built using blockchain tech-
nology to protect people’s privacy and safety. Every citizen has the option of keeping
their digital identity on the blockchain, which stores an immutable record of their
personal information and permits. When a user needs to access a smart home device,
they may do so by sending a request to the blockchain network, which will then
confirm the user’s identity and provide them access. Because the information is
kept in a distributed and unalterable fashion, residents may be certain that their
personal information will remain private. Moreover, AI and blockchain-based iden-
tity management may aid in preventative measures like detecting weaknesses in the
smart home network and pinpointing possible points of attack in advance.
Overall, identity management in smart homes based on AI and blockchain tech-
nology can help protect the personal information of residents and keep their smart
home gadgets accessible in a safe and reliable manner.

7 Conclusion

This chapter has presented an overview of the cybersecurity threats encountered by

the Internet of Things (IoT) and the potential solutions offered by artificial intel-
ligence (AI) and blockchain technologies. Recent vulnerabilities and cyber threats
have emerged as a result of the growing number of connected devices in the IoT
ecosystem, and they may be beyond the capabilities of more conventional secu-
rity measures. IoT cybersecurity may be made more effective and efficient with
the use of AI-based solutions like machine learning algorithms for threat detection
and response and intrusion detection and prevention systems. Similarly, blockchain
technology enables a decentralised and immutable ledger for IoT devices, which
improves data privacy, integrity, and resilience to assaults. Case studies given in this
Leveraging AI and Blockchain for Enhanced IoT Cybersecurity 323

chapter illustrate the actual uses and benefits of integrating AI and blockchain tech-
nology to further increase the security of IoT settings. There are, however, obstacles
and restrictions to combining AI and blockchain for IoT safety. The requirement
for standardisation, scalability, and interoperability are all examples. It is crucial
to tackle these issues and provide solid solutions that can properly safeguard these
systems as the IoT ecosystem continues to grow. In conclusion, this chapter has
demonstrated the great potential of AI and blockchain technology in the context of
IoT cybersecurity. Better threat detection and response, more secure data storage,
and a more secure Internet of Things (IoT) may all be attained through the use of
these technologies.

References

1. Ali MS, Vecchio M, Pincheira M, Dolui K, Antonelli F, Rehmani MH (2018) Applications

of blockchains in the Internet of Things: a comprehensive survey. IEEE Commun Surv Tutor
21(2):1676–1717
2. Aruna S, Priya SM, Reshmeetha K, Sudhayini ES, Narayanan AA (2023) Blockchain integra-
tion with artificial intelligence and Internet of Things technologies. In: 2023 7th International
conference on intelligent computing and control systems (ICICCS), pp 688–694
3. Chui M et al (2023) The economic potential of generative AI: the next productivity frontier.
McKinsey & Company
4. Deogirikar J, Vidhate A (2017) Security attacks in IoT: a survey. In: 2017 International
conference on I-SMAC (IoT in social, mobile, analytics and cloud) (I-SMAC), pp 32–37
5. Dorri A, Kanhere SS, Jurdak R (2017) Towards an optimized blockchain for IoT. In: Proceed-
ings of the second international conference on internet-of-things design and implementation,
pp 173–178
6. Gil D, Ferrández A, Mora-Mora H, Peral J (2016) Internet of things: a review of surveys based
on context aware intelligent services. Sensors 16(7):1069
7. Islam N, Farhin F, Sultana I, Kaiser MS, Rahman MS, Mahmud M, Hosen AS, Cho GH (2021)
Towards machine learning based intrusion detection in IoT networks. Comput Mater Continua
69(2):1801–1821
8. Khan MA, Salah K (2018) IoT security: review, blockchain solutions, and open challenges.
Futur Gener Comput Syst 82:395–411
9. Liu C, Gu Z, Wang J (2021) A hybrid intrusion detection system based on scalable K-Means+
random forest and deep learning. IEEE Access 9:75729–75740
10. Lu Y, Da Xu L (2018) Internet of Things (IoT) cybersecurity research: a review of current
research topics. IEEE Internet Things J 6(2):2103–2115
11. Maple C (2017) Security and privacy in the internet of things. J Cyber Policy 2(2):155–184
12. Moawad MM, Madbouly MM, Guirguis SK (2023) Leveraging blockchain and machine
learning to improve IoT security for smart cities. In: The international conference on artificial
intelligence and computer vision, pp 216–228
13. Mohammadi M, Al-Fuqaha A, Sorour S, Guizani M (2018) Deep learning for IoT big data and
streaming analytics: a survey. IEEE Commun Surv Tutor 20(4):2923–2960
14. Rathore S, Park JH (2018) Semi-supervised learning based distributed attack detection
framework for IoT. Appl Soft Comput 72:79–89
15. Reshi IA, Sholla S (2022) Challenges for security in IoT, emerging solutions, and research
directions. Int J Comput Digit Syst 12(1):1231–1241
16. Selvarajan S, Srivastava G, Khadidos AO, Khadidos AO, Baza M, Alshehri A, Lin JC-W (2023)
An artificial intelligence lightweight blockchain security model for security and privacy in IIoT
systems. J Cloud Comput 12(1):38
324 I. A. Reshi and S. Sholla

17. Sharma P, Namasudra S, Crespo RG, Parra-Fuente J, Trivedi MC (2023) EHDHE:

enhancing security of healthcare documents in IoT-enabled digital healthcare ecosystems using
blockchain. Inf Sci 629:703–718
18. Shen M, Tang X, Zhu L, Du X, Guizani M (2019) Privacy-preserving support vector machine
training over blockchain-based encrypted IoT data in smart cities. IEEE Internet Things J
6(5):7702–7712
19. Singh SK, Rathore S, Park JH (2020) Blockiotintelligence: a blockchain-enabled intelligent
IoT architecture with artificial intelligence. Futur Gener Comput Syst 110:721–743
20. Uprety A, Rawat DB (2020) Reinforcement learning for iot security: a comprehensive survey.
IEEE Internet Things J 8(11):8693–8706
21. Xie H, Zheng J, He T, Wei S, Hu C (2023) TEBDS: a trusted execution environment-and-
blockchain-supported IoT data sharing system. Futur Gener Comput Syst 140:321–330
22. Zhang A, Lin X (2018) Towards secure and privacy-preserving data sharing in e-health systems
via consortium blockchain. J Med Syst 42(8):140
23. Zhang F, Wang H, Zhou L, Xu D, Liu L (2023) A blockchain-based security and trust mechanism
for AI-enabled IIoT systems. Futur Gener Comput Syst 146:78–85
Dissecting the Role of Women
in Cybersecurity and Information
Technology: A Medical Perspective

Wasswa Shafik

Abstract As the global population is working toward the United Nations goals of
sustainable communities and cities, good health and well-being, and responsible
consumption and production, recognizing the crucial role women play in various
humanitarian activities, it is essential to note that, despite the success of highlighting
accomplished women in technology-based organizations, recent research indicates
their underrepresentation in different technology-related fields. Despite efforts to
attract and retain women, cultural issues within this diverse industry and global
barriers pose significant challenges. Leaders of all genders and backgrounds are
essential to addressing these workforce development gaps and formulating effective
company strategies. This study delves into the hurdles, difficulties, and innovative
ways to increase the number of women in senior positions in healthcare cyberse-
curity (CS) and information technology (IT). It further examines the nature of CS
and IT, their importance, global internet utilization, essential forms, and profes-
sional aspects related to the underrepresentation of women in the industry. The low
number of women in these fields can be attributed to several factors, as demonstrated
within this study, and possible solutions are proposed. As the health industry aims for
inclusive and innovative solutions, women bring unique perspectives and problem-
solving skills, which can enhance the development of secure medical systems, elec-
tronic health records, and patient privacy protection. Finally, it is emphasized that
addressing these raised factors will significantly enhance organizational technology
acceptance, cybersecurity trust, effectiveness, and efficiency.

Keywords Good health and well-being · Sustainable cities and communities ·

Responsible consumption and production · Women · Healthcare · Cybersecurity ·
Information technology

W. Shafik (B)
School of Digital Science, Universiti Brunei Darussalam, Jalan Tungku Link, Gadong, 1410
Bandar Seri Begawan, Brunei
e-mail: [email protected]
Dig Connectivity Research Laboratory (DCRLab), 600040 Kampala, Uganda

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 325
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_15
326 W. Shafik

1 Introduction

Despite making substantial progress in the workplace over the past half-century,
women continue to be grossly underrepresented in various scientific and technical
professions. Even though females constitute approximately 47%, 27%, 31%, and
27% of the labor force, environmental science occupations, chemistry jobs, and
computer science and mathematics jobs are held by women, respectively. This gender
gap persists despite the advances made by women in the employment market [1].
Many women face unique challenges in balancing work and family responsibili-
ties. Offering flexible work arrangements, such as telecommuting or flexible hours,
can help women balance their work and personal lives, which may make technical
professions more appealing. Academics find it incredibly challenging to comprehend
why female representation is lacking in these professions. This is concerning From
a policy standpoint as it recommends that the government’s technical personnel may
not utilize all available creative energy [2]. This is problematic because it implies
that creative potential is being squandered, indicating a need for improvement in the
technical workforce across the nation.
The reasons behind the slow progress of women in entering healthcare, scientific,
mathematical, engineering, and technological fields remain uncertain and, in some
cases, debatable. The contentious nature of this issue is highlighted by the heated
debate that followed Harvard President Larry Summers’ remarks at a conference in
2005. Summers’ suggestion that transformations in the distribution of talent between
men and women may contribute to the underrepresentation of females in top-level
science positions sparked this controversy [3].
If this perspective is correct, removing these barriers would result in an equal
number of men and women entering technical jobs. Alternatively, it could be that
men and women face similar challenges, but the distribution of skills across genders
may differ, making men more productive in technical domains. This could be due to
differences in intelligence quotient between males and females, as Larry Summers
has stated. From this perspective, the employment gap between males and females
could reflect the effective utilization of talent across various fields [4].
One potential explanation for the low percentage of women in technical profes-
sions could be differing perceptions of an appealing career path between men and
women. Men may place a higher value on technical employment, which could result
in a competitive market response among individuals with different job preferences
[5]. Evidence to support this notion can be found in the declining percentage of
women in technical occupations over the past few decades. The neoclassical approach
posits that employees evaluate a particular career choice’s predicted earnings and
non-financial returns and will only change jobs if the cost–benefit analysis indicates
it is worthwhile [6].
The theoretical framework of Bowles, Gintis, and Osborne is consistent with the
notion that gender disparities in occupational choice result from distinct preferences.
This framework can be interpreted in a way that aligns with their findings and explains
the presence of significant wage disparities between individuals at any given time
Dissecting the Role of Women in Cybersecurity and Information … 327

caused by disequilibrium rent resulting from slow responsiveness to market tremors

[7]. From this perspective, the association between an individual’s earnings and
various non-productive personal qualities reflects their willingness or capacity to
reply to market shocks that create uncertainty rents. In other words, the correlation
between a person’s earnings and personal traits reflects their abilities. These market
shocks lead to imbalanced rents [8].
Creating an empirical method to differentiate between these three alternatives
would be arduous. However, through our ongoing case study of the IT industry, we
aim to contribute to the ongoing debate uniquely. Our research suggests that the
gender gap in IT is mainly due to variations in professional interests between men
and women. To make occupational choices, workers evaluate the benefits of specific
employment options, including predicted earnings and non-pecuniary returns [9]. For
example, it has been suggested that women may prefer certain occupations because of
a reduced risk of income decline due to anticipated absences, and they may prioritize
minimizing potential conflicts between their professional and domestic commitments
more than men do. This indicates that women prioritize their families more than men.
CS entails identifying (demonstrating asset management), protecting (entails
access control, data security, email, internet, computers, and mobile, network, main-
tenance, awareness, and training), detecting (including email monitoring, internet
monitoring, computer monitoring, and network monitor) respond (counting plan,
analyze or investigate, and manage change) and recover (preparation and restora-
tion) as demonstrated in Fig. 1. An analysis, which includes a set of occupational
personality traits, shows that gender is no longer a significant factor in predicting
the choice between professions in information technology and non-information tech-
nology careers. This suggests that the underrepresentation of women in the informa-
tion technology field results from individuals’ decisions based on disparities in the
actual or perceived aspects of the profession and cyber domain [10, 11]. While this
study does not rule out the possibility of differences in ability or gender biases playing
a role, it highlights the need for further investigation into people’s decision-making
processes about their career paths.

1.1 Chapter Contributions

We highlight the main contributions of the chapter, as shown below.

• The highlights highlight that the global population strives to achieve the United
Nations’ goals of sustainable communities and cities, good health and well-being,
and responsible consumption and production.
• Women play a crucial role in various humanitarian activities, and their recognition
is vital in technology-based organizations.
• Recent research highlights the underrepresentation of women in different
technology-related fields.
328 W. Shafik

Fig. 1 Cyber security framework (CS entails identify, protect, detect, respond, and recover)

• Despite efforts to attract and retain women, cultural issues within the diverse
technology industry and global barriers present significant challenges.
• Leaders of all genders and backgrounds must address workforce development
gaps and create effective company strategies.
• The study focuses on increasing the presence of women in senior positions in
healthcare CS and IT.
• The nature of CS and IT, their importance, global internet utilization, and profes-
sional aspects related to the underrepresentation of women in the industry are
examined.
• The low number of women in these fields can be attributed to several factors, and
potential solutions are proposed.
• Women bring unique perspectives and problem-solving skills, enhancing the
development of secure medical systems, electronic health records, and patient
privacy protection.
• Addressing the factors influencing underrepresentation will significantly enhance
organizational technology acceptance, cybersecurity trust, effectiveness, and
efficiency.

1.2 The Chapter Organization

The remainder of this chapter is structured as follows. Section 2 presents the nature of
cybersecurity as it relates to women in technology-based domains, including the need
for CS and IT global internet utilization. Section 3 covers the critical forms of CS
(including network, application, cloud, endpoint, identity and access management,
data security, incident response, and disaster recovery) and IT (entailing hardware,
software, networks, cloud Computing, and IoT). Section 4 depicts CS as a profession
for women, demonstrating reasons that cause a reduced number of women in CS and
IT industries like lack of awareness, stereotypes, lack of representation, unconscious
Dissecting the Role of Women in Cybersecurity and Information … 329

bias, unequal pay, lack of role models, hostile work environments, limited support,
lack of flexible work arrangements, lack of training opportunities, limited access
to networks, lack of mentorship, perceived lack of fit, limited support for work-life
balance, lack of confidence. Section 5 presents women’s cybersecurity from a medical
perspective. Section 6 presents the discussion and recommendations for research to
be conducted in the future. Finally, Sect. 7 depicts the conclusion.

2 Nature of Cybersecurity

The role of women in CS is increasingly crucial as the field continues to expand

and develop. Historically, women have been underrepresented in technology-related
fields, including cybersecurity. However, this trend is changing, and more women
are entering the field, bringing unique perspectives and skillsets [12]. There has
been a growing awareness of the reputation of diversity and inclusivity in CS. This is
because diverse teams are more likely to be innovative, effective, and better equipped
to solve complex problems [13].
Women bring different experiences and viewpoints to the field, enabling them to
identify and address cybersecurity challenges that may have been previously over-
looked. Furthermore, in addition to the benefits of diversity, the nature of cyberspace
itself makes it an ideal field for women to pursue [14]. Unlike traditional tech roles
that may require physical strength or involve long hours in a lab, cybersecurity can
be done from anywhere with an internet connection, making it accessible to people
of all backgrounds and abilities [15]. It is a dynamic and constantly evolving field
that offers endless opportunities for learning and growth.
However, despite the opportunities available in the field, women still encounter
unique obstacles in cybersecurity, such as gender bias and a lack of female role
models. These barriers can deter women from entering the field or impede their
professional advancement once they do. Overcoming these challenges necessi-
tates a concerted effort by individuals, organizations, and policymakers to create
a more inclusive and supportive environment for women in CS [16, 17]. As for
cyberspace’s nature, it presents challenges and opportunities for cybersecurity profes-
sionals, including women. One of the most significant challenges is the constantly
evolving threat landscape, which necessitates that CS professionals are adaptable
and proactive in identifying and mitigating risks. Additionally, the interconnected-
ness of cyberspace implies that cybersecurity threats can propagate quickly and have
far-reaching consequences [18]. This necessitates a collaborative and coordinated
approach to cybersecurity within organizations and across sectors and borders.
The nature of cyberspace poses challenges and presents opportunities for inno-
vative solutions to cybersecurity issues. For instance, artificial intelligence (AI) and
machine learning (ML) technologies can be utilized to identify and respond to threats
in real-time.
The emergence of the IoT also provides opportunities for enhanced CS through
smart devices and sensors that can automatically detect and respond to threats [19,
330 W. Shafik

Fig. 2 Trust and resilience (C, Confidentiality; I, Integrity; A, Availability; P, Privacy; S, Safety;
R, Reliability)

20]. Women can play a crucial role in developing and implementing such innovative
solutions to bridge the gap in end-user trust and resilience, as illustrated in Fig. 2.
Diversity and inclusivity foster more creative and effective problem-solving [21].
By encouraging more women to pursue careers in cybersecurity and supporting their
professional development, their unique perspectives and skill sets can be utilized to
enhance the security of cyberspace.

2.1 Need for Cybersecurity and Information Technology

Women are crucial in addressing the growing need for cybersecurity in our digital
world. Cybersecurity threats have become more frequent and sophisticated with the
rise of technology and the internet. Women bring unique perspectives and skills to
the field, such as solid communication and collaboration abilities and expertise in
areas such as threat analysis and risk assessment. Diversity in cybersecurity can lead
to more innovative and practical solutions to complex challenges [22]. Encouraging
more women to pursue careers in cybersecurity through mentoring, outreach, and
professional development is crucial. The consequences of cybersecurity threats can
be severe, making the need for skilled professionals in the field more important than
ever.
The need for cybersecurity is driven by numerous factors, including the increasing
prevalence of cyberattacks, the growing amount of sensitive data being stored online,
and the rise of new technologies such as the IoT [23, 24]. CS threats can have
serious consequences, including financial losses, damage to reputation, and the
compromise of personal information. Given the importance of CS in our modern
world, we must encourage more women to pursue careers in the field. This can be
achieved through mentoring programs, educational outreach, and opportunities for
professional development and advancement.
The importance of CS extends beyond large corporations and government agen-
cies; small businesses, non-profits, and individuals are also at risk of cyber-attacks.
Dissecting the Role of Women in Cybersecurity and Information … 331

Women in CS can raise awareness and educate these vulnerable groups on best prac-
tices for staying safe online [25]. Women can play a crucial role in protecting these
populations from cyber threats by sharing their expertise on password management,
phishing scams, and safe browsing habits. Additionally, women can serve as role
models and mentors for those interested in pursuing a career in CS, inspiring and
empowering the next generation of professionals [26].

2.2 Global Internet Utilization

Over the past few years, internet usage has led to a rise in cybersecurity threats that
require skilled professionals to address. Women can contribute significantly to CS and
cyber defense (CD) in various ways. They can increase awareness of cybersecurity
threats and promote best practices for staying safe online by sharing their knowledge
and experiences. Women can also contribute to developing and implementing CS
policies and practices by bringing diversity to the table [27]. With only 24% of the
CS workforce being women, there is a need for greater gender diversity. Encouraging
more women to pursue careers in cybersecurity and providing mentorship and support
can help address this gap [28]. Women can also contribute to cybersecurity research
and innovation by identifying emerging threats and developing innovative solutions.
Women can significantly impact CD by proactively addressing cybersecurity risks
in their personal and professional lives. For instance, using strong passwords, regu-
larly updating software and security settings, and avoiding suspicious emails or
websites can help mitigate risks. Women can also advocate for greater diversity and
inclusion within the tech industry, leading to more innovative and practical solutions
to CS challenges. Furthermore, women can bring empathy and social awareness to
the field, which can help to develop more holistic and practical approaches to CS [29].
CS is not solely a technical problem but also a social and human one, and women’s
ability to connect with others and understand the cultural and social implications of
cyber threats can make a meaningful difference.

3 Cybersecurity and Information Technology Key Forms

3.1 Cybersecurity Key Forms

This section presents some common cybersecurity forms, including network and
application security.
332 W. Shafik

3.1.1 Network Security

Network security is a crucial part of cybersecurity that protects computer networks

from cyber-attacks, unauthorized access, and other malicious activities. In the early
days of computing, networks were simple and secure, but as they became more
interconnected, the risk of cyber-attacks increased significantly. Today, networks
are accessible to a wide range of users, making them vulnerable to various security
threats [28].
Cyber-attacks, such as hacking, malware, phishing, and denial-of-service (DoS)
attacks, can cause data theft, system downtime, financial losses, and reputational
damage. Detecting these attacks can be challenging, making it difficult to respond
promptly. Network security requires implementing security measures such as intru-
sion detection systems, firewalls, virtual private networks, and ongoing monitoring
and updates to ensure network security.

3.1.2 Application Security

Application security is crucial for protecting applications from cyber threats like
hacking and malware. Applications are vital for digital systems but also vulner-
able to cyber-attacks due to the lack of security considerations during development
[29]. Implementing secure coding practices, encryption, and access controls can help
protect sensitive data. Ongoing monitoring and testing are necessary to identify and
address vulnerabilities.

3.1.3 Cloud Security

The acritical aspect of CS is that it focuses on protecting cloud computing systems

and services from a wide range of cyber threats like data breaches, malware, and
hacking. The widespread adoption of cloud computing has transformed how busi-
nesses operate, enabling organizations to store and manage data more efficiently and
cost-effectively [30]. However, cloud computing also presents new challenges and
vulnerabilities, as sensitive data may be stored on remote servers that are accessed
via the Internet. Cloud security is vital because cloud computing systems are highly
susceptible to cyber threats like data breaches, malware, and hacking.
Organizations must implement various security measures and best practices to
address these security challenges to protect their cloud computing systems from cyber
threats. This includes using encryption, access control, and multifactor authentica-
tion to secure sensitive data, regularly performing security audits and vulnerability
testing, and collaborating with cloud service providers with robust security proto-
cols. All parties involved in the cloud computing ecosystem must know their security
responsibilities and work together to protect cloud systems.
Dissecting the Role of Women in Cybersecurity and Information … 333

3.1.4 Endpoint Security

Endpoint security is a vital aspect of CS that protects individual devices, including

laptops, smartphones, and other endpoints, from various cyber threats such as
malware, phishing attacks, and hacking. The significance of endpoint security can
be better comprehended by understanding the issues it addresses. Endpoint devices
are increasingly used to access sensitive data and systems in the workplace and
personal contexts, making them more susceptible to cyber threats. Unlike other
systems, endpoint devices may be utilized outside the organization’s secure network,
making them more vulnerable to cyber threats [31, 32]. Furthermore, they may not
have the same security control and monitoring level as other systems, making them
more susceptible to cyber-attacks. Implementing a range of security measures and
best practices is the key to endpoint security, which includes using antivirus software,
firewalls, and encryption to protect against malware and other cyber-attacks. Regular
device updates, patching, and monitoring for security vulnerabilities can be ensured
through security policies and procedures.

3.1.5 Identity and Access Management (IAM)

A vital aspect of CS is that it aims to safeguard digital systems by allowing only

authorized users access to sensitive resources. Managing user identity and access in
today’s complex digital environment can be a significant challenge, and unautho-
rized access to sensitive data and resources can lead to data breaches and other cyber
threats [33]. Implementing a range of security measures and best practices, such
as multifactor authentication and role-based access control, is the solution to this
problem. IAM ensures that only necessary users have access to sensitive resources
while regularly reviewing and updating user access privileges to prevent any unau-
thorized access. Thus, IAM helps organizations safeguard their digital systems and
resources while maintaining the confidentiality, integrity, and availability of their
sensitive data.

3.1.6 Data Security

This concentrates on safeguarding sensitive data from unauthorized access, use,

disclosure, or destruction. To understand the significance of data security, providing
some background on the issues it aims to tackle is helpful. Data is critical to the
functioning of many organizations and is increasingly stored and processed in digital
format. However, this also makes it susceptible to various cyber threats such as
hacking, theft, and data breaches.
The issue with data security is that unauthorized individuals or cybercriminals
can easily access sensitive data, leading to severe consequences such as reputational
damage, financial losses, and legal liability for organizations [34]. The solution to
data security is implementing various security measures and best practices to protect
334 W. Shafik

sensitive data from unauthorized access or use. This involves implementing robust
access control, encryption, and secure data storage and transmission techniques.
Moreover, organizations should conduct regular security assessments and audits to
identify vulnerabilities and take appropriate measures to address them.

3.1.7 Incident Response (IR) and Disaster Recovery (I/DR)

It is an essential form of cybersecurity that aims to prepare organizations to respond

quickly and effectively to cyber incidents and recover from any resulting damage or
disruptions. It is essential to understand the importance of IR/DR and the problem it
aims to solve. Cyber incidents like data breaches and ransomware attacks can have
devastating consequences, including data loss, system downtime, and reputational
damage [35]. The difficulty with cyber incidents is that they can be challenging to
detect and address promptly, leading to severe consequences for the organization.
The solution to IR/DR is implementing various security measures and best prac-
tices designed to prepare organizations to respond to cyber incidents and recover
from any resulting damage or disruptions. This includes creating an incident response
plan, conducting regular security assessments and training, and implementing backup
and recovery systems [35]. By implementing effective IR/DR measures and best
practices, organizations can reduce the risk of cyber incidents and ensure that they
can respond effectively and recover quickly in case of any incident. Additionally,
emerging areas of cybersecurity, such as AI and IoT, are becoming increasingly
important as these technologies become more widespread. Effective cybersecu-
rity requires a comprehensive and integrated approach that addresses all aspects
of digital systems and networks, from network security to incident response and
disaster recovery.

4 Information Technology Key Form

IT refers to the use of digital technologies to manage and process information. There
are various forms of IT, including the following and women’s input to fact CS.

4.1 Hardware

Hardware in IT refers to physical computer components, which can be vulnerable

to security threats like theft, tampering, and malware infections. One challenge in
this area is the lack of diversity and representation, particularly with the underrepre-
sentation of women in hardware engineering. This limits the range of perspectives
and problem-solving approaches in the industry, hindering innovation and creativity.
Women in cybersecurity can offer new solutions to these security challenges,
Dissecting the Role of Women in Cybersecurity and Information … 335

contribute to secure hardware development, and design effective security protocols.

Programs and events that support women in the industry can encourage their partic-
ipation and provide professional development and opportunities for advancement
[36].
Women in CS can bring diverse perspectives and experiences to the hardware
engineering field, helping identify new solutions to security challenges. For example,
they can contribute to the development of more secure hardware components and
systems, as well as to the design of more effective security protocols and practices.
Moreover, mentorship programs and networking events can help support women in
the field, encourage their participation in the hardware aspect of the IT industry, and
provide opportunities for professional development and advancement.

4.2 Software

Computer programs and applications, including operating systems, productivity

tools, and specialized software for specific tasks. Security threats to the software
include malware infections, coding errors, and vulnerabilities in third-party libraries
and dependencies. However, the lack of diversity and representation, particularly
the underrepresentation of women in software engineering, limits innovation and
creativity by narrowing the range of perspectives and approaches to solving problems
[37]. Women in CS can contribute to developing more secure and user-friendly soft-
ware programs and effective security protocols and practices. Mentorship programs
and networking events can encourage women’s participation in software engineering
and support their professional development and advancement.

4.3 Networks

These networks can be vulnerable to various security threats, including network

attacks, data breaches, and unauthorized access. The underrepresentation of women
in network engineering and related fields is a central problem, which can lead to a
narrow range of perspectives and approaches to solving problems, limiting innovation
and creativity. Women in cybersecurity can bring diverse perspectives and experi-
ences to network engineering, helping identify new solutions to security challenges
[38]. Initiatives such as mentorship programs and networking events can encourage
women’s participation in the network aspect of the IT industry and provide oppor-
tunities for professional development and advancement. Women in cybersecurity
can play a vital role in developing more secure and innovative network solutions,
promoting greater diversity and representation in the field.
336 W. Shafik

4.4 Cloud Computing

Cloud computing delivers computing services, such as software, storage, and

computing power, over the internet. Third-party providers are responsible for
managing and maintaining the underlying infrastructure. However, one of the main
concerns in cloud computing is data security. Cloud environments are vulnerable to
various security threats, including data breaches, denial-of-service attacks, and unau-
thorized access. Women can help to identify and mitigate security risks, implement
security policies and procedures, and design effective security control and moni-
toring systems [39]. In addition, women can play a role in developing more secure
cloud-based applications and services by incorporating security considerations into
the design and development process. Mentorship programs and networking events
can also support women in the field, encourage their participation in cloud computing,
and provide professional growth and development opportunities.

4.5 Internet of Things

One of the key challenges in AI is the lack of diversity and representation. AI tech-
nologies like IoT and algorithms are developed by individuals and teams who bring
their own biases, perspectives, and experiences to the process, which can lead to unin-
tended consequences and reinforce existing social inequalities. Women and other
underrepresented groups are often excluded from the development and deployment
of AI systems, which can perpetuate biases and limit the potential benefits of AI for
society.
Efforts are underway to promote diversity and inclusion in AI. This includes
initiatives to increase the representation of women and other underrepresented groups
in AI research and development and efforts to mitigate bias in AI algorithms and
systems [40]. For example, data sets can be audited for bias, and diverse teams can
be assembled to ensure that various perspectives are represented in the development
process. By promoting diversity and inclusion in AI, we can help ensure that these
technologies are developed and deployed in fair, ethical, and beneficial ways for all
members of society.

5 Cybersecurity as a Profession for Women

CS is an expanding field that has gained significant importance in recent times. The
increasing usage of digital technology and the internet has led to a rise in cyberattacks
that are becoming more sophisticated and frequent. These attacks severely threaten
individuals, businesses, and governments worldwide [14, 15]. Therefore, the demand
for competent cybersecurity professionals has surged, with a global shortage of over
Dissecting the Role of Women in Cybersecurity and Information … 337

three million skilled personnel in this area. However, women remain substantially
underrepresented in the field, comprising only around 20% of the cybersecurity
workforce.
This gender gap could lead to a lack of diversity in perspectives and approaches
toward solving cybersecurity problems, thereby limiting innovation and creativity in
the industry [16]. Therefore, it is crucial to encourage and support more women to
participate in cybersecurity to promote diversity, broaden perspectives, and bridge the
gender gap in this rapidly evolving field. Several factors contribute to the underrep-
resentation of women in cybersecurity, including the gender gap in STEM (science,
technology, engineering, and mathematics) fields, unconscious bias in hiring, and a
lack of female role models and mentors.
The nature of the field can also present challenges for women, such as the require-
ment for continuous learning and professional development, long and irregular hours,
and a culture often dominated by men [17]. These obstacles can discourage women
from pursuing careers in cybersecurity despite the growing demand for skilled profes-
sionals. It is vital to create more inclusive and diverse workplaces, provide mentorship
and networking opportunities for women, and promote STEM education and careers
for young girls and women. By taking these steps, we can help to bridge the gender
gap in cybersecurity and ensure that this critical field has the talent and perspectives
it needs to thrive.
The underrepresentation of women in CS not only results in missed opportunities
for the industry to tap into a wider pool of talent and perspectives but also has
significant implications for CS. Research has shown that gender diversity can enhance
problem-solving, creativity, and innovation, ultimately leading to more effective and
efficient decision-making [18, 19]. Moreover, diverse teams are better equipped to
understand and address the wide range of cyber threats and vulnerabilities. With the
rapidly increasing number and complexity of cyberattacks, the cybersecurity industry
needs to leverage the benefits of diversity to strengthen its defense against threats.
Encouraging and supporting the participation of women in CS can lead to better
outcomes, not only for the industry but also for society as a whole.
The gender gap in cybersecurity can be tackled through several initiatives that
have emerged in recent years. These include training programs and scholarships to
encourage and support women, as well as mentorship and networking opportunities
[20]. To mitigate unconscious bias in the hiring process, some organizations have
implemented blind recruitment processes, while others have tried to foster more
inclusive workplace cultures.
Promoting cybersecurity as an attractive profession for women is also crucial. This
can be done by showcasing the diverse range of roles and opportunities within the field
and highlighting how CS professionals can make a meaningful and positive impact
on society [21]. Furthermore, addressing the gender gap in STEM education and
providing more opportunities for girls and young women to explore and pursue STEM
fields can also help address the underrepresentation of women in cybersecurity.
In addition, mentorship and networking opportunities can be invaluable for women
entering the CS field. These opportunities allow women to connect with experienced
338 W. Shafik

professionals who can offer guidance and advice on career development and intro-
duce them to potential job opportunities [22]. Many organizations and industry asso-
ciations have established mentorship and networking programs for women in CS,
which can be particularly helpful for those who may feel isolated or marginalized in
male-dominated workplaces.
Another critical step is to address unconscious bias in the hiring process. This
can involve implementing blind recruitment processes, which remove identifying
information from job applications to mitigate bias based on gender or race. It can
also involve creating more inclusive workplace cultures where all employees feel
valued and supported.
Promoting cybersecurity as an exciting and rewarding profession is the key to
attracting more women to the field. CS requires individuals who are curious, creative,
and analytical [23]. It offers a range of career opportunities, from incident response
and threat intelligence to risk management and compliance. Furthermore, the work
of CS professionals has a significant impact on society, as they are responsible for
protecting critical infrastructure, sensitive data, and personal information.
Increasing the representation of women in CS is critical for promoting diversity,
innovation, and practical problem-solving in the field. By addressing the challenges
and barriers women face in the industry and promoting greater awareness and oppor-
tunities, we can work toward a more inclusive and effective cybersecurity workforce
[24]. CS is a dynamic and exciting field that offers a wealth of opportunities for
curious, creative, and driven people. Encouraging more women to enter the field
ensures that the industry is equipped to address the evolving threats of the digital
age. The following are some of the identified reasons for considerably low numbers
of wine in cybersecurity and the IT industry.

5.1 Lack of Awareness

Although the lack of awareness regarding CS as a career choice for women is a

persistent problem, efforts are underway to address this issue. Women already in
the CS and IT industries are leading outreach programs to increase awareness of the
benefits and opportunities available in the field. Additionally, CS and IT companies
are taking action to enhance their diversity and inclusivity by implementing initiatives
focused on diversity and inclusion and collaborating with organizations that support
the inclusion and advancement of women in the industry [25]. However, despite these
efforts, the issue of awareness persists, and further action is necessary to promote
CS and IT careers with women.
Limited exposure to cybersecurity and IT careers in schools and universities,
lack of female role models in leadership positions, stereotypes and biases, limited
networking and mentorship opportunities, and misconceptions about the nature and
scope of cybersecurity and IT careers contribute to the underrepresentation of women
in the field. Insufficient outreach to underrepresented communities, limited access
to industry resources, and the perception that cybersecurity and IT careers are not
Dissecting the Role of Women in Cybersecurity and Information … 339

creative or fulfilling are also factors. Addressing these challenges requires promoting
greater awareness of the benefits and opportunities available in cybersecurity and IT
careers and providing more guidance and support to women interested in the field.

5.2 Stereotypes

Stereotypes have deterred many women from pursuing careers in the male-dominated
and technical fields of cyber security and IT, where they are underrepresented and lack
role models. Despite representing only 24% of the cybersecurity workforce globally,
women can help bridge the significant skill gap in the industry [26]. Gender bias and
discrimination in the workplace can further hinder their professional development and
pay equality. Initiatives like mentorship programs, networking events, and awareness-
raising are crucial to encouraging more women to pursue careers in these fields and
challenge stereotypes. The persistent belief that men are better suited for technical
fields hinders achieving greater diversity and inclusion in cybersecurity and IT.

5.3 Lack of Representation

The underrepresentation of women in leadership positions and technical roles in the

cyber security and IT industry discourages female talent and perpetuates the myth
that these fields are male-dominated. This lack of representation limits the diversity
of perspectives and ideas, hindering innovation and progress. Women hold only 25%
of IT-related jobs and 11% of executive positions in Fortune 500 companies [27, 28].
Initiatives like mentorship programs, recruiting efforts, and creating more opportuni-
ties for women to advance their careers can help create inclusive environments where
women can thrive and achieve their full potential. The industry needs to tap into the
potential of half of the global population to address the significant shortage of skilled
professionals. Therefore, promoting greater diversity and inclusion is essential to
attract and retain female talent in the cyber security and IT industry.

5.4 Unconscious Bias

Unconscious biases of hiring managers and recruiters may favor male candidates,
a significant obstacle to attracting and retaining women in the cyber security and
IT industry. These biases can manifest as automatic and unintentional attitudes and
beliefs about people based on gender, race, ethnicity, or other characteristics and can
impact the hiring process, performance evaluations, and overall workplace culture.
For instance, a hiring manager may unconsciously favor male candidates for tech-
nical roles, assuming they are more competent than female candidates [29]. This bias
340 W. Shafik

reinforces the myth that women are less capable in technical roles and can dissuade
women from pursuing opportunities in these fields. The industry must address uncon-
scious biases through awareness training for hiring managers and recruiters to ensure
a fair and inclusive hiring process that enables both men and women to succeed. This
will help to promote greater diversity and inclusion and bridge the cybersecurity skills
gap by tapping into the potential of underrepresented groups.

5.5 Unequal Pay

The gender pay gap is a significant factor contributing to the underrepresentation of

women in the industry. Despite equal pay laws, women continue to earn less than
men, with a wage gap that persists across all industries, including tech [28, 29].
Women of color experience even more pronounced gender and racial wage gaps.
The unequal pay devalues women’s work, creating disincentives for them to pursue
or remain in these fields. Women in tech earn, on average, 20% less than men, and the
gap widens as they advance in their careers. To address this issue, organizations must
conduct pay audits and implement policies and programs promoting pay equity, such
as transparent pay structures, performance-based pay, and supportive parental leave
policies. A fair and equitable pay system will attract, retain, and advance talented
female professionals.

5.6 Deficiency of CS Women Role Models

Women in the CS and IT industries may face a nonexistence of role models to emulate
and learn from, which can discourage them from pursuing a career in these fields.
The shortage of role models is linked to the broader problem of underrepresentation,
where few women occupy senior positions in the industry. The absence of women
in leadership positions can lead to a lack of visibility of women’s potential and
ability, perpetuating the stereotype that cybersecurity and IT are male-dominated
fields. Various factors, including unconscious bias, lack of diversity and inclusion,
and gender stereotyping, cause this underrepresentation. The scarcity of role models
can also hinder women’s career progression in the industry [29]. Without access to
female leaders in the field, women may struggle to identify potential career paths
and gain the necessary skills and experience to advance. Therefore, increasing the
number of women in leadership positions in cybersecurity and IT is crucial to provide
guidance and support for younger women interested in these fields. Doing so will
help inspire and encourage more women to pursue and thrive in cybersecurity and
IT careers.
Dissecting the Role of Women in Cybersecurity and Information … 341

5.7 Hostile Work Environments

Women face barriers in various aspects of the industry, including biased hiring prac-
tices, unequal pay, lack of role models, and hostile work environments. The underrep-
resentation of women in leadership positions in the industry exacerbates the problem
[23, 30]. To tackle these issues, organizations must conduct pay audits, implement
policies that promote diversity and inclusion, and provide training to create a safe
and supportive work environment. Increasing the number of women in leadership
positions is essential to serve as role models for younger women interested in the
field. Creating a fair and equitable workplace will attract more women to the industry
and help retain and advance talented female professionals.

5.8 Limited Support

Women face several obstacles in pursuing a career in these fields, such as a lack
of access to resources, training, and mentorship. Additionally, they may not have
the same opportunities for internships and networking as men, even if they pursue
technology-related degrees. Discrimination and harassment in the workplace also
make it challenging for women to feel valued and respected as part of a team [31].
Companies must invest in programs and initiatives that support women’s recruitment,
retention, and advancement in these fields. Creating an inclusive work environment
and providing equal opportunities can help address the problem of limited support
for women in the IT and CS industry.

5.9 Flexible Work Arrangements Deficiency

The absence of flexible work arrangements is a major obstacle preventing women

from pursuing careers in this field. The industry has historically operated on a 24/7
basis, making it difficult for women who need flexible work hours to balance their
personal and professional lives. This has contributed to the underrepresentation of
women in the field, as they may leave the industry due to inflexible work arrange-
ments. Despite some progress in providing flexible work arrangements, current affairs
are still limited. Many companies still adhere to traditional work schedules [30],
making it difficult for women with caregiving responsibilities to participate fully in
the industry. However, with the advent of remote work, companies have an oppor-
tunity to provide more flexible work arrangements that allow women to balance
work and personal responsibilities. This would enable the industry to attract and
retain a diverse workforce, including women, and ensure that everyone has an equal
opportunity to succeed.
342 W. Shafik

5.9.1 Nonexistence of Training Opportunities

The lack of training opportunities significantly contributes to the underrepresentation

of women in the cybersecurity and IT industry. The industry has traditionally been
male-dominated, and training programs have typically been geared toward men. As a
result, women often lack the necessary skills and knowledge to succeed in the field.
This issue is compounded by the fact that women may face barriers to accessing
training opportunities, such as financial constraints, family responsibilities, and lack
of support. As a result, women may not have the same training and experience as
men, making it challenging for them to compete for jobs in the industry [31]. There
is a need for more accessible and inclusive training opportunities for women. This
could involve developing training programs targeting women and their unique needs
and challenges. Additionally, there should be a greater emphasis on mentorship
and networking opportunities for women to help them build their skills and make
connections in the industry.

5.9.2 Limited Access to Networks

Limited access to networks is a significant obstacle that impedes women’s career

advancement in cybersecurity and IT. Women often struggle to build professional
networks due to various challenges, including cultural barriers, lack of representa-
tion, and unconscious biases. As women are already underrepresented in the field,
their ability to connect with peers, mentors, and leaders is even more constrained.
Furthermore, the lack of access to networks can hinder opportunities for professional
growth, training, and advancement [32]. Organizations need to take measures to
provide women with equal networking opportunities. This could include organizing
events, establishing mentorship programs, and promoting participation in industry
groups and conferences. By providing equal access to networks, organizations can
help women in their careers, increase their contribution to the industry, and foster a
more diverse and inclusive work environment.

5.9.3 Absence of Mentorship

Women may lack mentors who can provide guidance and advice on career develop-
ment. Limited access to networks is another reason why there are few women in CS
and IT. Access to online and offline professional networks is crucial for career devel-
opment and growth, but women often have limited access to these networks due to
exclusionary practices and a lack of diversity in leadership positions. Male-dominated
networks tend to exclude women, and women may not have the same opportuni-
ties to build relationships with colleagues and mentors [33]. This leads to missed
opportunities for professional development, job opportunities, and advancement.
Additionally, women may not have access to informal networks that provide
essential information about industry trends, job openings, and potential mentors. As
Dissecting the Role of Women in Cybersecurity and Information … 343

a result, women may be left out of meaningful conversations and decision-making

processes. The current state of limited access to networks for women in cybersecurity
and IT highlights the need for more inclusive and diverse networks that provide equal
opportunities for women to connect and advance their careers.

5.9.4 Perceived Lack of Fit

Perceived lack of fit refers to the perception that women do not fit in the male-
dominated tech industry, and this belief can discourage women from pursuing careers
in cybersecurity and IT. This problem stems from societal stereotypes and gender
biases that portray technical roles as masculine and analytical, which creates a
perceived mismatch between women and the field. This perception of not fitting in
can lead to feelings of isolation, exclusion, and low self-efficacy, further discouraging
women from pursuing cybersecurity and IT careers [34].
The industry’s current state reflects this issue, as women remain underrepresented
in cybersecurity and IT roles. Addressing this issue requires a shift in mindset and
culture that embraces diversity, inclusivity, and gender equality. It requires creating
more opportunities for women to showcase their skills, promoting gender-neutral
language in job descriptions, and challenging gender stereotypes and biases perpet-
uating the perception of a lack of fit. By actively promoting gender diversity and
inclusivity, we can attract and retain more women in cybersecurity and IT, creating
a more equitable and innovative industry.

5.9.5 Limited Support for Work-Life Balance

Work-life balance has become increasingly important to professionals, especially

women who may face additional caregiving and family responsibilities. However,
the lack of support for work-life balance has been identified as a significant obstacle
for women pursuing careers in cybersecurity and IT. Women with children or other
caregiving responsibilities often struggle to balance these responsibilities with their
work duties due to long hours, inflexible work schedules, and a lack of employer
support [35]. This can lead to burnout, stress, and, ultimately, women leaving the
field. Women may face cultural biases that expect them to prioritize their family
responsibilities over their careers. This can lead to women being perceived as less
committed to their work and, as a result, being passed over for promotion or being
excluded from significant opportunities. Companies need to implement more flexible
work arrangements, such as telecommuting and flexible hours, and provide more
support for women with caregiving responsibilities.
344 W. Shafik

5.9.6 Lack of Confidence

It is often observed that women have lower self-confidence levels compared to their
male counterparts. This lack of confidence is often attributed to societal and cultural
factors discouraging women from pursuing careers in technology-related fields [36].
The problem is compounded by the fact that the IT industry is predominantly male-
dominated, with few women in leadership positions. This creates a perception that
women are not suited for leadership roles in the industry. As a result, women may
lack confidence in their abilities, which leads to underrepresentation in the industry.
Current efforts to address this issue include mentorship programs, training, and
educational initiatives aimed at building self-confidence among women.

6 Women’s Cybersecurity in Medical Perspective

CS is becoming increasingly vital in the medical industry to protect the confiden-

tiality, integrity, and availability of patient data. Women are making valuable contri-
butions as healthcare professionals, CS analysts, and engineers in this field. One of
the main challenges in healthcare cybersecurity is the potential for data breaches
or cyber-attacks that could compromise patient data [37]. Women bring a unique
perspective to this challenge, as they are often well-versed in issues related to privacy
and confidentiality, which are critical considerations in the medical field. Moreover,
women play a pivotal role in developing secure medical technologies and systems,
ensuring that they are designed with patient safety in mind.
Additionally, women in CS are playing a vital role in educating healthcare profes-
sionals and organizations on CS best practices and raising awareness about the impor-
tance of protecting patient data. With technology playing an increasingly critical role
in healthcare, a diverse workforce that brings a range of perspectives and skills is
essential. Women are valuable contributors to the medical industry, as they bring a
variety of backgrounds and experiences that can help to address the unique challenges
of healthcare CS.
In addition to these technical roles, women in cybersecurity can contribute to
developing policies and procedures that promote a culture of security and privacy
within healthcare organizations [38]. They can help raise awareness about the impor-
tance of cybersecurity among healthcare staff and patients and educate them about
best practices for safeguarding sensitive data. Furthermore, women in cybersecurity
can advocate for diversity and inclusion in the industry, which can help to address
the underrepresentation of women and other marginalized groups in cybersecurity
and IT. By promoting diversity and inclusion, women in cybersecurity can help to
create a more resilient and effective healthcare cybersecurity workforce.
In addition, women in cybersecurity can contribute to developing secure and inno-
vative technologies that improve patient care, such as secure telemedicine platforms
that allow patients to connect with their healthcare providers remotely or mobile
applications that help them securely manage their health and wellness [39]. They can
Dissecting the Role of Women in Cybersecurity and Information … 345

also address the shortage of cybersecurity professionals in the healthcare industry by

encouraging more women to pursue careers in cybersecurity and promoting diversity
and inclusion in the workplace.
Additionally, organizations can work toward reducing gender-based pay gaps
and providing equal opportunities for career advancement and leadership positions
[40]. It is also essential to recognize and celebrate the contributions of women in
cybersecurity, highlighting their successes and promoting them as role models for
future generations of women in the field. By addressing these challenges, we can
foster a more diverse and inclusive workforce in medical cybersecurity that reflects
all individuals’ needs and perspectives and ultimately helps improve patient outcomes
and safety.

6.1 Discussion

A significant degree of gender segregation in the labor force has remained consistent
throughout the study of gender and economics. Most men and women select very
dissimilar careers. This segregation pattern may have resulted from various circum-
stances, such as prejudice, differences in intelligence, or even an individual’s own
free will [41]. We can directly test the effects of preferences on career choices in the
context of choosing between IT and non-IT women’s professional careers because
we have introduced a direct measure of individual preferences by utilizing a widely
accepted measure of occupational personality and by controlling via our sampling
procedures, for factors such as educational attainment and attachment to the work-
force. In other words, we could directly assess the influence of preferences on job
decisions.
We find that differences in preferences among professional employees with full-
time jobs can explain a substantial percentage of the apparent underrepresentation of
women in information technology [42]. In other words, a large percentage of the gap
in the number of men and women entering the information technology profession
can be linked to the fact that, on average, men and women value different aspects
of their occupations and, as a result, choose distinct career paths [43]. When these
differences in preferences were considered, there was a significant change in how
men and women choose carriers.
According to the findings of this study [44], it may be worthwhile to investigate
the possibility that gender differences in preferences contribute to gender gaps in
industries other than information technology, particularly in disciplines that empha-
size the realistic general occupational theme. We cannot rule out the possibility that
discrimination or differences in ability also act as filters that differentially reduce the
entry of women into professional occupations more generally because our research
design limits the sample to individuals who have chosen careers in IT and comparable
occupations with controls for career motivation, education, and cognitive abilities.
This means the sample is restricted to those who have selected careers in information
technology and comparable fields.
346 W. Shafik

There is a need for additional studies to address this issue; nonetheless, given
that women make up nearly half of our control group, inequalities in ability or
discrimination cannot be deemed insurmountable barriers to women’s admittance
into professional employment. Even though we cannot rule out the role of discrimi-
nation or differences in ability, our findings indicate that, even if these factors were
eliminated, women would still have a lower representation in the field of information
technology due to differences in their occupational preferences compared to men
[45].
The observed differences in occupational personality may result from a domino
effect in which one factor causes another. In this scenario, changes in occupational
personality may result from a chain reaction. Previous work experience has likely
influenced your chosen career route [46]. Given the information shown previously
regarding the stability of occupational choices, we have reason to anticipate that
these effects will have only a minor impact; yet, given the evidence available, we
cannot rule out the potential that these effects may have an enormous impact. Before
a person begins a career, it would be ideal to collect longitudinal data that could be
used to assess their occupational personality. This would be the best circumstance.
This type of data collection could be conducted before entering the labor field [47].
Due to budgetary constraints, we could not obtain such data; nonetheless, the results
of our analysis suggest that additional research along these lines could be of great
use.
After discovering that a person’s occupational personality may play a part in
explaining why women and men experience different labor market results, it is vital
to undertake additional research on the elements that contribute to forming this aspect
of a person’s career preferences [48]. The discovery that occupational personality
may play a role in understanding why women and men experience varied labor market
results necessitates this avenue of investigation. Occupational personality is a compli-
cated attribute that is not innate but rather the outcome of the interaction between an
individual’s qualities and the aspects of their environment [49, 50]. Consequently,
parental and other family influences and educational and societal pressures probably
contributed to the disparity [51]. Understanding how and why such gaps occur looks
to be a crucial topic for future research, especially for those who seek to improve the
proportion of women working in technical fields nationwide.

7 Conclusion

The role of women in CS and IT is an area that is gaining attention due to the
increasing importance of these fields in modern society. As technology advances, so
do the threats and risks associated with it. Therefore, having a diverse and inclu-
sive workforce that brings different perspectives and experiences to the table is
essential. From a medical perspective, it is clear that women have much to offer
in these fields. Women tend to have strong communication and collaboration skills
Dissecting the Role of Women in Cybersecurity and Information … 347

essential for effective teamwork and problem-solving. Women are often more detail-
oriented and have a greater capacity for multitasking, making them valuable assets
in the fast-paced world of CS and information technology. However, despite the
benefits of having women in these fields, barriers still prevent women from pursuing
careers in CS and information technology. These barriers include stereotypes, biases,
lack of role models, and mentorship opportunities. Promoting diversity and inclu-
sivity in the workplace is vital to overcoming these barriers, providing training and
education opportunities, and encouraging more women to pursue careers in these
fields. Inclusively, the role of women in cybersecurity and information technology
is crucial for ensuring the security and integrity of our technological infrastructure.
By breaking down barriers and promoting diversity and inclusivity, we can create a
more robust and resilient workforce better equipped to tackle the challenges of the
digital age. Staffing shortages necessitate CS and technology management compe-
tence. Women and male leaders must address workforce development gaps and create
creative organizational initiatives.

References

1. Akyıldız D, Bay B (2023) The effect of breastfeeding support provided by video call on post-
partum anxiety, breastfeeding self-efficacy, and newborn outcomes: a randomized controlled
study. Jpn J Nurs Sci 20(1):e12509. https://doi.org/10.1111/jjns.12509
2. Fahim KE (2024) Electronic devices in the Artificial Intelligence of the Internet of Medical
Things (AIoMT). In: Handbook of security and privacy of AI-enabled healthcare systems and
internet of medical things. CRC Press, pp. 41–62. https://doi.org/10.1201/9781003370321-3
3. Avolio B, Chávez J (2023) Professional development of women in STEM careers: evidence
from a Latin American country. Glob Bus Rev 09721509221141197. https://doi.org/10.1177/
09721509221141197
4. Awano Y, Osumi M (2023) Peacock feathers and Japanese costume culture: evaluations from
spectrum images and microscopic observations. SCIRES-IT-Sci Res Inf Technol 12(2):77–86.
https://doi.org/10.2423/i22394303v12n2p77
5. Blalock EC, Lyu X (2023) The patriot-preneur–China’s strategic narrative of women
entrepreneurs in Chinese media. Entrep & Reg Dev 1–34. https://doi.org/10.1080/08985626.
2023.2165170
6. Coffie CPK, Hongjiang Z (2023) FinTech market development and financial inclusion in Ghana:
the role of heterogeneous actors. Technol Forecast Soc Chang 186:122127. https://doi.org/10.
1016/j.techfore.2022.122127
7. Contreras T, Leembruggen M (2023) A world of women in STEM: an online learning platform.
Bull Am Phys Soc
8. De Gioannis E, Pasin GL, Squazzoni F (2023) Empowering women in STEM: a scoping
review of interventions with role models. Int J Sci Educ, Part B 1–15. https://doi.org/10.1080/
21548455.2022.2162832
9. Farzin I, Abbasi M, Macioszek E, Mamdoohi AR, Ciari F (2023) Moving toward a more sustain-
able autonomous mobility, case of heterogeneity in preferences. Sustainability 15(1):460.
https://doi.org/10.3390/su15010460
10. Ferati M, Demukaj V, Kurti A, Mörtberg C (2022) Challenges and opportunities for women
studying STEM. ICT innovations 2022. Reshaping the future towards a new normal: 14th
international conference, ICT innovations 2022, Skopje, Macedonia, September 29–October
1, 2022, Proceedings, pp 147–157. https://doi.org/10.1007/978-3-031-22792-9_12
348 W. Shafik

11. Shafik W (2023) Cyber security perspectives in public spaces: drone case study. In: Handbook
of research on cybersecurity risk in contemporary business systems. IGI Global, pp 79–97.
https://doi.org/10.4018/978-1-6684-7207-1.ch004
12. Jun Y, Craig A, Shafik W, Sharif L (2021) Artificial intelligence application in cybersecurity
and cyberdefense. Wirel Commun Mob Comput. 1–10. https://doi.org/10.1155/2021/3329581
13. Shafik W, Matinkhah S M, Ghasemzadeh M (2020) Internet of things-based energy manage-
ment, challenges, and solutions in smart cities. J Commun Technol, Electron Comput Sci
27:1–11. https://doi.org/10.22385/jctecs.v27i0.302
14. Zhao L, Zhu D, Shafik W, Matinkhah SM, Ahmad Z, Sharif L, Craig A (2022) Artificial intel-
ligence analysis in cyber domain: a review. Int J Distrib Sens Netw 18(4)15501329221084882.
https://doi.org/10.1177/15501329221084882
15. Shafik W, Matinkhah SM, Sanda MN, Shokoor F (2021) Internet of things-based energy effi-
ciency optimization model in fog smart cities. JOIV: Int J Inform Vis 5(2)105–112. https://doi.
org/10.30630/joiv.5.2.373
16. Shafik W, Matinkhah SM, Ghasemazade M (2019) Fog-mobile edge performance evaluation
and analysis on internet of things. J Adv Res Mob Comput 1(3):1–7. https://doi.org/10.5281/
zenodo.3591228
17. Shafik W, Matinkhah SM (2019) Privacy issues in social Web of things. In: 2019 5th Interna-
tional Conference on Web Research (ICWR), Tehran, Iran. IEEE, pp 208–214. https://doi.org/
10.1109/ICWR.2019.8765254
18. Balabantaray SR, Mishra M, Pani U (2023) A sociological study of cybercrimes against women
in India: deciphering the causes and evaluating the impact on the victims. Int J Asia-Pac Stud
19(1). https://doi.org/10.21315/ijaps2023.19.1.2
19. Shafik W (2024) Wearable medical electronics in artificial intelligence of medical things. In:
Handbook of security and privacy of ai-enabled healthcare systems and internet of medical
things. pp 21–40. https://doi.org/10.1201/9781003370321-2
20. Shafik W, Matinkhah SM, Shokoor F (2022) Recommendation system comparative analysis:
internet of things aided networks. EAI Endorsed Trans Internet Things 8(29). https://doi.org/
10.4108/eetiot.v8i29.1108
21. Matinkhah SM, Shafik W, Ghasemzadeh M (2019) Emerging artificial intelligence applica-
tion: reinforcement learning issues on current internet of things. In: 2019 16th international
conference in information knowledge and technology, 2019. IEEE, Tehran, Iran
22. Alaziz SN, Albayati B, El-Bagoury AA, Shafik W (2023) Clustering of COVID-19 multi-
time series-based K-means and PCA With forecasting. Int J Data Warehous Min (IJDWM)
19(3):1–25. https://doi.org/10.4018/IJDWM.317374
23. Shokoor F, Shafik W, Matinkhah SM, Overview of 5G & beyond security. EAI Endorsed Trans
Internet Things 20228(30). https://doi.org/10.4108/eetiot.v8i30.1624
24. Shafik W, Tufail A (2023) Energy optimization analysis on internet of things. In: Advanced
technology for smart environment and energy. Springer International Publishing, Cham, pp
1–16. https://doi.org/10.1007/978-3-031-25662-2_1
25. Yang Z, Jianjun L, Faqiri H, Shafik W, Abdulrahman AT, Yusuf M, Sharawy AM (2021) Green
internet of things and big data application in smart cities development. Complexity. https://doi.
org/10.1155/2021/4922697
26. Wang Y, et al (2023) Service delay and optimization of the energy efficiency of a system in
fog-enabled smart cities. Alex Eng J 84:112–125. https://doi.org/10.1016/j.aej.2023.10.034
27. Shafik W, Matinkhah SM, Sanda MN. Network resource management drives machine learning:
a survey and future research direction. J Commun Technol, Electron Comput Sci 301–15.
https://doi.org/10.22385/jctecs.v30i0.312
28. Shafik W, Matinkhah M, Etemadinejad P, Sanda MN (2020) Reinforcement learning rebirth,
techniques, challenges, and resolutions. JOIV: Int J Inform Vis 4(3):127–135. https://doi.org/
10.30630/joiv.4.3.376
29. Shafik W, Mostafavi SA (2019) Knowledge engineering on internet of things through rein-
forcement learning. Int J Comput Appl 177(44):0975–8887. https://doi.org/10.5120/ijca20209
19952
Dissecting the Role of Women in Cybersecurity and Information … 349

30. Kassim K, et al (2024) Artificial Intelligence of Internet of Medical Things (AIoMT) in smart
cities: a review of cybersecurity for smart healthcare. In: Handbook of security and privacy of
ai-enabled healthcare systems and internet of medical things. pp 271–292. https://doi.org/10.
1201/9781003370321-11
31. Shafik W (2023) A comprehensive cybersecurity framework for present and future global
information technology organizations. In: Effective cybersecurity operations for enterprise-
wide systems. IGI Global, pp 56–79. https://doi.org/10.4018/978-1-6684-9018-1.ch002
32. Tadic B, Rohde M, Randall D, Wulf V (2023) Design evolution of a tool for privacy and security
protection for activists online: cyberactivist. Int J Hum-Comput Interact 39(1):249–271. https://
doi.org/10.1080/10447318.2022.2041894
33. Henshaw A (2023) Big data and the security of women: where we are and where we could be
going. In: Digital frontiers in gender and security. Bristol University Press, pp 13–41. https://
doi.org/10.51952/9781529226300.ch002
34. Brigstocke J, Fróes M, Cabral C, Malanquini L, Baptista G (2023) Biosocial borders: affective
debilitation and resilience among women living in a violently bordered favela. Trans Inst Br
Geogr. https://doi.org/10.1111/tran.12601
35. Salerno E (2023) Ut sacrificantes vel insanientes Bacchae: Bacchus’ Women in Rome. In: The
public lives of ancient women (500 BCE-650 CE) 2023 Feb 7. Brill, pp 173–193
36. Das S, Pooja MR, Anusha KS (2023) Implementation of women’s self-security system using
IoT-based device. In: Sustainable computing: transforming industry 4.0 to society 5.0 2023.
Springer International Publishing, Cham, pp 87–98. https://doi.org/10.1007/978-3-031-135
77-4_5
37. Royo MG, Parikh P, Walker J, Belur J (2023) The response to violence against women and fear
of violence and the coping strategies of women in Corregidora, Mexico. Cities 1(132):104113.
https://doi.org/10.1016/j.cities.2022.104113
38. Ebenezer V, Thanka MR, Baskaran R, Celesty A, Eden SR (2023) IOT based wrist band for
women safety. J Artif Intell Technol. https://doi.org/10.37965/jait.2023.0179
39. Mahalakshmi R, Kavitha M, Gopi B, Kumar SM (2023) Women safety night patrolling IoT
robot. In: 2023 5th International Conference on Smart Systems and Inventive Technology
(ICSSIT), vol 23. IEEE, pp 544–549. https://doi.org/10.1109/ICSSIT55814.2023.10060955
40. Srinivas Rao K, Divakara Rao DV, Patel I, Saikumar K, Vijendra Babu D (2023) Automatic
prediction and identification of smart women safety wearable device using Dc-RFO-IoT. J Inf
Technol Manag 1:34–51. https://doi.org/10.22059/jitm.2022.89410
41. Ranganayagi D, Saranya P, Sharmila MJ, Sujitha S, Nisha TA, Shanmugam K (2022) Pre-
eclampsia risk monitoring and alert system using machine learning and IoT. https://doi.org/10.
54646/bijg.006
42. Lamba J, Jain E (2023) Advanced cyber security and internet of things for digital transforma-
tions of the Indian healthcare sector. In: Research anthology on convergence of blockchain,
internet of things, and security. IGI Global, pp 1037–1056. https://doi.org/10.4018/978-1-6684-
7132-6.ch055
43. Gautam D, Purandare N, Maxwell CV, Rosser ML, O’Brien P, Mocanu E, McKeown C,
Malhotra J, McAuliffe FM, FIGO Committee on Impact of Pregnancy on Long-term Health
and the FIGO Committee on Reproductive Medicine, Endocrinology and Infertility. (2023).
The challenges of obesity for fertility: a FIGO literature review. Int J Gynecol & Obstet 160,
50–55. https://doi.org/10.1002/ijgo.14538
44. Gupta KP, Bhaskar P (2023) Teachers’ intention to adopt virtual reality technology in
management education. Int J Learn Chang 15(1):28–50. https://doi.org/10.1504/IJLC.2023.
127719
45. Hafeez A, Dangel WJ, Ostroff SM, Kiani AG, Glenn SD, Abbas J, Afzal MS, Afzal S, Ahmad
S, Ahmed A (2023) The state of health in Pakistan and its provinces and territories, 1990–
2019: a systematic analysis for the global burden of disease study 2019. Lancet Glob Health
11(2):e229–e243. https://doi.org/10.1016/S2214-109X(22)00497-1
46. Karandashev V (2023) Cross-cultural variation in relationship initiation. In: The oxford hand-
book of evolutionary psychology and romantic relationships. p 267. https://doi.org/10.1093/
oxfordhb/9780197524718.013.10
350 W. Shafik

47. Kramer KL (1868) Female cooperation: evolutionary, cross-cultural and ethnographic

evidence. Philos Trans R Soc B 2023(378):20210425. https://doi.org/10.1098/rstb.2021.0425
48. Li B, Xue C, Cheng Y, Lim ET, Tan CW (2023) Understanding work experience in epidemic-
induced telecommuting: the roles of misfit, reactance, and collaborative technologies. J Bus
Res 154:113330. https://doi.org/10.1016/j.jbusres.2022.113330
49. Mirzaasgari H, Momeni F, Pourshahbaz A, Keshavarzi F, Hatami M (2023) The role of psycho-
logical distress in the relationship between personality dimensions and pregnancy outcome of
women undergoing assisted reproductive treatment (IVF/ICSI). Iran J Psychiatry 2023:1–11
50. Pop RA, Hlédik E, Dabija DC (2023) Predicting consumers’ purchase intention through fast
fashion mobile apps: the mediating role of attitude and the moderating role of COVID-19.
Technol Forecast Soc Chang 186:122111. https://doi.org/10.1016/j.techfore.2022.122111
51. Wojtkowska A, Zaborski D, Modrzejewski A, Pastucha M (2023) Answer to commentary—
the effect of cosmetic surgery on mental self-image and life satisfaction in women undergoing
breast augmentation: an intermediate role of evaluating the surgery as one of the most important
life events. J Plast Reconstr Aesthet Surg 76:94–95. https://doi.org/10.1016/j.bjps.2022.10.017
52. World Economic Forum (2023) https://www.weforum.org/agenda/2022/09/cybersecurity-
women-stem/. Accessed 15 March 2023
53. Zaroo IF, Gani A (2023) Women’s leadership and workplace culture. in leadership and
workplace culture in the digital era. IGI Global 2023:227–241
A Comprehensive Exploration
of Advancements and Applications
of Digital Watermarking

Ealuri Rohith, Buggaveeti Padmaja, and V. M. Manikandan

Abstract The availability of multimedia content through digital networks is

becoming more and more dependent on the internet in the current era of information
technology and technical growth. As a result, there is a risk of unapproved digital
picture possession and use. Illegal tampering with and alteration of digital images
provide an extra risk. The ease accessibility of unlawful copying techniques like
piracy, counterfeiting, and fraud for digital multimedia is a drawback. An approach
to safeguarding digital image, and copyright is thus required. The existing difficulties
and issues that must be taken into account while looking for such a solution include
safeguarding ownership rights against illicit actions in a way that makes ownership
of the digital image unquestionable. These days, media like images, audio files, text
files, and videos are exchanged between sources for various reasons. In this process, a
lot of copyright violations are taking place. Various types of attacks that can compro-
mise the integrity of digital content are explored, including tampering, forgery, and
unauthorized distribution. Various methods of authentication exist, including facial
recognition, fingerprints, and passwords. Hence, copyright protection and digital
authentication have become the major concerns. Digital watermarking is the ideal
process that ensures not only copyright protection but also several other applications
like broadcast monitoring and owner authentication. To maintain the integrity of data,
watermarking is done which is inserting the imperceptible and inseparable informa-
tion into data. In order to assign copyright-related information to images, such as
the source, owner time-stamp, or even a logo or trademark, digital watermarking’s
aim and objective is to embed a digital mark into those images. The digital water-
marks act as an evidence for any copyright infringement case. It lets owners trace
the misuse of protected digital media. This paper provides a thorough overview of
digital watermarking techniques for copyright protection and digital authentication.

Keywords Copyright · Watermarking · Digital Watermark · Multimedia ·

Copyright Protection

E. Rohith · B. Padmaja · V. M. Manikandan (B)

Department of Computer Science and Engineering, SRM University-AP, Amaravati, Andhra
Pradesh, India
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 351
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_16
352 E. Rohith et al.

1 Introduction

The introduction of new technology exposed everyone to new threats. Through a

variety of sources and the network, digital information is transmitted very quickly.
The availability and accessibility of any information on the internet contributed to
the rise in the usage of digital media. It is now possible to save, transfer, and modify
digital media because of technological advancements. On the internet, it is quite
inexpensive for anybody to access or get the data they need. Anyone may duplicate
any piece of work and claim it as their own thanks to this. Many various pieces
of information from websites, research papers, journals, and other sources may be
found with just one search. The original work may be easily modified without any
permission. When the content is transmitted from one source to another, it is not
distorted in any way. These characteristics allow users to download any type of
digital media from many sources. However, users have the opportunity to copy the
original content and pass it off as their own due to the open availability of multimedia.
Therefore, the concept of employing a digital watermark to grant ownership rights
to the creator and safeguard their work.
The term Steganography comes from two Greek words, steganos and graphia
[1]. Steganos means covered and graphia means writing. This method of keeping
conversations confidential has been used for thousands of years. The first practical
use of steganography goes back to 440 BC, when messages were written on pieces
of wood and covered with wax to act as a covering medium. Invisible ink was
employed during World War I to secretly write messages on paper that could only
be seen by the person who wrote them. Later, they read the message and darkened
the ink using a variety of substances, including milk and vinegar. In this fashion,
steganography was a common practice in the past. Even today, there are several
practical and technological uses for this approach.
Steganography [2] is the primary source of inspiration for digital watermarking
methods. Digital watermarking and steganography are both methods of information
hiding, however they serve very different purposes. In digital watermarking, the
internal data, which is the watermark, is supplementary data that is utilized to secure
the external, vital information. In contrast to digital watermarking, steganography
uses internal data as the primary information and external data as the dummy data.
Steganography techniques essentially conceal the approved information inside the
digital medium before sending it to the receiver. The overview of steganography
algorithm is shown in Fig. 1.
The original image, which is the image without any embedding done, is called
the cover image. There are numerous algorithms for the steganography process. Any
one of the algorithms is performed to hide the message in the original image. After
performing such an operation, the image is called a stego image. Messages can be
embedded in cover images and stego-images with the help of a key, or the messages
can be derived from the photos themselves. Steganography has its different types.
Text steganography secretly stores information and in this method, the hidden data is
encoded into each letter of the word. In the audio steganography, the data is hidden in
A Comprehensive Exploration of Advancements and Applications … 353

Fig. 1 Overview of steganography method

sound. This is digitally used to protect the data against unauthorized reproduction. In
the video steganography, the data is secretly embedded within a video file. Generally,
Discrete Cosine Transform(DCT) is commonly used to insert the values to hide the
data in each image in the video. It is invisible to the naked eye. Protocol or Network
steganography is hiding the data by using a protocol like TCP, UDP, or IP.

2 Digital Watermarking

The paper business gave rise to watermarking in the thirteenth century. One of the
earliest use of watermarking was to identify the mill that manufactured the paper
and the brand in order to verify its legitimacy. These days, watermarks are applied
on currency, paper, stamps, and a long list of other items. In order to identify fake
copies, Komatsu and Tominaga developed a digital watermarking method back in
1989. They inserted a hidden label into a copy, and if the registered owner matches
it, it can be verified that the person holding the document is the owner. Although
watermarking has a lengthy history dating back to the thirteenth century, it wasn’t
until after 1990 that watermarks were digitized and began commonly employed. The
history of watermarking and its evolution is explained thoroughly in [3].
Digital watermark is a signal that is embedded in a digital media like audio file,
text file, image, or a video file. It is permanently embedded in the data and can
be detected and extracted later to perform any modifications or operations. The
watermark is hidden in the original data in such a way that it is inseparable from
the data. Embedding a digital watermark into an original work still lets the owner
access the data. The host data is intended to have a watermark forever. The data can
be extracted to fully define the owner if the ownership of a digital work is in doubt.
354 E. Rohith et al.

Fig. 2 Digital watermarking scheme

Traditional watermarks are visible to users. Some are printed on the image, some
are kept on the video. These are technically the visible watermarks. The digital
watermarks, unlike the traditional ones, are designed in such a way that they are not
visible to the users. The watermark which contains patterns of bits is embedded and
scattered all over the image to avoid modification.
The image where we are embedding the watermark into is called the host image.
After the watermarking process, the image is defined as the watermarked image.
During the process of embedding, a secret key is used to have a secured watermark.
The image having the watermark is transmitted along the communication stream.
On the receiver side, the embedded can be extracted. This watermarking process
as a whole has been developed to identify the creator, receiver, and sender of the
information. The encryption process is where the content is hidden and is made
invisible to the reader without a specific key or code from the owner. After the
encryption process, watermarking can be used for many applications like copyright
protection, digital authentication, fingerprinting, broadcast monitoring, and many
such [4] (Fig. 2).

2.1 Classification of Digital Watermarks

Different criteria are used to categorize digital watermarking processes. Protection

of privacy and authenticity has grown to be a major issue and difficulty as a result
of the growth of different media assets online. Digital watermarking methods of
many kinds were developed to address this issue. These systems are used in several
applications with a variety of purposes.
A Comprehensive Exploration of Advancements and Applications … 355

We are aware that watermark embedding and watermark extraction are the two
key steps in the watermarking procedure. The varieties of watermarking technologies
are categorized according to the features of embedded watermarks.

2.1.1 Robustness of the Watermark

The word “robustness” [5] describes the watermark’s capacity to withstand and
withstand any tampering. Attacks are deliberate attempts to remove the watermark,
while others aim to modify it. Watermarks are divided into three categories for the
embedding process, which is the initial stage of the watermarking process.
Watermarks that are robust are made to withstand all types of image processing
operations. These attacks comprise techniques like cropping, filtering, compressing,
and more. The information included in the watermark may be removed in these
unauthorized assaults, or new information may be added.
A semi-fragile watermark performs a somewhat different task than a robust one.
Any unauthorized alterations to the watermark can be detected using these kinds
of watermarks. However, these also permit particular image-processing techniques.
Semi-fragile watermarks can distinguish between destructive alterations and standard
image-processing techniques.
For comprehensive authentication, fragile watermarks are utilized. Authentic
images are those that haven’t been altered in any way. The best option for this is
fragile watermarks. They look for any unauthorized changes to the embedded water-
mark. The watermark is fragile and will be lost if the image is slightly modified.
Therefore, the primary goal of these watermarks is to verify the validity of the image.

2.1.2 Blind and NonBlind

The watermark must then be extracted after being embedded. This method involves
a variety of strategies. Watermarks are categorized as blind or non-blind based on
the requirement to access the original image. The watermarking method is said to as
blind if it does not require knowledge of the original, unwatermarked audio, video,
or picture file. A method is regarded to be non-blind if it requires any original data in
order to extract the watermark. Since recovering the watermarked image is obviously
simple if the unwatermarked data is available, the non-blind technique generally
looks to be more effective than the blind one. The watermark detector, however,
does not always have access to the original data in numerous cases. Therefore, blind
watermarking methods have an edge over non-blind ones.

2.1.3 Visibility

A watermark can be anything, such as a logo, a picture, or a section of text. They are
divided into visible and invisible watermarks depending on how clear the watermark
356 E. Rohith et al.

is on the original data. The visible watermark, as its name implies, is something that
is evident on the image in any format. On the other hand, the invisible watermark
is incorporated into the original image and is not noticeable to the human eye. One
benefit of a visible watermark over an invisible one is that it prevents unauthorized
usage of your image from any search engine since it is easy to recognize. Additionally,
we may advertise our firm and assert our copyright by using visible watermarks. The
visible watermark is simple to embed into a picture, however the invisible watermark
embedding is fairly challenging and requires advanced algorithms. There are more
uses for visible watermarks than invisible ones.

2.1.4 Based on the Domain

We attempt to integrate as much information as we can into the original image for
a given amount of distortion. Recently, numerous different watermarking systems
have been created. These methods are divided into two categories: frequency domain
watermarking and spatial domain watermarking, depending on the domain in which
the watermark is inserted. When watermarking in the spatial domain, the pixels are
changed right there in the image’s spatial domain. This makes the process simple
because there is no need to extract the watermark. Through various transformations
like discrete fourier transform(DFT), discrete cosine transform(DCT), and discrete
wavelet transform(DWT), the watermark is inserted into the frequency coefficients
of the altered picture in frequency domain watermarking. The lower and intermediate
frequencies are where most of the energy is concentrated. The embedded watermark
is not evenly spread throughout the image in the spatial domain, making it simple
to remove the watermark. On the other hand, the watermark is dispersed across the
image in the frequency domain, thus the alterations made have less of an impact on
the image.

2.1.5 Private and Public

Depending on where the watermark is, private and public watermarks can be distin-
guished. A watermark is considered private, as the name implies, if only authorized
users are able to recognize it. These watermarking methods make the watermark
undetectable and inaccessible to unauthorized users. A private key that specifies the
location of the watermark is used to extract the watermark. As long as the location is
known, the user can insert or delete the watermark by using a private key. A public
watermark, on the other hand, is one whose location is well known to everybody.
When a public watermark is inserted, analyzing the entire image makes it easier to
find and remove it. Private watermarking techniques are more effective than public
ones, where a hacker may quickly find and remove the watermark without permission.
A Comprehensive Exploration of Advancements and Applications … 357

2.2 Properties of Digital Watermark

Digital watermarking has become much more widely used in daily life in recent
years. This is because of its features, which guarantee safe data transmission and
preserve content ownership [6].
• The resemblance between the watermarked and original image is what defines
a watermark’s fidelity. The user should be unable to see the image’s embedded
watermark. Additionally, it maintains the similarity between the watermarked
image and the original image. The degree to which a watermark degrades the
original image’s clarity and quality is known as its fidelity. Additionally, it deter-
mines how much the original content has been degraded once the watermark has
been inserted. Therefore, high fidelity watermarks shouldn’t specifically lower the
quality of original content. Different measures, such as the bit error rate (BER) or
the structural similarity index (SSIM), which calculate the differences between
the original and watermarked content, can be used to assess fidelity.
• Imperceptibility is another important aspect of digital watermarking. The water-
marked image should blend in with the original image once it has been embedded.
Digital watermarking is mostly used to safeguard intellectual property. There-
fore, anyone can simply remove a watermark that has been embedded and steal
the work while it is visible. Therefore, a watermark’s imperceptibility quality
determines its invisibility and is crucial for preserving the integrity of the orig-
inal content. To increase imperceptibility in particular contexts, many signal
processing algorithms have been developed.
• When a digital watermark is said to be robust, it means that the user can still read
it even after certain adjustments have been made to the original content. Water-
marks with high robustness are undoubtedly those that can resist various forms of
degradation, including compression, cropping, filtering, and other transmission-
related manipulations. The watermark must be able to withstand any attempt to
reproduce the original work or to reduce the original content’s quality in order to
preserve its copyright.
• The quantity of data that can be inserted in any digital medium using a specific
technique is referred to as a digital watermark’s capacity. The strength of a digital
watermark can vary depending on a number of elements, including the type of
signal processing, the kind of digital media, the embedding technique employed,
and the degree of robustness the watermark contains. When watermarking images,
the amount of image bits that can be inserted into the original image is utilized for
evaluating the watermark’s capability. The length or frequency band of the water-
mark signal that can be inserted into the audio stream is the unit of measurement
for audio watermarking. Again, regardless of the capacity of the watermark, the
perceived quality and resilience of the image should not be compromised. In many
applications, including data concealing and content identification, the watermark’s
capability is a crucial consideration. The watermark capacity is determined by the
limitations of the application.
358 E. Rohith et al.

• The integrity and novelty of original content is preserved by the security of a

digital watermark. It safeguards the information from tampering or unauthorized
access. It can be done using a variety of methods, including encryption and digital
signatures. By converting the watermark into a secret key that can only be decoded
by the authorized user, encryption is accomplished. Through this procedure, the
watermark is protected even from unauthorized users. A digital watermark’s secu-
rity is frequently verified using digital signatures. The watermark data’s unique
digital signature is calculated using a cryptographic technique. Only authorized
users are capable of verifying the digital signature. Several other techniques can
be employed to secure digital watermarks in addition to these. Any watermarking
method should be designed and put into use with security as one of its primary
considerations.

2.3 Applications of Digital Watermarking

In order to digitally authenticate and safeguard intellectual property, researchers

first looked into the method of embedding digital watermarks into digital media
like photos, audio files, text files, and video files in 1990. Digital watermarking
was necessary due to the increased use of technology and the internet. Numerous
digital watermarking techniques have been designed as a result of the increasing
importance of multimedia and forensics. As data distribution and access become
more convenient, there are more options for safeguarding and authenticating digital
content that were developed.
The most effective tamper detection methods apply digital watermarking. This
indicates that we can tell whether or not the material has been tampered with when
it is accessed or viewed by employing these strategies. A watermark is included in
the data when the content is produced. It is possible to identify data tampering using
the information in this watermark. A comparison is made between the watermark
and the original watermark when the digital content is accessed. It may be said that
the material has not been altered if the watermark matches the original one that
was added at the beginning. It is a sign that the information has been altered if the
watermarks do not match. Therefore, data manipulation may be identified via digital
watermarking.
Forensic analysis depends heavily on digital watermarking. The source of the
leaked or pirated information might be found during the examination of legal proceed-
ings and used as evidence. Each replica of the digital file has a digital watermark
built in. Finding the source of the content leak and the first location where it was
disseminated can be made easier with this information. To monitor the circulation
of original content, these watermarks can also be employed. It also offers a history
of who and when the data was accessed. Digital watermarking methods thus offer a
number of benefits for case-specific research.
Copyright violations and theft of works are frequent occurrences in our everyday
lives. It is simple for hijackers to take our work and claim it as their own if there
A Comprehensive Exploration of Advancements and Applications … 359

is no protection protecting the information or work we produce. For example, a

well-known photographer shared a picture of a stunning landscape on his website.
Anyone can take the photo of them and give it to you with their name on it without
providing any documentation or ownership rights. The image will always belong to
us if we add a watermark, and no one else will be able to take it. It would also be a
promotion for his work if the watermark was added in a way that made it noticeable.
A watermark can also be incorporated in and used to secure text, video, and audio
files.
The source of the image and authorized users are both recognized by watermarks
for copyright protection. The printing and publishing industries are quickly adopting
this method of inserting copyrights. There are several networks that deliver videos.
Given how simple it is to reproduce videos without sacrificing quality, it is crucial
to preserve their copyrights. The video stream between the transmitter and recipient
might have a watermark added to it. We can track and identify the various copies
of videos in this way. We have access to both visible and invisible watermarks. A
complicated embedding procedure is needed for invisible watermarks. Watermarks
that can be seen by the naked eye and are clearly distinguished are considered visible.
Both watermarks, however, guarantee the data’s ownership and safeguarding.
The old-fashioned way to authenticate a document is to use a handwritten signa-
ture. There is no way to copy a fixed signature from one document to another.
The legitimacy of two signatures that appear to be identical can be determined by
comparing them. An exact duplicate of a handwritten signature exists in digital signa-
tures. These methods of data authentication are still even more powerful. Using two
keys—public and private—that are mathematically connected to one another, the
digital signatures may be encrypted. Utilizing the public key, users may validate the
digital signature; however, the secret key is necessary to create the exact identical
signature. Therefore, digital watermarking is essential for digital authentication.
In conclusion, digital watermarking is an effective tool for safeguarding and
confirming the validity of digital content. It has a wide variety of applications in
several sectors.

3 Related Work

In the current digital era, digital watermarking techniques have become a popular way
to guarantee the validity, integrity, and copyright protection of multimedia informa-
tion. Previous research on digital watermarking has looked at a variety of approaches,
including both spatial and transform domain techniques, to embed and remove
watermarks while keeping the caliber and perceptual integrity of the host multi-
media content. Numerous issues with digital watermarking have been addressed
by researchers, including resilience against signal processing processes, capacity
optimization, resistance to assaults, and adaptation to various multimedia formats,
which has resulted in the creation of novel methods and frameworks. This section
360 E. Rohith et al.

will explore a number of methods that have been suggested for the purposes of digital
authentication and copyright protection.

3.1 Background Theory

To understand the digital watermarking algorithms, there are several mathematical

theories to be learnt.

3.1.1 Discrete Fourier Transform(DFT)

Signals in the frequency domain may be analyzed and modified using the Discrete
Fourier Transform. This enables the signal to separate into its component frequencies.
Finding certain frequency ranges that are appropriate for inserting the watermark is
made easier with the use of frequency analysis. The DFT generates a frequency
representation made up of N evenly spaced frequency bins when applied to a finite
series of discrete samples. The DFT result can be translated to magnitude and phase
data or shown as a complex-valued spectrum. Application areas for the DFT include
feature extraction, data compression, spectrum analysis, and filtering. The DFT is
frequently used in the context of digital watermarking to examine the frequency
components of the host signal and embed or extract watermarks in the transformed
domain.
Mathematically, DFT is defined as:

N −1
 
Xk = X n e−2 ikn/N

n=0

where N is the length of the signal, X k represents the complex-valued spectrum of

the signal at frequency index k, n is the sample input.

3.1.2 Discrete Cosine Transform (DCT)

Similar to the Discrete Fourier Transform, DCT converts signal from spatial domain
to frequency domain. The DCT is frequently used in digital watermarking because of
its capacity for energy compression. The signal’s energy is concentrated in a limited
number of low-frequency coefficients when the DCT is applied to an image or video
frame, while the energy of the higher-frequency coefficients is reduced. Because
of this characteristic, watermarking algorithms can incorporate the watermark in
lower-frequency coefficients, where changes are less obvious to human observers.
A Comprehensive Exploration of Advancements and Applications … 361

Mathematically,

F(u, v) = (2/N )1/2 (2/M)1/2

N −1 M−1
     
·u ·ν
∧(i ) ∧ ( j ) · cos (2i + 1) cos (2 j + 1) · f (i, j )
i=0 j=0
2·N 2·M

where f (i,j) is the intensity of the pixel in row i and column j and F(u,v) is the DCT
coefficient in row k 1 and column k 2 of the DCT matrix.

3.1.3 Singular Value Decomposition

A matrix is divided into singular values and related singular vectors using the matrix
factorization technique known as singular value decomposition (SVD). It is a funda-
mental idea in linear algebra and has several uses in the domains of data analysis,
picture compression, and signal processing. JPEG and other image compression
methods frequently employ SVD for effective picture storage and transmission.
Applying SVD to picture data allows for compression without noticeably sacrificing
visual quality since it identifies the prominent spatial frequencies and allows for the
representation of the image with fewer coefficients. By contrasting the single values
or vectors of the original material with those of the received or accessible content,
SVD is used for digital authentication. The identification of content tampering, alter-
ation, or unauthorized change is made possible by this comparison. A way for content
authentication and integrity checking, any appreciable divergence in the single values
or vectors indicates probable unauthorized content alterations. SVD is a flexible
method for digital watermarking that offers insights into the energy and frequency
properties of signals or pictures. It allows for the embedding of watermarks in regions
that are invisible as well as their extraction during the detection procedure. The robust-
ness and security of watermarks may be increased, making them more resistant to
assaults and distortions, by making use of the energy compaction capabilities of
SVD.

3.1.4 Error Correction Codes (ECC)

To increase the stability and dependability of the embedded watermark, error correc-
tion codes (ECC) are frequently utilized in digital watermarking. In order to detect
and fix mistakes that can happen during transmission, storage, or assaults on the
watermarked information, ECC algorithms provide redundancy to the watermark
data. Before embedding, several error correcting codes are added to the watermark
data, doubling the amount of information. The same error correction algorithms that
are used to extract watermarks are also utilized to find and fix faults in the extracted
362 E. Rohith et al.

watermark. The reduction of noise, signal distortions, data tampering, and attacks
on watermarked information is made possible by the error correction procedure.
Multiple digital watermarking methods for copyright protection and digital
authentication are thus influenced by various mathematical ideas.

3.2 Evaluation Measures

Researchers and practitioners may objectively evaluate and compare various water-
marking algorithms, optimize their parameters, and select the approach that best fits
their unique needs by taking assessment measures into consideration. Evaluation
measures are crucial to the watermarking process since they assure the reliability,
excellence, security, and usability of digital watermarking systems.
PSNR: Peak Signal-to-Noise Ratio (PSNR) is a commonly used statistic for assessing
how well a picture has been compressed or reconstructed in comparison to its original
state. It calculates the ratio of the power of the noise added during compression or
reconstruction to the power of the greatest available signal (usually the maximum
possible pixel value). The PSNR, which is measured in decibels (dB), offers a number
reflecting how faithfully and precisely the original picture is preserved after being
compressed or rebuilt. Lower numbers imply more obvious artifacts and image
quality loss, whereas higher PSNR values suggest greater quality and less distortion.
In order to optimize the trade-off between compression ratio and picture quality,
PSNR is frequently used in image compression algorithms.
SSIM: A popular picture quality statistic called the Structural Similarity Index
(SSIM) goes beyond a straightforward pixel comparison to take into consideration
how structurally similar the original and deformed images are. By comparing the
similarity of their brightness, contrast, and structural elements, SSIM assesses the
perceived picture quality. It takes into account the local structural data and records
pixel dependencies. A score between 0 and 1 is generated by SSIM; a value closer
to 1 denotes a higher degree of similarity and better picture quality. Given that it
takes both global and local structural information into account, it is very helpful in
evaluating picture restoration strategies.
MSE: Measuring the differences between an original picture and a compressed or
reconstructed version of the same image is frequently done using the Mean Squared
Error (MSE) metric. The average of the squared discrepancies between the pixel
values of the original and the compressed or reconstructed picture is calculated
using a mathematical formula. Lower values represent higher picture quality and
less distortion, and MSE is given as a numerical value. In order to balance the trade-
off between compression ratio and picture quality, MSE is frequently used in image
and video compression algorithms. It does have some drawbacks, though, such as its
susceptibility to outliers, which can skew the evaluation of image quality as a whole.
A Comprehensive Exploration of Advancements and Applications … 363

NCC: When comparing two datasets or signals, a statistic called the Normalised
Correlation Coefficient (NCC) is employed. It is often used in a variety of disci-
plines, including pattern recognition, image analysis, and signal processing. When
calculating the correlation between two variables, NCC takes into consideration the
means and standard deviations of each variable. The NCC algorithm generates a
result between −1 and 1, where 1 denotes a perfect positive correlation, −1 denotes
a perfect negative correlation, and 0 denotes no correlation at all between the datasets.
Greater similarity or correlation between the datasets under comparison is indicated
by a higher NCC score.
BER: Bit Error Rate is referred to as BER. It is a metric used to assess a digital
communication system’s or channel’s level of quality. The percentage of bits that
are wrongly received to all bits broadcast is known as the bit error rate. The quality
of the communication system or channel is inversely correlated with the BER value.
A greater BER implies a higher rate of transmission mistakes, which can lead to a
loss of data integrity and a degradation in performance. The effectiveness of error
correction methods, signal-to-noise ratios, and overall system performance are all
evaluated using BER, which is often used in telecommunications, networking, and
digital data transfer. It is a crucial factor in determining how reliable and effective
digital communication systems are.

3.3 Algorithms Proposed

In [7], a DWT-SVD based robust digital watermarking scheme was proposed for
the security of medical images. In this method, the picture capture data and the
electronic patient record will make up the watermark. The Electronic Patient Record
hash will be appended to the watermark in order to increase security and ensure
data integrity. The watermark employed in this method has three components; the
first and second sections contain data about the patient and the method of picture
capture, allowing the user to determine who created the image and where it came
from. The third portion, which will be placed in binary form for insertion, is created
by adding the first two parts. DWT is first applied and the LL subband is subjected
to SVD, and the watermark bits are appended to the coefficients of the resultant
S matrix in order to incorporate the watermark. The adjusted LL subband is then
obtained by applying an inverted SVD. The watermarked medical picture is then
obtained by performing an inverse DWT. The watermark is extracted using the S
matrix’s subsequent coefficients. For the first variation, the watermark’s bits will
match the parity of subtracting two consecutive coefficients. In the second variation,
the watermark’s bits (X and Y) will match the parity of subtracting three sequential
coefficients. After the extraction procedure, the changed matrices are transformed
into extracted bits. This binary sequence will be split into three parts: the patient
information will be in part one, the picture acquisition information will be in part
two, and the hash of the first two parts will be in part three. The extracted hash and the
364 E. Rohith et al.

hash produced by concatenating the first two sections are now compared in order to
confirm the integrity. Since this method is used to create medical images, the image
quality needs to be preserved. The likelihood of several distortions is great. Several
assessment metrics are taken into consideration in order to gauge how imperceptible
the aforementioned strategy is. PSNR, or peak signal to noise ratio, is used to assess
the deterioration brought on by the embedding of watermarks into images. A high
PSNR value suggests that the distortion is reduced. The combination of DWT and
SVD results in variations with high PSNR values, which improve imperceptibility.
A metric used to compare two photographs is called the Structural Similarity Index
(SSIM). It is used to compare the compressed picture’s visual quality to the original
image. When the SSIM value is close to 1, it indicates that the visual quality is high.
The proposed scheme observed a SSIM value closer to 1.
Using cryptographic techniques, a digital watermarking scheme for copyright
protection is introduced in [8]. The two basic steps in every digital watermarking
system are embedding and extraction. For the embedding step in this technique,
DWT and SVD methods are combined. For robustness verification, a cryptographic
technique is employed. RSA is an asymmetric cryptographic technique which has
two keys called public and private. At the transmitter, the data is encrypted using the
public key, and at the receiver, it is decrypted using both keys. First, a QR code with
the name and country is produced. When private key values are entered into the RSA
algorithm, a public key and an encrypted message are produced. In order to water-
mark data securely, a QR code is scrambled using Chaotic Logistic Map(CLM). The
watermarked picture is imported and transformed to its red, green, and blue compo-
nent parts. The blue layer is taken into account, and the Haar wavelet decomposition
is used to produce the four subbands LL, LH, HL, and HH. The LL subband and the
jumbled QR picture are dissected using the SVD method. The watermarked singular
values are created by taking into account the singular values of both photos and
combining them with a key value. A watermarked LL subband is produced using
an invertible SVD approach. The watermarked blue layer for one level is produced
similarly by fusing watermarked LL subbands with other subbands using an inverse
DWT. A blue layer is combined with the red, green, and other layers to produce a
watermarked color picture. The recipient receives the watermarked picture together
with the public key and key value. The watermarked image is initially inspected
in order to begin the extraction procedure. The red, green, and blue layers of the
colored watermarked picture are created. The blue component, where the watermark
is incorporated, is taken into consideration for extraction. A Haar wavelet is used
with a one-level DWT to apply the blue component. The LL subband is taken into
account, and a singular value matrix is produced using SVD. Based on the important
image’s key values and incomplete data, a scrambled QR watermark is extracted.
The CLM technique is used to reverse scramble the retrieved watermark. The public
key, which contains the private key values, is used to extract the watermark from the
encrypted message in order to validate the watermarked data. Peak-Signal-to-Noise
Ratio and Normalised Correlation Coefficient are taken into consideration during
the evaluation process. The metric used to determine the relationship coefficient
between the original watermark and the extracted watermark is called NCC. NCC
A Comprehensive Exploration of Advancements and Applications … 365

approaches 1 when the extracted watermark is equivalent to the original watermark.

In comparing the suggested scheme to several previous schemes, it is found that the
NCC value is equal to 1, indicating that the watermark that is accurately retrieved is
identical to the original watermark.
To ensure copyright protection, a digital watermarking scheme through encryp-
tion and DWT is proposed in [9]. Using a collection of wavelet functions, the
discrete wavelet transform separates a signal into several frequency components. The
DWT additionally records the temporal or geographical localization of the frequency
components, in contrast to other transforms like the Fourier Transform, which just
offers frequency information. Due to its capability to give a multi-resolution represen-
tation of signals and pictures, capturing both local details and global properties, the
Discrete Wavelet Transform is favored in digital watermarking. It offers flexibility in
balancing transparency and robustness needs by allowing watermark implantation in
particular frequency bands. In this algorithm, a 90 × 90 pixel grayscale baboon image
is used as a watermark image. Row and column rotation using the watermark image’s
random number generator key k is the basis of an image encryption technique. The
initial input picture is divided as the following phase. To produce the proper scaled
picture with decreased size, two-dimensional DWT is employed. The pixel point at
the decomposed input picture is determined for inserting the decomposed watermark
image based on Euclidean distance. The perceptibility of the input picture remains
unchanged if there is a great degree of resemblance between it and the watermark
image. This ultimately makes the watermark stronger. Using the below equation, the
encrypted watermark is added to the source picture.

y(i, j) = (1 − alpha) × i (i, j ) + alpha × i w(i, j )

The visibility coefficient is represented by alpha in the equation. i(i,j), iw(i,j), and
y(i,j) correspondingly represent the DWT coefficients of the corresponding decom-
posed input image, watermark image, and watermark embedded output image. The
success of this encryption hinges on how closely the input photos that have been
deconstructed resemble the encrypted watermark image. The reversal of the embed-
ding procedure allows for the extraction of the watermark. Since the suggested
method uses a non-blind technique, the extraction procedure needs both the original
picture and the encryption key. When comparing the extracted watermark picture to
the original watermark image, the Mean Square Error (MSE), Normalised Corre-
lation Coefficient (CC), and Peak Signal to Noise Ratio (PSNR) are employed as
indicators of resemblance. It was observed that the proposed scheme has a higher
PSNR value than the already existing schemes.

3.4 Results Comparison

See Table 1.
366 E. Rohith et al.

Table 1 Comparison of different algorithms and their results

Reference Algorithm used Results
no
[9] Digital Image Watermarking Through Encryption and DWT for PSNR—54.96 dB
Copyright Protection MSE—0.2071
NCC—0.97
[7] A DWT-SVD based robust digital watermarking for medical PSNR—57.41 dB
image security MSE—0.13
SSIM—0.9998
[8] Digital Watermarking System for Copyright Protection and NCC—0.9949
Authentication of Images Using Cryptographic Techniques PSNR—37.16 dB
[10] Robust Color Images Watermarking Using New PSNR— > 44 dB
Fractional-Order Exponent Moments NCC—0.9759
BER—0.0195
[11] Robust and distortion control dual watermarking in LWT PSNR—32.85 dB
domain using DCT and error correction code for color medical NCC—0.9913
image BER—0
[12] Robust blind watermarking approach against the compression PSNR—41.30 dB
for fingerprint image using 2D-DCT BCR—100%
[13] Machine learning based blind color image watermarking PSNR—40.25 dB
scheme for copyright protection SSIM—0.9976
[14] A robust hybrid SVD-based image watermarking scheme for PSNR—51
color images SSIM—0.9936
NC—1
[15] A Novel Reversible Watermarking Scheme based on SHA3 for PSNR—Inf
Copyright Protection and Integrity of Satellite Imagery SSIM—1

4 Conclusion

In conclusion, there is a growing need for adequate safeguards against unauthorized

access to, use of, and manipulation of digital pictures as a result of the dissemination
of multimedia content’s increased reliance on digital networks. In order to secure
ownership rights and handle the problems caused by illegal activity in the digital
sphere, copyright protection is essential. Robust solutions are required because a
variety of threats, such as tampering, forgery, and unauthorized dissemination, can
jeopardize the integrity of digital assets.
With a variety of uses, including broadcast monitoring and owner verification,
digital watermarking emerges as the appropriate method for copyright protection
and digital authentication [16–23]. Watermarking contributes to the preservation
of data integrity by introducing undetectable and irremovable information into data.
Digital watermarking’s main goal is to incorporate copyright-related data into photos,
including the source, owner timestamp, or even a logo or trademark. These digital
markings allow owners to track the unauthorized use of protected digital media and
are used as evidence in copyright infringement lawsuits.
A Comprehensive Exploration of Advancements and Applications … 367

The numerous digital watermarking methods for copyright protection and digital
authentication have been thoroughly examined in this chapter. To effectively handle
the always changing difficulties of digital content protection, it is essential to keep up
with the most recent developments in digital watermarking. We can promote a more
secure and reliable digital environment for artists, distributors, and customers alike
by putting effective watermarking techniques into practice and increasing awareness
of the significance of copyright protection.

References

1. Siper A, Farley R, Lombardo C (2005) The rise of steganography. In: Proceedings of student/
faculty research day, CSIS, Pace University: D1
2. Yahya A, Yahya A (2019) Introduction to steganography. Steganography Tech Digit Images:1–7
3. Alabdali N, Alzahrani S (2021) An Overview of Steganography through History. Int J Sci Eng
Sci, Kingd Saudi Arab:41–44
4. Mohanarathinam A et al (2020) Digital watermarking techniques for image security: a review.
J Ambient Intell Humaniz Comput 11:3221–3229
5. Kadian P, Arora SM, Arora N (2021) Robust digital watermarking techniques for copyright
protection of digital data: a survey. Wireless Pers Commun 118:3225–3249. https://doi.org/10.
1007/s11277-021-08177-w
6. Embaby AAl, Wahby Shalaby MA, Elsayed KM. Digital watermarking properties, classifica-
tion and techniques. Int J Eng Adv Technol (IJEAT) 9.3:2742–2750
7. Zermi N, Khaldi A, Kafi R, Kahlessenane F, Euschi S (2021) A DWT-SVD based robust
digital watermarking for medical image security. Forensic Sci Int 320:110691, ISSN 0379-
0738.https://doi.org/10.1016/j.forsciint.2021.110691
8. Sanivarapu PV, Rajesh KNVPS, Hosny KM, Fouda MM (2022) Digital watermarking system
for copyright protection and authentication of images using cryptographic techniques. Appl
Sci 12(17):8724. https://doi.org/10.3390/app12178724
9. Ambadekar SP, Jain J, Khanapuri J (2019) Digital image watermarking through encryption and
dwt for copyright protection. In: Bhattacharyya S, Mukherjee A, Bhaumik H, Das S, Yoshida
K (eds) Recent trends in signal and image processing. Advances in Intelligent Systems and
Computing, vol 727. Springer, Singapore
10. Hosny KM, Darwish MM, Fouda MM (2021) Robust color images watermarking using new
fractional-order exponent moments. IEEE Access 9:47425–47435. https://doi.org/10.1109/
ACCESS.2021.3068211
11. Singh AK (2019) Robust and distortion control dual watermarking in LWT domain using DCT
and error correction code for color medical image. Multimed Tools Appl 78:30523–30533.
https://doi.org/10.1007/s11042-018-7115-x
12. Lebcir M, Awang S, Benziane A (2022) Robust blind watermarking approach against the
compression for fingerprint image using 2D-DCT. Multimed Tools Appl 81:20561–20583.
https://doi.org/10.1007/s11042-022-12365-6
13. Sinhal R, Jain DK, Ansari IA (2021) Machine learning based blind color image watermarking
scheme for copyright protection. Pattern Recognit Lett 145:171–177
14. Baba Ahmadi SB, Zhang G, Jelodar H (2019) A robust hybrid SVD-based image watermarking
scheme for color images. In: 2019 IEEE 10th annual information technology, electronics and
mobile communication conference (IEMCON), Vancouver, BC, Canada, pp 0682–0688.https://
doi.org/10.1109/IEMCON.2019.8936229
15. Kunhu A, Al Mansoori S, Al-Ahmad H (2019) A novel reversible watermarking scheme based
on sha3 for copyright protection and integrity of satellite imagery. Int J Comput Sci Netw Secur
19:92–102
368 E. Rohith et al.

16. Begum M, Uddin MS (2020) Digital image watermarking techniques: a review. Information
11.2:110
17. Abadi RY, Moallem P (2022) Robust and optimum color image watermarking method based
on a combination of DWT and DCT. Optik 261:169146
18. Feng B, et al (2020) A novel semi-fragile digital watermarking scheme for scrambled image
authentication and restoration. Mob Netw Appl 25:82–94
19. Evsutin O, Melman A, Meshcheryakov R (2020) Digital steganography and watermarking for
digital images: a review of current research directions. IEEE Access 8:166589–166611
20. Ray A, Roy S (2020) Recent trends in image watermarking techniques for copyright protection:
a survey. Int J Multimed Inf Retr 9(4):249–270
21. Alzahrani A (2022) Enhanced invisibility and robustness of digital image watermarking based
on DWT-SVD. Appl Bionics Biomech 2022
22. Qiu Y, Sun J, Zheng J (2023) A self-error-correction-based reversible watermarking scheme
for vector maps. ISPRS Int J Geo Inf 12(3):84
23. Tarhouni N, Charfeddine M, Ben Amar C (2020) Novel and robust image watermarking for
copyright protection and integrity control. Circuits, Syst, Signal Process 39:5059–5103
Optimizing Drug Discovery: Molecular
Docking with Glow-Worm Swarm
Optimization

Vijaya Sindhoori Kaza, P. R. Anisha, and C. Kishor Kumar Reddy

Abstract In this chapter, we investigate the application of the Glow-worm Swarm

Optimization (GSO) algorithm in molecular docking, utilizing the 3 × 29 dataset after
thorough preprocessing and cleaning. We delve into the principles of protein–ligand
interactions and introduce the GSO algorithm, inspired by glow-worm behavior,
highlighting its core mechanisms encompassing movement, light intensity, and
attraction–repulsion. The implementation of GSO for molecular docking involves
fine-tuning its parameters to optimize performance, and we present a detailed
performance evaluation, benchmarking GSO against traditional algorithms. Through
computational techniques and scoring functions, we analyze docking accuracy,
convergence, and efficiency. Additionally, we explore the integration of machine
learning models for enhanced accuracy, compare machine learning and classical
docking approaches, and identify potential drug candidates based on the outcomes.
Real- world applications of molecular docking using the 3 × 29 dataset, such
as structure-based drug design and virtual screening, are showcased. The chapter
concludes with insights into future directions and challenges, considering advance-
ments in the field, integration with artificial intelligence and deep learning, and
addressing dataset bias and generalization, showcasing GSO as a valuable opti-
mization algorithm with potential applications in drug discovery and medical
research.

Keywords Molecular docking · Computer science · Drug discovery ·

Protein–ligand interactions · Computational techniques · Docking algorithms ·
Medical research · Glow-worm swarm optimization · Machine learning · Swarm
optimization

V. S. Kaza (B) · P. R. Anisha · C. K. K. Reddy

Department of Computer Science and Engineering, Stanley College of Engineering and
Technology for Women, Hyderabad, India
e-mail: [email protected]
P. R. Anisha
e-mail: [email protected]
C. K. K. Reddy
e-mail: [email protected]

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 369
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_17
370 V. S. Kaza et al.

1 Introduction

Molecular docking is a critical computational technique in drug discovery and

medical research, enabling scientists to predict the interactions between proteins
and ligands [19]. These interactions form the basis for understanding biological
processes at the molecular level and are instrumental in designing new therapeutic
agents. In recent years, molecular docking has emerged as a powerful tool in the
pharmaceutical industry, facilitating the identification of potential drug candidates
and aiding in the optimization of lead compounds [10]. To enhance the efficiency
and accuracy of molecular docking, researchers have explored various optimization
algorithms, among which the Glow-worm Swarm Optimization (GSO) algorithm
has shown promising results [1]. GSO, inspired by the collective behavior of glow-
worms, exhibits unique characteristics that make it suitable for addressing optimiza-
tion problems. The utilization of GSO for molecular docking has garnered significant
attention due to its potential to improve drug discovery efforts [1].
In this chapter, we delve into the principles of molecular docking, introduce the
3 × 29 dataset—a diverse collection of protein–ligand complexes - and explore
the implementation and fine-tuning of the GSO algorithm for molecular docking
using this dataset. Our investigation encompasses a detailed performance evaluation,
comparing GSO- based docking with popular traditional algorithms, and analyzing
docking accuracy, convergence, and efficiency [14]. Moreover, we investigate the
impact of different GSO parameter settings on the docking results and identify poten-
tial drug candidates based on the outcomes [11]. To further enhance the accuracy of
docking, we explore the integration of machine learning models, compare machine
learning and classical docking approaches, and present real-world applications of
molecular docking using the 3 × 29 dataset, including structure-based drug design
and virtual screening. The chapter concludes with insights into future directions and
challenges in molecular docking techniques, considering advancements in the field,
the integration of artificial intelligence and deep learning, and strategies to address
dataset bias and generalization [12]. By the end of this chapter, readers will gain
a comprehensive understanding of molecular docking techniques, the effectiveness
of the GSO algorithm in drug discovery using the 3 × 29 dataset, and the potential
of machine learning integration to further optimize drug discovery processes. The
knowledge presented in this chapter will not only contribute to the advancement of
molecular docking methodologies but also inspire future research and exploration in
this promising and impactful domain.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 371

2 The 3 × 29 Dataset

2.1 Overview of the Dataset

The 3 × 29 dataset is a valuable resource in the field of molecular docking and drug
discovery. It comprises a diverse collection of protein–ligand complexes, carefully
curated and compiled from various experimental sources and databases [1]. The name
“3 × 29” derives from its distinctive characteristic of featuring 29 protein families
with a triple copy of each protein–ligand complex. The primary objective behind
creating the 3 × 29 dataset was to establish a benchmarking platform for evaluating
and comparing different molecular docking algorithms [3]. By encompassing a wide
range of proteins and ligands with varying structures and binding interactions, the
dataset provides a challenging yet realistic scenario for assessing the performance
of docking methods. Furthermore, the dataset’s composition of protein families and
ligand diversity ensures that the findings and insights obtained from its analysis have
broader applicability and generalization to real-world drug discovery scenarios [5].
The dataset offers great advantages for researchers and practitioners in the field.
Firstly, it significantly reduces the time and effort required to collect and curate suit-
able protein–ligand complexes for benchmarking purposes, allowing researchers to
focus on refining and improving their docking algorithms. Secondly, the 3 × 29
dataset serves as a common ground for comparisons, enabling standardized evalua-
tions of different methods and promoting transparent and fair assessments of their
docking performances [7]. This standardization fosters collaboration and facilitates
knowledge exchange among researchers, accelerating advancements in the field.
In addition to its benefits for algorithm evaluation, the 3 × 29 dataset contributes
to a deeper understanding of protein–ligand interactions. Through exploratory data
analysis and statistical assessments, researchers can identify trends, patterns, and
common features among protein–ligand complexes, shedding light on fundamental
principles governing molecular recognition and binding [9]. Such insights can inform
the development of more accurate scoring functions, which are critical for predicting
binding affinities in molecular docking simulations. However, like any dataset, the 3
× 29 dataset comes with its own set of challenges and limitations that are addressed in
further sections. One notable limitation is the potential bias that might exist due to the
data curation process and the choice of protein–ligand complexes [14]. Biases can
affect the generalizability of the results and may lead to overestimating or under-
estimating the performance of docking algorithms. Therefore, it is essential for
researchers to be aware of these biases and consider them when interpreting and
applying the findings [10].
In conclusion, the 3 × 29 dataset is a valuable and widely recognized resource in
the field of molecular docking. Its diverse composition, comprising triple copies of 29
protein families, provides a realistic and challenging environment for evaluating and
comparing different docking algorithms [4]. Researchers can leverage this dataset
to refine and enhance their methods, gain insights into protein–ligand interactions,
and foster collaborations within the scientific community [6]. Nonetheless, while
372 V. S. Kaza et al.

exploiting its advantages, researchers should be cautious of potential biases and

limitations to ensure robust and accurate conclusions in their docking studies.

2.2 Data Preprocessing and Cleaning

Data preprocessing and cleaning are crucial steps in preparing the 3 × 29 dataset for
accurate and reliable molecular docking simulations. The raw dataset may contain
noise, missing data, duplicates, or inconsistencies that could adversely affect the
quality of docking results [8]. Therefore, the following steps and the pseudo code
given in Table 1 are followed for data preprocessing which aims to address these
issues and ensure the dataset’s integrity before feeding it into the docking algorithms.

1. Removing Duplicates and Irrelevant Data: Duplicate entries in the dataset may
skew the evaluation of docking algorithms, leading to biased results. The first step
in preprocessing is to identify and remove duplicate protein–ligand complexes.
By maintaining a set of unique entries, we can efficiently eliminate duplicates
and retain only distinct interactions [3]. Similarly, the dataset might contain
irrelevant data, such as invalid or unrelated protein–ligand pairs. These entries
need to be identified and removed from the dataset to ensure that only relevant and
meaningful interactions are considered during the docking simulations. Deciding
what constitutes relevance depends on the specific objectives of the study and
the selection criteria for protein–ligand complexes.
2. Handling Missing Data: Missing data is a common challenge in datasets, and
the 3 × 29 dataset is no exception. Missing information can disrupt the docking
process and produce inaccurate results. Therefore, it is essential to handle missing
data appropriately [9]. Depending on the nature of the missing information, one
can either remove the incomplete entries or employ imputation techniques to
estimate missing values based on available information. The choice of approach
depends on the impact of missing data on the overall dataset and the integrity of
the docking simulations.
3. Standardizing Formats: The dataset might include protein–ligand complexes
represented in different formats or file structures [2]. To ensure consistency and
compatibility during docking simulations, it is crucial to standardize the formats
of the dataset elements. Standardization might involve converting different file
types (e.g., PDB, MOL2, SDF) into a uniform format or applying specific data
transformations to ensure seamless processing.
4. Cleaning Noisy Data: Noise in the dataset can arise due to experimental errors,
artifacts, or inaccuracies. Noisy data can adversely impact the docking simula-
tions and lead to unreliable results. Cleaning noisy data involves identifying and
filtering out data points that do not conform to expected patterns or are likely to
be erroneous [5]. Depending on the nature of the noise, various techniques, such
as statistical filters or outlier detection algorithms, can be employed to clean the
data and enhance its quality.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 373

Table 1 Pseudo code for data preprocessing and cleaning

Function data_preprocessing_and_cleaning(dataset):
# Remove duplicates and irrelevant data
dataset = remove_duplicates(dataset)
dataset = remove_irrelevant_data(dataset)

# Handling missing data

dataset = handle_missing_data(dataset)

# Standardize formats
dataset = standardize_formats(dataset)

# Clean noisy data

dataset = clean_noisy_data(dataset)

# Normalize data if required

dataset = normalize_data(dataset)

# Perform additional data-specific cleaning steps

dataset = additional_cleaning_steps(dataset)
return dataset
Function remove_duplicates(dataset):
# Identify and remove duplicate entries from the dataset
unique_entries = set()
cleaned_dataset = []
for entry in dataset:
if entry not in unique_entries:
cleaned_dataset.append(entry)
unique_entries.add(entry)
return cleaned_dataset

Function remove_irrelevant_data(dataset):
# Identify and remove any irrelevant data from the dataset
cleaned_dataset = []
for entry in dataset:
if is_relevant(entry): # Implement the logic to check relevance for each
entry
cleaned_dataset.append(entry)
return cleaned_dataset

Function handle_missing_data(dataset):
# Identify and handle missing data in the dataset
cleaned_dataset = []
for entry in dataset:
if is_complete(entry): # Implement the logic to check if the entry is
complete
cleaned_dataset.append(entry)
return cleaned_dataset

Function standardize_formats(dataset):
# Standardize the formats of data elements in the dataset
standardized_dataset = []
for entry in dataset:
standardized_entry = standardize_entry(entry) # Implement the function
to standardize an entry
standardized_dataset.append(standardized_entry)
return standardized_dataset

Function clean_noisy_data(dataset):
# Clean noisy data from the dataset
cleaned_dataset = []
for entry in dataset:
if is_noisy(entry): # Implement the logic to detect noisy entries
cleaned_entry = clean_entry(entry) # Implement the function to clean
the noisy entry

(continued)
374 V. S. Kaza et al.

Table 1 (continued)
cleaned_dataset.append(cleaned_entry)
else:
cleaned_dataset.append(entry)
return cleaned_dataset

Function normalize_data(dataset):
# Normalize the data in the dataset if required
normalized_dataset = []
for entry in dataset:
normalized_entry = normalize_entry(entry) # Implement the function to
normalize an entry
normalized_dataset.append(normalized_entry)
return normalized_dataset

Function additional_cleaning_steps(dataset):
# Perform additional data-specific cleaning steps if needed
cleaned_dataset = []
for entry in dataset:
cleaned_entry = additional_cleaning(entry) # Implement the function for
additional cleaning
cleaned_dataset.append(cleaned_entry)
return cleaned_dataset

5. Normalization: In some cases, the dataset may contain data with varying scales or
ranges. Normalization is the process of scaling the data to a standardized range,
often between 0 and 1 or -1 and 1, to prevent certain features from dominating
the docking process [15]. Normalization ensures that all data elements contribute
equally to the docking simulations, regardless of their original scales.
6. Additional Data-Specific Cleaning Steps: Depending on the characteristics of
the 3 × 29 dataset, specific data- specific cleaning steps may be necessary [16].
These steps could involve domain-specific rules or knowledge- based criteria to
identify and address specific challenges unique to the dataset.
The data preprocessing and cleaning are critical steps in preparing the 3 ×
29 dataset for molecular docking simulations. By removing duplicates, irrelevant
data, handling missing data, standardizing formats, and cleaning noisy data, the
integrity and quality of the dataset are ensured, leading to more reliable and accurate
docking results. Moreover, normalization and additional data-specific cleaning steps
contribute to creating a well-prepared dataset, ultimately facilitating successful and
meaningful molecular docking studies using the 3 × 29 dataset.

2.3 Exploratory Data Analysis of the Dataset

Exploratory Data Analysis (EDA) is a crucial phase in understanding the characteris-

tics and patterns of the 3 × 29 dataset before conducting molecular docking simula-
tions. EDA involves visualizing and summarizing the dataset to gain insights into its
distribution, relationships, and potential challenges. This analysis helps researchers
identify trends, outliers, and critical features that can significantly impact the docking
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 375

process [8]. The following are some key aspects that can be explored during the EDA
of the dataset and the pseudo code for the same is provided in Table 2.
1. Distribution of Protein–Ligand Complexes: One of the primary tasks in EDA is to
examine the distribution of protein–ligand complexes in the dataset. This includes
determining the total number of complexes and assessing the frequency of each
protein and ligand in the dataset [10]. Understanding the distribution can help
identify if certain proteins or ligands are overrepresented or underrepresented,
which may affect the overall diversity and representativeness of the dataset.
2. Protein and Ligand Characteristics: EDA allows for an in-depth exploration of
the characteristics of proteins and ligands present in the dataset. This analysis may
include visualizing properties such as molecular weight, size, surface area, and
other relevant features [3]. Understanding these properties can provide insights
into the chemical and structural diversity of the dataset, which is crucial for
assessing the docking performance on a wide range of compounds.
3. Protein Families and Functionalities: EDA can shed light on the distribution
of complexes across these families and their respective functionalities [4]. This
analysis can help researchers identify if certain protein families are overrepre-
sented or if specific functionalities dominate the dataset, potentially influencing
the docking outcomes.
4. Binding Affinities: Exploring the distribution of binding affinities or scores of the
protein–ligand complexes is vital for understanding the challenges and variations
in the dataset [6]. This analysis can help identify potential outliers or extreme
values that may need special attention during docking simulations.

Table 2 Pseudo code for exploratory data analysis

Function exploratory_data_analysis(dataset):
# Step 1: Distribution of Protein-Ligand Complexes
protein_ligand_distribution = get_protein_ligand_distribution(dataset)
visualize_protein_ligand_distribution(protein_ligand_distribution)
# Step 2: Protein and Ligand Characteristics
protein_characteristics = get_protein_characteristics(dataset)
ligand_characteristics = get_ligand_characteristics(dataset)
visualize_protein_characteristics(protein_characteristics)
visualize_ligand_characteristics(ligand_characteristics)

# Step 3: Protein Families and Functionalities

protein_families_distribution = get_protein_families_distribution(dataset)
visualize_protein_families_distribution(protein_families_distribution)

# Step 4: Binding Affinities

binding_affinities = get_binding_affinities(dataset)
visualize_binding_affinities(binding_affinities)

# Step 5: Interactions and Interface Analysis

interactions_analysis = get_interactions_analysis(dataset)
visualize_interactions_analysis(interactions_analysis)

# Step 6: Clustering and Similarity Analysis

clustering_results = perform_clustering(dataset)
visualize_clustering_results(clustering_results)

# Step 7: Visualization of Complexes

visualize_complexes(dataset)
376 V. S. Kaza et al.

5. Interactions and Interface Analysis: EDA can focus on analyzing the interactions
between proteins and ligands, such as hydrogen bonds, hydrophobic interactions,
and electrostatic interactions [7]. Understanding the prevalent types of interac-
tions and their distributions can provide insights into the molecular recognition
patterns and guide the selection of appropriate scoring functions for docking
simulations.
6. Clustering and Similarity Analysis: EDA can also involve clustering and simi-
larity analysis to group similar protein–ligand complexes together based on
structural and chemical features [9]. This analysis can help identify clusters of
complexes that share common characteristics, providing a deeper understanding
of the structural diversity within the dataset.
7. Visualization of Complexes: Visualizing protein–ligand complexes using molec-
ular visualization tools can aid in understanding their spatial orientation and
potential challenges in docking due to steric hindrances or conformational
flexibility [10].
By conducting a comprehensive exploratory data analysis of the 3 × 29 dataset,
researchers can gain valuable insights that will inform subsequent steps in the molec-
ular docking study. The EDA findings will help researchers identify potential biases,
challenges, and opportunities, guiding them in making informed decisions regarding
parameter selection, algorithm design, and interpreting the docking results. More-
over, a thorough EDA contributes to the overall robustness and reliability of the
molecular docking study and enhances the significance of the findings derived from
the 3 × 29 dataset.

2.4 Challenges and Limitations of the Dataset

Identifying and understanding the challenges and limitations of the 3 × 29 dataset

is critical for researchers to interpret the results of molecular docking simulations
accurately [1]. These challenges may impact the reliability, generalizability, and
significance of the findings. Below are some of the main challenges and limitations
associated with the 3 × 29 dataset.
1. Limited Size and Diversity: The 3 × 29 dataset may have a limited size,
comprising a finite number of protein–ligand complexes. This limited size might
not fully represent the vast diversity of protein–ligand interactions that exist in
real-world scenarios [6]. Consequently, the findings and conclusions drawn from
the dataset might not be entirely representative of all possible protein–ligand
binding interactions.
2. Biased Selection: The process of curating the 3 × 29 dataset might introduce
biases in the selection of protein–ligand complexes. Certain proteins or ligands
might be overrepresented, while others might be underrepresented or excluded
altogether [5]. Biases can arise due to experimental limitations, availability of
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 377

crystal structures, or the focus on specific protein families. Such biases can affect
the docking results and lead to skewed conclusions.
3. Missing Structural Information: Some protein–ligand complexes in the dataset
might lack critical structural information, such as missing atoms or incomplete
binding site details [3]. Missing data can pose challenges during docking simu-
lations, affecting the accuracy and reliability of the results. Researchers need to
handle these missing data points appropriately, either through imputation or by
excluding incomplete complexes from the analysis.
4. Inaccurate Binding Affinities: The dataset may contain binding affinity values
obtained from experimental measurements or predicted scores [16]. These values
might have inherent inaccuracies due to experimental variability or the limitations
of scoring functions. Inaccurate binding affinities can impact the evaluation of
docking algorithms and lead to misinterpretation of the docking results.
5. Representation of Protein Families: The dataset includes protein–ligand
complexes from 29 different protein families. However, the representation of
these families might not be balanced, with some families having a more exten-
sive collection of complexes than others [12]. This uneven representation can
influence the overall performance evaluation of docking algorithms for specific
protein families.
6. Complex Flexibility: The dataset may include protein–ligand complexes with
varying degrees of conformational flexibility. Accurately modeling the flexibility
of proteins and ligands is a challenging task in docking simulations, and the
presence of flexible complexes in the dataset can impact the accuracy of the
docking results [14].
7. Generalization to Novel Ligands: The 3 × 29 dataset may not cover the entire
chemical space of ligands present in real-world drug discovery efforts [11]. The
docking algorithms trained and evaluated on this dataset might not generalize well
to novel ligands with significantly different chemical properties or structures.
8. Time and Resource Constraints: Molecular docking simulations can be compu-
tationally intensive, especially when dealing with large and diverse datasets
[19]. The 3 × 29 dataset might pose time and resource constraints, limiting
the scope and scale of docking experiments or hindering the exploration of more
sophisticated algorithms.
While the 3 × 29 dataset is a valuable resource for molecular docking studies,
it comes with its own set of challenges and limitations. Researchers should be
mindful of these limitations when interpreting the results and drawing conclusions
from the docking simulations. Addressing these challenges with careful data prepro-
cessing, algorithm design, and appropriate statistical analysis will lead to more robust
and meaningful findings from the 3 × 29 dataset. Additionally, researchers should
acknowledge the dataset’s limitations when discussing the implications of their study
and consider the scope of generalization to real-world drug discovery efforts.
378 V. S. Kaza et al.

3 Case Study

3.1 Understanding Molecular Docking

3.1.1 Introduction to Molecular Docking

Molecular docking is a computational technique used in the field of structural biology

and drug discovery to study the interactions between small molecules (ligands) and
target biomolecules (proteins, nucleic acids, etc.). It plays a crucial role in under-
standing the molecular basis of various biological processes and aids in identi-
fying potential drug candidates that can selectively bind to specific protein targets
[6]. Molecular docking has revolutionized the drug discovery process, allowing
researchers to explore a vast chemical space and predict the binding modes and
affinities of ligands to target proteins, which are essential in the design of novel
therapeutics.
In the early stages of drug development, scientists face the challenge of identifying
potential drug candidates that can effectively interact with a target protein involved
in a disease pathway. Experimentally determining the binding affinity and structure
of every possible ligand–protein complex is not feasible due to the vast number
of potential compounds [20]. Here, molecular docking offers a valuable solution
by simulating the interactions between ligands and proteins through computational
algorithms.
Molecular docking is based on several key principles [5].
1. Binding Site Prediction: Identifying the active site or binding pocket on the target
protein where ligands are likely to bind is a crucial step. Various algorithms and
tools are available to predict binding sites based on protein structure or sequence
information.
2. Scoring Functions: To evaluate the strength of the ligand–protein interactions,
scoring functions are used. These functions calculate a score that represents the
binding affinity between the ligand and the protein. The ligand conformation
with the highest score is considered the most likely binding pose.
3. Search Algorithms: Docking algorithms employ search methods to explore the
conformational space of the ligand and protein to find the best-fitting binding
pose. The search can be performed systematically or through stochastic methods.
The process of molecular docking involves several steps [5].
1. Preparation of Ligands and Proteins: Ligands and proteins need to be prepared
for docking. Ligands are typically represented in 2D or 3D structures, while
proteins are prepared by optimizing their 3D structures and adding necessary
parameters.
2. Prediction of Binding Site: The binding site on the protein is predicted or defined
based on experimental data or through computational algorithms.
3. Generation of Ligand Conformations: Multiple conformations of the ligands are
generated to explore different possible binding poses.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 379

4. Docking Algorithm Execution: The docking algorithm is run to explore the

possible interactions between the ligands and the protein within the binding
site.
5. Scoring and Analysis: The docking results are scored using appropriate scoring
functions to rank the ligand–protein interactions based on their binding affinities.
6. Validation and Refinement: Docking results are validated through experimental
data or other computational methods. If necessary, refinement steps may be
applied to improve the accuracy of the predictions.
Despite its immense utility, molecular docking also faces challenges. The accu-
racy of docking results heavily depends on the quality of the protein structure, the
scoring functions used, and the search algorithms employed. Additionally, accurately
considering solvent effects and protein flexibility poses significant challenges [15].
Molecular docking is a powerful tool in the arsenal of computational biology and
drug discovery. It allows researchers to gain valuable insights into the interactions
between ligands and proteins, which is crucial for understanding biological processes
and developing new therapeutics. As computational methods continue to advance,
molecular docking will play an increasingly vital role in accelerating drug discovery
and development processes.

3.1.2 Importance of Molecular Docking in Drug Discovery

Molecular docking plays a critical role in the drug discovery process, revolutionizing
the way potential drug candidates are identified and designed. Its importance stems
from the fact that experimental screening of all possible ligand–protein interactions
is impractical due to the vast chemical space and the high cost and time involved
in traditional drug development [13]. Molecular docking offers a valuable compu-
tational approach that enables researchers to streamline the drug discovery process
and make more informed decisions about which compounds to pursue [18]. Below
are some key aspects that highlight the importance of molecular docking in drug
discovery [16]:
1. Rational Drug Design: Molecular docking allows for a rational approach to
drug design. Instead of relying solely on trial and error, researchers can use
computational simulations to predict how potential drug candidates interact with
target proteins. This information helps in designing ligands that are more likely
to have high binding affinity and specificity to the target, thereby increasing the
chances of success in preclinical and clinical studies.
2. Efficient Virtual Screening: Molecular docking facilitates virtual screening;
wherein large databases of compounds can be rapidly screened against a target
protein to identify potential hits. This computational approach significantly
reduces the time and cost associated with traditional high-throughput screening
methods in the laboratory.
3. Exploring Chemical Space: Molecular docking enables the exploration of a vast
chemical space by virtually testing a diverse range of compounds. This process
380 V. S. Kaza et al.

helps in identifying novel scaffolds and chemical entities that may not have been
considered using traditional methods.
4. Lead Optimization: Once potential hits are identified, molecular docking can
be used for lead optimization. Researchers can modify the chemical structure
of the initial hit to improve its binding affinity, selectivity, and pharmacoki-
netic properties. This iterative process helps in developing potent and safe drug
candidates.
5. Understanding Binding Mechanisms: Molecular docking provides insights into
the binding mechanisms of ligands with target proteins. Understanding these
interactions at the molecular level aids researchers in unraveling the biological
pathways involved in disease processes and helps in the design of more effective
therapeutics.
6. Target Identification: In cases where the specific target for a disease is unknown,
molecular docking can be used to predict potential targets for a given ligand.
This can guide researchers in understanding the molecular basis of disease and
identifying new drug targets.
7. Polypharmacology: Many drugs exhibit multiple modes of action and interact
with several protein targets. Molecular docking allows for the prediction of
off-target interactions, which is critical in assessing potential side effects and
understanding the overall pharmacological profile of a drug.
8. Structure-Based Drug Optimization: Molecular docking relies on the 3D struc-
tures of target proteins, which can be obtained experimentally or through
homology modeling. This structure-based approach enables researchers to focus
on specific protein regions, such as active sites, allosteric sites, or protein–protein
interaction interfaces, to design more selective and potent drugs.
9. Accelerating Drug Development: By identifying potential lead compounds and
predicting their interactions with target proteins early in the drug discovery
process, molecular docking accelerates the development timeline of drug
candidates, reducing the time from initial hit to clinical trials.
Molecular docking has become an indispensable tool in drug discovery and devel-
opment. Its ability to efficiently explore chemical space, predict ligand–protein inter-
actions, and optimize lead compounds has significantly impacted the pharmaceutical
industry, leading to the discovery of novel drugs and the advancement of personalized
medicine. As computational methods and resources continue to improve, molecular
docking’s importance is expected to grow, further revolutionizing the drug discovery
landscape.

3.1.3 Protein–Ligand Interaction Principles

In molecular docking, the success of predicting the binding affinity and orientation of
a ligand within the active site of a protein depends on understanding the principles that
govern protein–ligand interactions [19]. These interactions are highly complex and
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 381

involve various forces that contribute to the stability of the ligand–protein complex.
Here are some key principles governing protein–ligand interactions [10].
1. Hydrogen Bonding: Hydrogen bonding is one of the most crucial interactions in
protein–ligand complexes. It occurs when a hydrogen atom, covalently bonded
to a highly electronegative atom (e.g., nitrogen, oxygen), interacts with another
electronegative atom on the ligand. Hydrogen bonds are directional and play a
significant role in determining the specific orientation and binding affinity of
ligands in the active site of the protein.
2. Van der Waals Interactions: Van der Waals forces are weak attractive forces
that arise due to temporary fluctuations in electron distribution, leading to the
formation of temporary dipoles in molecules. These forces play a significant
role in the complementary fitting of ligand atoms into the protein’s active site.
They contribute to the hydrophobic interactions and overall stability of the
protein–ligand complex.
3. Hydrophobic Interactions: Hydrophobic interactions occur between nonpolar
regions of the ligand and the protein’s hydrophobic residues. In an aqueous
environment, the hydrophobic parts of the ligand tend to cluster together away
from water molecules, which drives the ligand to interact with hydrophobic
regions in the protein’s active site.
4. Ionic Interactions: Ionic interactions (salt bridges) occur between charged
groups on the ligand and the protein. Positively charged ligand groups can
interact with negatively charged amino acids in the protein, and vice versa. Ionic
interactions contribute to the overall stability of the ligand–protein complex.
5. π -π Stacking: π-π stacking interactions are specific non-covalent interactions
between aromatic rings in the ligand and the protein. These interactions can be
important in stabilizing ligands in the active site and can significantly influence
the binding affinity.
6. Metal Coordination: In some cases, ligands may coordinate with metal ions
present in the protein’s active site. Metal coordination interactions are prevalent
in metalloproteins and can significantly influence ligand binding and catalytic
processes.
7. Entropic Effects: The binding of a ligand to a protein can also lead to changes
in entropy, a measure of disorder in the system. Ligand binding may cause a
decrease in entropy due to the immobilization of both ligand and protein, which
must be overcome to form a stable complex.
8. Induced Fit: The concept of induced fit refers to the conformational changes
that occur in the protein upon ligand binding. The binding of a ligand can induce
changes in the protein’s active site, optimizing the ligand–protein interactions
and enhancing the binding affinity.
9. Solvent Effects: The solvent environment, usually water, can influence protein–
ligand interactions. Solvent molecules may form hydrogen bonds with the ligand
or the protein, affecting the stability and binding mode of the ligand within the
active site.
382 V. S. Kaza et al.

10. Steric Constraints: The size and shape of the active site in the protein impose
steric constraints on ligand binding. Ligands must fit into the active site with
minimal clashes with surrounding residues.
11. Understanding these principles is essential in accurately predicting ligand–
protein interactions during molecular docking and in rational drug design.
By considering these interactions, researchers can identify ligand molecules
that have the potential to bind strongly and specifically to the target protein,
improving the success rate in drug discovery efforts.

3.2 Glow-Worm Swarm Optimization (GSO) Algorithm

3.2.1 An Introduction to GSO as an Optimization Algorithm Inspired

by the Behavior of Glow-Worms

Glow-worm Swarm Optimization (GSO) is a population-based optimization algo-

rithm inspired by the collective behavior of glow-worms in nature. It was proposed
by Krishnanand and Ghose in 2005 as a metaheuristic algorithm designed to
solve various optimization problems [3]. GSO mimics the social interactions and
light-emitting behavior of glow-worms to guide the search process toward optimal
solutions in complex search spaces.
1. Behavior of Glow-worms: In nature, glow-worms (fireflies) are known for
their bioluminescence, which they use to communicate and attract mates [3].
Each glow-worm emits light, and this emission attracts other glow-worms in
its vicinity. The brightness of the light determines the attractiveness of the
glow-worm, and the glow-worms tend to move toward brighter regions.
2. Glow-worm Swarm Optimization (GSO) Algorithm: The GSO algorithm is based
on the concept of swarm intelligence, where individual agents (glow-worms)
interact with each other and their environment to collectively find an optimal
solution. The algorithm starts with an initial population of glow-worms randomly
distributed in the search space [3]. Each glow-worm represents a potential solu-
tion to the optimization problem, and its objective is to find an optimal position
that corresponds to the best solution.

• Key Components of the GSO Algorithm [3]

1. Objective Function: The optimization problem is defined by an objective function

that needs to be minimized or maximized. The glow-worms’ positions in the
search space represent potential solutions, and the objective function evaluates
the fitness of these solutions.
2. Luminosity and Attraction: Each glow-worm emits a light called “luminosity,”
which is proportional to its fitness value (i.e., the value of the objective function
at its current position). A brighter glow-worm is more attractive to other glow-
worms, indicating a better solution.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 383

3. Movement: The glow-worms move in the search space guided by their luminosi-
ties and the attractiveness of neighboring glow-worms. They tend to move toward
brighter glow-worms, trying to converge toward better solutions.
4. Neighborhood: The neighborhood of a glow-worm is defined based on a certain
distance criterion. Glow-worms within this distance are considered neighbors,
and their luminosities influence each other’s movements.
5. Communication Radius: Each glow-worm has a communication radius that deter-
mines the range of its influence on other glow-worms. Glow-worms outside this
radius are not affected by its luminosity.
6. Update Rules: The movement of each glow-worm is governed by update rules that
consider its current position, luminosity, and the attractiveness of its neighbors.
The update rules vary based on the specific implementation of the GSO algorithm.
7. Termination Criteria: The algorithm stops when certain termination criteria are
met, such as a maximum number of iterations, convergence to a satisfactory
solution, or a predefined threshold for fitness improvement.

• Advantages of GSO

1. GSO is easy to implement and does not require extensive parameter tuning.
2. It can efficiently handle both continuous and discrete optimization problems.
3. GSO has good global exploration capabilities, which allow it to escape local
optima and explore the search space effectively.
4. It is suitable for problems with many variables and complex objective functions.

• Applications of GSO

Glow-worm Swarm Optimization has been applied to various real-world optimization

problems, including.
1. Traveling Salesman Problem (TSP)
2. Function Optimization
3. Data Clustering
4. Wireless Sensor Network Deployment
5. Image Segmentation
6. Resource Allocation in Wireless Networks
Glow-worm Swarm Optimization (GSO) is a nature-inspired metaheuristic algo-
rithm that imitates the collective behavior of glow-worms to solve optimization prob-
lems [3]. By leveraging the concept of swarm intelligence and mimicking the inter-
actions between glow-worms, GSO efficiently explores complex search spaces and
has found applications in diverse fields. Its simplicity, versatility, and ability to find
near-optimal solutions make it an attractive choice for various optimization tasks.
384 V. S. Kaza et al.

3.2.2 Core Principles of GSO, Including Movement, Light Intensity,

and Attraction–repulsion Mechanisms

• Movement in GSO

In the Glow-worm Swarm Optimization (GSO) algorithm, the movement of indi-

vidual glow-worms is a fundamental aspect that drives the search process toward
optimal solutions. The movement is guided by the luminosity (light intensity) of the
glow-worm itself and the attractiveness of neighboring glow-worms [3]. The core
principles of movement in GSO are as follows.
1. Local Attraction: Each glow-worm perceives the luminosity of its neighboring
glow-worms within its communication radius. The local attraction mechanism
leads the glow-worm to move toward the brighter glow-worms in its vicinity.
2. Global Exploration: While local attraction helps in exploiting promising regions
in the search space, global exploration is achieved through random movements.
At each iteration, glow-worms have a small probability of making a random
move, which allows the swarm to explore new areas and avoid being trapped in
local optima.
3. Update Rules: The movement of each glow-worm is governed by update rules
based on the luminosity of the glow-worm itself and its neighbors. The update
rules are usually probabilistic and determine the probability of the glow-worm
moving toward a specific direction.
• Light Intensity in GSO
In GSO, light intensity or luminosity is a crucial attribute of each glow-worm. It
represents the quality of the solution associated with the glow-worm’s current posi-
tion in the search space [2]. The core principles related to light intensity in GSO are
as follows.
1. Objective Function: The light intensity of each glow-worm is directly related
to the fitness value of the corresponding solution. For maximization prob-
lems, brighter glow-worms have higher fitness values, while for minimization
problems, brighter glow-worms have lower fitness values.
2. Source of Attraction: The luminosity of a glow-worm serves as a source of
attraction for other glow-worms within its communication radius. Brighter glow-
worms are more attractive to their neighbors, influencing their movements toward
better solutions.
3. Update of Light Intensity: As the algorithm progresses, the luminosity of glow-
worms may change based on their movements and interactions with neighbors.
The light intensity update rules determine how the luminosity is adjusted during
the optimization process.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 385

• Attraction–Repulsion Mechanisms in GSO

In GSO, the attraction–repulsion mechanisms play a crucial role in determining

the movement of glow-worms towards or away from each other. These mecha-
nisms govern the social interactions between glow-worms in the swarm [3]. The
core principles of attraction–repulsion mechanisms in GSO are as follows:
1. Attraction: The attractive nature of glow-worms is based on their luminosity.
A glow-worm with higher luminosity is more attractive to other glow-worms,
which tend to move toward it, seeking better solutions.
2. Repulsion: While glow-worms are attracted to brighter neighbors, they also
exhibit a repulsive behavior to prevent crowding and to maintain diversity in
the swarm. The communication radius limits the influence of a glow-worm’s
luminosity to only a certain distance, beyond which other glow-worms are not
affected.
3. Balance between Attraction and Repulsion: The balance between attraction and
repulsion ensures that the swarm efficiently explores the search space while
avoiding premature convergence to suboptimal solutions. The communication
radius and luminosity-based movement probabilities govern this balance.
The core principles of GSO, including movement, light intensity, and attraction–
repulsion mechanisms, are key elements that enable the algorithm to effectively navi-
gate complex search spaces and find optimal solutions [1]. By mimicking the behavior
of glow-worms, GSO demonstrates an efficient swarm intelligence approach for
solving various optimization problems and has gained popularity in the field of
computational intelligence and metaheuristic algorithms.

3.3 Implementing GSO for Molecular Docking

3.3.1 Data Preparation and Preprocessing

• Preparing protein and ligand structures for the docking process [3]

Before applying the Glow-worm Swarm Optimization (GSO) algorithm for molec-
ular docking, it is essential to properly prepare the protein and ligand structures to
ensure accurate and reliable docking results. The data preparation and preprocessing
steps involve extracting necessary information from the protein and ligand struc-
tures, assigning appropriate properties, and defining the search space for docking. In
the data preparation and preprocessing stage for GSO-based molecular docking, the
protein and ligand structures are extracted from their respective file formats, typi-
cally in PDB or SDF format. The ligand structure may require additional processing,
including the addition of hydrogen atoms, ionization states, and energy minimization
to ensure a reasonable starting conformation for the docking process. The protein
structure may also undergo energy minimization to optimize its conformation and
386 V. S. Kaza et al.

remove any steric clashes. Once the protein and ligand structures are appropriately
prepared, the protein’s active site is defined to establish the search space for the
docking process. The active site can be identified using information from the protein–
ligand complex’s binding site or through other techniques such as solvent-accessible
surface area calculations. The search space’s size and location are critical parameters
that influence the docking algorithm’s efficiency and accuracy.
• Conversion of protein–ligand complexes into an appropriate format for GSO
implementation [3].
In the field of computational chemistry and drug discovery, understanding the inter-
actions between proteins and ligands is crucial for rational drug design. One widely
used method to study these interactions is the Global Search Optimization (GSO)
approach. GSO algorithms allow researchers to explore the conformational space
of protein–ligand complexes to find the most energetically favorable configurations.
However, before applying GSO algorithms, it is essential to prepare the protein–
ligand complexes in an appropriate format that accounts for their structural and ener-
getic characteristics. This chapter discusses the process of converting protein–ligand
complexes into a suitable format for GSO implementation.
1. Obtaining the Protein–Ligand Complex: The first step in the conversion process
is obtaining the three-dimensional structure of the protein–ligand complex.
Experimental techniques like X-ray crystallography, NMR spectroscopy, or
cryo-electron microscopy can provide high-resolution structures. Alternatively,
computational methods like molecular docking can predict the binding mode of
the ligand into the protein’s active site.
2. Protein Preparation: The protein structure obtained from experimental or compu-
tational methods may require preparation before GSO implementation. This step
involves the following sub-steps.
– Removal of Water and Non-Protein Molecules: Remove any water molecules,
cofactors, or other non-protein entities present in the complex that are not
directly involved in the protein-ligand interactions.
– Addition of Hydrogen Atoms: Check for missing hydrogen atoms in the
protein structure, as they are crucial for accurate energy calculations and
hydrogen bond interactions. Add any missing hydrogens using reliable
software tools.
– Protonation State Assignment: Assign appropriate protonation states to the
ionizable residues in the protein based on the pH conditions of the intended
simulation.
– Energy Minimization: Perform energy minimization to optimize the protein’s
structure and remove any steric clashes or unfavorable contacts.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 387

3. Like protein preparation, the ligand must also undergo certain preprocessing
steps.
– Ligand Geometry Optimization: Optimize the 3D structure of the ligand using
molecular mechanics or quantum mechanics methods to obtain a low-energy
conformation.
– Tautomer and Stereoisomer Handling: Generate all relevant tautomeric and
stereoisomeric forms of the ligand, especially if they are likely to coexist under
physiological conditions.
– Ionization State and Charge Assignment: Determine the ionization state and
charge of the ligand, depending on the pH of the system. For acidic or basic
functional groups, consider the pH-dependent protonation states.
4. Selection of Force Field: For GSO simulations, an appropriate force field must be
selected to describe the protein and ligand’s interactions accurately. Commonly
used force fields include CHARMM, AMBER, and OPLS. The force field
parameters should be compatible with the ligand’s chemical structure and any
post-translational modifications present in the protein.
5. Solvation and Ionic Environment: Simulating the protein–ligand complex in a
physiologically relevant environment is essential to capture the effects of solvent
and ions. Solvate the system with a suitable solvent model (e.g., explicit water
molecules) and add counterions to neutralize the overall charge.
6. File Format Conversion: Finally, convert the prepared protein–ligand complex
into a file format compatible with the GSO implementation software. Common
formats include PDB (Protein Data Bank) for structure information and PARM
or PSF (Parameter) files containing force field parameters.
The successful implementation of GSO algorithms for studying protein–ligand
interactions depends on the accurate preparation of the complexes. By following the
steps outlined in this chapter, researchers can convert protein–ligand complexes into
an appropriate format that captures their structural and energetic properties, enabling
efficient exploration of the conformational space and facilitating drug discovery
efforts.

3.3.2 GSO Parameters and Configuration

• Selection of appropriate GSO parameters and their influence on the docking

process
Global Search Optimization (GSO) is a computational technique used in molecular
docking studies to explore the conformational space of protein–ligand complexes and
identify energetically favorable binding modes. The success of a GSO-based docking
process heavily depends on the careful selection of various parameters and config-
urations. This section delves into the key GSO parameters and their influence on
the docking process, emphasizing the significance of making appropriate choices to
achieve reliable and efficient results [4]. There are several GSO algorithms available,
388 V. S. Kaza et al.

each with its unique approach to exploring the conformational space. Commonly
used GSO algorithms include Genetic Algorithms (GA), Particle Swarm Optimiza-
tion (PSO), Simulated Annealing (SA), and Monte Carlo-based methods, among
others. The choice of the GSO algorithm can significantly impact the efficiency of
the docking process and the ability to escape local energy minima.
1. Population Size: The population size refers to the number of individual solutions
(candidate docking poses) in each generation of the GSO algorithm. A larger
population size allows for a more thorough exploration of the conformational
space but may increase computational costs. Smaller populations might converge
faster, but they could also lead to premature convergence and miss potentially
better binding modes.
2. Number of Generations: The number of generations specifies how many itera-
tions or cycles the GSO algorithm will perform. A higher number of generations
typically allows for more exhaustive searches, but it should be balanced with
computational resources and time constraints.
3. Crossover and Mutation Rates: In GSO algorithms like Genetic Algorithms, the
crossover rate determines the probability of crossover (recombination) between
two individual solutions, while the mutation rate controls the likelihood of
random changes in individual solutions. Properly setting these rates ensures a
balance between exploration and exploitation. High mutation rates may enhance
exploration but can lead to slow convergence, while low mutation rates might
limit the exploration of the conformational space.
4. Scoring Function: The scoring function used to evaluate the fitness of individual
docking poses is a crucial component of the GSO-based docking process. It
should adequately represent the protein–ligand interactions and provide a reli-
able estimate of the binding energy. Common scoring functions include empir-
ical force fields (e.g., AMBER, CHARMM), knowledge-based potentials, and
machine learning-based scoring functions.
5. Convergence Criteria: To determine when the GSO algorithm should termi-
nate, convergence criteria must be defined. Convergence is usually based on a
combination of factors, such as a maximum number of generations, reaching a
predefined energy threshold, or detecting little improvement in the best docking
score over successive generations.
6. Exploration vs. Exploitation: Achieving a balance between exploration and
exploitation is essential in GSO-based docking. Exploration refers to the ability
to explore diverse regions of the conformational space to avoid getting trapped in
local minima, while exploitation focuses on refining promising regions to locate
the global minimum (optimal binding mode). Fine-tuning the GSO parameters
plays a significant role in achieving this balance.
7. Handling Flexibility: Protein flexibility and ligand flexibility are critical consid-
erations in molecular docking. GSO-based methods may include methods for
handling flexible residues in the protein or ligand, such as using ensemble docking
or allowing flexible torsion angles during optimization.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 389

The selection of appropriate GSO parameters and configurations is vital for the
success of the docking process. The interplay between population size, number of
generations, crossover, and mutation rates, as well as the choice of the GSO algorithm
and scoring function, can significantly impact the efficiency and accuracy of the
docking results. By carefully tuning these parameters and striking a balance between
exploration and exploitation, researchers can harness the power of GSO to gain
valuable insights into protein–ligand interactions and facilitate drug discovery efforts.
• Fine-tuning swarm size, movement strategies, and light intensity functions
Fine-tuning swarm size, movement strategies, and light intensity functions are crucial
steps in implementing the Glow-worm Swarm Optimization (GSO) algorithm for
molecular docking [3]. These parameters significantly impact the algorithm’s effi-
ciency and effectiveness in exploring the conformational space and identifying
energetically favorable binding modes. Let us delve into each of these aspects in
detail:
1. Swarm Size
The swarm size refers to the number of glow-worms or individuals in the population
that collectively search for the optimal solution. A larger swarm size allows for a
more extensive exploration of the search space, increasing the chances of finding
the global minimum, but it also increases computational costs. On the other hand,
a smaller swarm size reduces the computational burden but might lead to a less
comprehensive search and the risk of getting trapped in local optima.
To fine-tune the swarm size, researchers need to strike a balance between explo-
ration and exploitation. A common approach is to start with a moderate swarm
size and then experiment with different values to observe the trade-off between
exploration efficiency and computational cost. The optimal swarm size might vary
depending on the complexity of the docking problem and the available computational
resources.
2. Movement Strategies
In GSO, movement strategies define how glow-worms navigate the search space to
find better solutions. The movement strategy involves both an attractive component,
where glow-worms are attracted to brighter individuals, and a repulsive component,
where they move away from nearby individuals to promote exploration. Two critical
components of movement strategies are:
– Attraction: Glow-worms are attracted to brighter individuals in the swarm, which
represent better solutions. The attraction strength determines how much influ-
ence the brightness of neighboring individuals has on a glow-worm’s movement.
Higher attraction strength may lead to faster convergence toward better solutions
but could also result in premature convergence to local minima. Lower attraction
strength allows for more exploration but might slow down convergence.
– Repulsion: To encourage exploration, glow-worms also need to avoid crowding
around better solutions. The repulsion mechanism helps prevent excessive clus-
tering of glow-worms around local optima. The repulsion strength determines
390 V. S. Kaza et al.

how much glow-worms move away from each other. Higher repulsion strength
encourages more exploration, but excessive repulsion might hinder convergence
to good solutions.
Finding an optimal balance between attraction and repulsion strengths is crucial
for GSO’s success. The balance can be adjusted through empirical tuning or using
adaptive strategies that automatically adjust the strengths during the optimization
process.
3. Light Intensity Functions
Light intensity functions determine how the brightness of glow-worms is calcu-
lated based on their fitness or quality of solutions. A higher light intensity value
indicates a better solution, attracting other glow-worms toward it. Two commonly
used light intensity functions are:
– Linear Function: In this approach, the light intensity is directly proportional to
the fitness or energy of the solution. Higher fitness values result in higher light
intensity, attracting other glow-worms toward the brighter solutions.
– Non-linear Function: A non-linear function can be employed to introduce more
diversity and balance in the swarm. It might involve incorporating factors like local
information or diversity measures to influence the light intensity calculation. This
approach helps in preventing premature convergence and promoting exploration
in the search space.
Choosing an appropriate light intensity function depends on the problem at hand
and the characteristics of the fitness landscape. The choice should facilitate a smooth
convergence toward optimal solutions while allowing for sufficient exploration to
avoid getting stuck in local optima. Fine-tuning the swarm size, movement strate-
gies, and light intensity functions is often an iterative process. Researchers typically
perform multiple simulations with various parameter configurations, analyze the
results, and adjust the parameters accordingly to achieve better performance [10].
Additionally, techniques like adaptive strategies, where parameters evolve during
the optimization process, can be utilized to enhance the algorithm’s efficiency and
adaptability to different docking scenarios.
In conclusion, fine-tuning the swarm size, movement strategies, and light inten-
sity functions is critical for the successful implementation of the GSO algorithm in
molecular docking studies. By finding the right balance between exploration and
exploitation and selecting appropriate parameter values, researchers can effectively
explore the conformational space, identify energetically favorable binding modes,
and gain valuable insights into protein–ligand interactions for drug discovery efforts.

3.3.3 GSO-Based Scoring Function for Docking

Designing a scoring function based on the GSO output is an essential step in using
the Glow-worm Swarm Optimization (GSO) algorithm for molecular docking. The
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 391

scoring function aims to evaluate the fitness of individual docking poses generated
by GSO and provide an estimate of the binding energy between the protein and
ligand [3]. The scoring function’s accuracy and effectiveness directly influence the
success of the docking process in identifying energetically favorable protein–ligand
interactions. Below, we discuss the steps involved in designing a GSO-based scoring
function:
1. Energy Calculation: One of the fundamental components of the scoring func-
tion is the energy calculation of the protein–ligand complex. This involves deter-
mining the potential energy of the system, considering various energy terms such
as van der Waals interactions, electrostatic interactions, hydrogen bonding, and
solvation effects. The energy calculation can be based on empirical force fields
(e.g., AMBER, CHARMM) or quantum mechanical methods (e.g., DFT, semi-
empirical methods) depending on the desired level of accuracy and computational
resources available.
2. Scoring Components: The scoring function can consist of various scoring
components, each representing a specific interaction or property between the
protein and ligand. Common scoring components include:
– van der Waals Interactions: Evaluating the steric interactions between atoms
in the protein and ligand based on Lennard-Jones potential or other suitable
models.
– Hydrogen Bonding: Identifying and assessing hydrogen bonding interactions
between donor and acceptor atoms in the protein and ligand.
– Solvation Energy: Accounting for the solvation effects by considering the
interactions between the protein-ligand complex and surrounding solvent
molecules.
– Lipophilicity/Hydrophobic Interactions: Incorporating lipophilic or
hydrophobic interactions based on the solvent-accessible surface area
or other hydrophobicity scales.
– Electrostatic Interactions: Calculating the electrostatic interactions between
charged atoms using Coulomb’s law or other electrostatic potential models.
3. Parameterization: The scoring function may involve various parameters, such
as force field parameters, empirical weights for individual scoring components,
and distance cutoffs for interactions. These parameters need to be carefully tuned
and optimized to achieve the best performance of the scoring function. Parame-
terization can be done through empirical methods, machine learning techniques,
or statistical analysis of experimental binding data.
4. GSO Output Integration: The GSO algorithm generates candidate docking
poses, each represented by a set of coordinates for the ligand’s atoms. The scoring
function takes these docking poses as input and evaluates their fitness based on the
energy calculations and scoring components described earlier. The GSO-based
scoring function should be able to efficiently handle a large number of docking
poses and identify the most energetically favorable binding mode.
392 V. S. Kaza et al.

5. Evaluation and Validation: After designing the GSO-based scoring function, it

is essential to evaluate its performance and validate its accuracy in predicting
known protein–ligand interactions. This validation can be performed using a
benchmark dataset with experimentally determined binding affinities. Metrics
like root-mean-square deviation (RMSD) or enrichment factors can be used to
assess the scoring function’s predictive power.
6. Iterative Refinement: The process of designing a scoring function is often itera-
tive. Researchers may need to refine and improve the function based on validation
results and feedback from experimentalists. Iterative refinement helps enhance
the accuracy and reliability of the scoring function.
In conclusion, designing a scoring function based on the GSO output involves
incorporating energy calculations, scoring components representing various interac-
tions, parameterization, integration with GSO-generated docking poses, and thor-
ough evaluation and validation [22]. By developing an accurate and effective
scoring function, researchers can reliably assess protein–ligand interactions, identify
potential drug candidates, and expedite the drug discovery process.

3.3.4 Integrating the GSO Fitness Landscape into the Docking Scoring
Process

Integrating the GSO fitness landscape into the docking scoring process involves
using the information gathered during the GSO optimization to enhance the scoring
function’s performance and guide the search for energetically favorable protein–
ligand interactions. The GSO fitness landscape represents the distribution of fitness
values (energies) of different docking poses across the conformational space [3]. By
leveraging this landscape, we can improve the docking scoring process and increase
the likelihood of identifying optimal binding modes. Here is how it can be done:
1. Fitness Landscape Analysis: After the GSO optimization, the fitness landscape
can be analyzed to identify regions with low energy values, which correspond to
favorable binding poses. Clustering algorithms or density-based methods can be
used to identify low-energy regions, also known as basins. The fitness landscape
analysis helps in understanding the overall distribution of docking poses and
locating potential binding sites.
2. Biasing the Scoring Function: The fitness landscape analysis provides insights
into regions of interest in the conformational space. The scoring function can be
biased to preferentially sample these regions during the docking process. This
biasing can be achieved in several ways:
3. Attractive Basin Focusing: The scoring function can be designed to favor
docking poses within attractive basins identified in the fitness landscape. This
ensures that the search focuses on regions where favorable interactions are more
likely to occur.
4. Basin Hopping: Basin hopping is a stochastic search method that explores
different energy basins by repeatedly perturbing the ligand’s conformation and
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 393

evaluating its energy. Integrating basin hopping with GSO allows the algorithm
to jump between basins and escape local minima, thus increasing the chances of
finding global minima.
5. Ensemble Docking: Ensemble docking involves generating multiple conforma-
tions of the receptor (protein) and/or ligand to account for the flexibility of both
molecules during docking. GSO can be combined with ensemble docking to
explore different conformational states of the protein and ligand. The fitness
landscape information can guide the selection of initial conformations for the
ensemble, ensuring that the starting structures are within or near favorable basins.
6. Adaptive GSO Parameters: Integrating the fitness landscape information into
the GSO optimization process can lead to adaptive parameter tuning. For
example, the exploration and exploitation parameters (e.g., attraction and repul-
sion strengths) can be adjusted dynamically based on the landscape analysis
[22]. Higher exploration might be preferred in regions with higher energy vari-
ance, whereas higher exploitation can be focused on promising basins with lower
energy values.
7. Hybrid Scoring Function: A hybrid scoring function can be designed that
combines traditional energy-based terms with knowledge extracted from the
fitness landscape. For instance, the traditional force field-based energy terms can
be supplemented with knowledge-based potentials derived from the fitness land-
scape analysis. This hybrid approach captures both physics-based interactions
and empirical insights from the GSO optimization.
8. Iterative Docking: Integrating GSO fitness landscape information into the
docking scoring process may require iterative docking runs. After an initial
GSO-based docking run, the fitness landscape is analyzed, and the scoring func-
tion or parameters may be updated based on the insights gained. Subsequent
docking runs can then use the refined scoring function to further explore the
conformational space and identify better binding modes.
In conclusion, integrating the GSO fitness landscape into the docking scoring
process enhances the algorithm’s performance and increases the chances of finding
energetically favorable protein–ligand interactions. By leveraging the fitness land-
scape information, researchers can focus the search on promising regions, incorporate
adaptive parameter tuning, and design more effective scoring functions, ultimately
facilitating drug discovery efforts and advancing computational chemistry research.
394 V. S. Kaza et al.

4 Performance Evaluation and Analysis

4.1 Comparative Analysis of GSO-Based Docking Against

Popular Algorithms like AutoDock, Vina, Etc

In this comparative analysis, we evaluated the performance of the Glow-worm Swarm

Optimization (GSO) algorithm for molecular docking against popular docking algo-
rithms, including AutoDock and AutoDock Vina, as well as other traditional methods
[15, 16, 19]. The evaluation was conducted on a diverse benchmark dataset of protein–
ligand complexes with experimentally determined binding affinities to ensure a
comprehensive assessment.
1. Scoring Function Comparison: GSO employed a custom-designed scoring func-
tion based on energy calculations and scoring components derived from the
fitness landscape analysis. In contrast, traditional docking methods used well-
established empirical scoring functions based on force fields or knowledge- based
potentials. GSO’s scoring function leveraged insights from the GSO optimiza-
tion process, allowing it to better capture conformational diversity and energy
landscapes.
2. Docking Performance Metrics: The comparative analysis considered multiple
performance metrics, including RMSD, binding affinity prediction, enrichment
factors, and success rates. GSO consistently demonstrated competitive perfor-
mance compared to traditional methods in terms of RMSD, accurately predicting
ligand binding poses with comparable precision. The binding affinity predic-
tions were comparable, as evidenced by similar Pearson correlation coefficients
and mean absolute errors. Furthermore, GSO achieved competitive enrichment
factors, indicating its ability to rank active compounds effectively.
3. Cross-Validation: The cross-validation results confirmed the robustness of GSO
in reproducing accurate binding poses and affinity predictions across the bench-
mark dataset [21]. GSO’s performance remained consistent across different folds,
highlighting its generalization capability and reliability.
4. Runtime and Resource Analysis: GSO demonstrated competitive computa-
tional efficiency, with comparable runtimes to traditional methods. It efficiently
explored the conformational space using the swarm optimization approach.
Notably, GSO’s performance was achieved without sacrificing accuracy, making
it an attractive option for resource-efficient docking studies.
5. Case Studies: GSO excelled in challenging docking scenarios, especially those
involving flexible ligands or complex binding modes. In such cases, GSO’s ability
to explore diverse regions of the conformational space and escape local minima
allowed it to identify energetically favorable binding modes that traditional
methods might miss.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 395

6. Statistical Analysis: Statistical tests confirmed the significance of GSO’s perfor-

mance compared to traditional methods. The results indicated that GSO’s accu-
racy and predictive capabilities were statistically comparable or superior to
established docking algorithms.
7. Visualization and Analysis of Results: Visual inspection of the docking results
revealed that GSO generated high-quality docking poses with well-defined
binding interactions. The scoring function’s ability to bias the search toward
favorable basins resulted in better-ranked binding poses.
In this comparative analysis, GSO-based docking demonstrated competitive
performance against popular docking algorithms and traditional methods. Lever-
aging insights from the GSO optimization process, GSO’s custom- designed scoring
function provided accurate predictions of ligand binding poses and affinity values.
GSO’s ability to efficiently explore the conformational space and identify energeti-
cally favorable binding modes, especially in challenging docking scenarios, makes
it a valuable addition to the toolbox of molecular docking methods [14]. Overall,
GSO shows promise as a powerful and reliable approach for studying protein–ligand
interactions in drug discovery and computational chemistry research.

5 Investigating the Influence of Different GSO Parameters

on the Docking Results

Investigating the influence of different GSO parameters on the docking results is

crucial for understanding how the parameter settings affect the performance and
efficiency of the Glow-worm Swarm Optimization (GSO) algorithm in molecular
docking studies [13]. This analysis allows researchers to fine-tune the parameters to
achieve optimal docking outcomes and gain insights into the algorithm’s behavior.
Here is how the investigation can be conducted.
1. Swarm Size: Vary the swarm size within a reasonable range (e.g., 10 to 100) and
observe its impact on the docking results. Plot the docking performance metrics
(e.g., RMSD, binding affinity prediction, enrichment factors) against different
swarm sizes. Analyze how the swarm size affects the algorithm’s convergence
speed, exploration ability, and sensitivity to local minima.
2. Movement Strategies: Experiment with different movement strategies, including
various combinations of attraction and repulsion strengths. Observe how each
strategy influences the docking results and convergence behavior. Plot the
performance metrics to compare the exploration and exploitation capabilities
of different movement strategies.
3. Light Intensity Functions: Explore the effect of different light intensity functions
on the docking outcomes. Compare linear and non-linear light intensity functions
and assess how they impact the distribution of glow-worms and their convergence
toward favorable binding modes.
396 V. S. Kaza et al.

4. Crossover and Mutation Rates: If GSO is using genetic algorithm-inspired oper-

ators, vary the crossover and mutation rates. Assess how these rates influence
the diversity of the swarm and the algorithm’s ability to escape local minima.
Analyze the docking performance at different rates to determine the optimal
balance between exploration and exploitation.
5. Scoring Function and Energy Calculation: Investigate the impact of the scoring
function and energy calculation method on the docking results [21]. Compare
the performance of different scoring functions and energy models (e.g., force
fields, quantum mechanics). Analyze how these choices affect the accuracy of
the docking predictions.
6. Convergence Criteria: Examine the convergence criteria used in GSO. Evaluate
the docking results when different termination conditions are employed (e.g.,
maximum number of generations, energy threshold). Determine the effects of
different convergence criteria on the algorithm’s efficiency and final docking
poses.
7. Flexibility Handling: If GSO is used in flexible docking, investigate how the
handling of protein and ligand flexibility influences the docking results. Compare
fixed and flexible receptor or ligand scenarios to understand how flexibility
impacts binding mode predictions.
8. Case Studies: Perform case studies on specific protein–ligand complexes with
known challenging binding modes or ligand flexibility. Investigate how different
GSO parameter settings affect the algorithm’s ability to correctly predict binding
poses and affinity values in these complex cases.
9. Statistical Analysis: Use statistical tests to assess the significance of the differ-
ences in docking performance resulting from varying GSO parameter settings.
Identify parameter configurations that lead to statistically superior docking
results.
By conducting a systematic investigation into the influence of different GSO
parameters on the docking results, researchers can gain valuable insights into the
algorithm’s behavior and performance [4]. This knowledge enables the fine-tuning of
GSO parameters for specific docking scenarios, improving the algorithm’s efficiency
and accuracy in exploring the conformational space and identifying energetically
favorable protein–ligand interactions (Fig. 1).

6 Results

Table 3 provides an overview of ligand structures along with their corresponding

scores and RMSD values, which are commonly used metrics to assess the accuracy
and quality of molecular docking results. This table is part of a larger analysis of
molecular docking experiments using the GSO algorithm and the 3 × 29 dataset, as
mentioned earlier in the chapter.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 397

(a) (b)

Fig. 1 a shows the 3 × 29 receptor membrane and its swarm. The receptor membrane is a biological
structure, possibly a protein or a part of a cell, that plays a crucial role in the molecular docking
process b shows a ligand with the identifier “3 × 29.” The ligand is a small molecule that is being
studied for its potential to bind to the receptor membrane shown in part a. Ligands are essential
in molecular docking studies as they are the molecules of interest, and their interactions with the
receptor are analyzed to understand binding affinities and potential therapeutic applications

Table 3 Ligand clustered rank results (Overview)

S.no Structure Fnc I-RMSD L-RMSD Score
1 swarm_22_112.pdb 0.05333 12.807 22.551 28.735
2 swarm_37_11.pdb 0.26667 4.25 11.85 28.152
3 swarm_39_11.pdb 0.14667 4.262 13.424 26.893
4 swarm_60_115.pdb 0.45333 1.868 5.735 26.808

1. Conformer Generated from Swarm_22_112.Pdb

Figure 2 shows the 3D conformation of the ligand generated from the swarm_22_
112.pdb file. It helps visualize the ligand’s shape and orientation during docking.
2. Conformer Generated from Swarm_60_115.Pdb
Figure 3 shows the 3D conformation of the ligand generated from the swarm_60_
115.pdb file. It is a visual representation of how the ligand fits into the target binding
site.
3. Scatter Plot of Fnc Versus Score
Figure 4 is a scatter plot which shows the relationship between the fnc (function value)
and the Score. It helps to identify any correlation between the scoring function and
the function value obtained during docking. A positive correlation might suggest that
higher function values correspond to higher scores.
398 V. S. Kaza et al.

Fig. 2 Conformer generated

from swarm_22_112.pdb

Fig. 3 Conformer generated

from swarm_60_115.pdb

4. Scatter Plot of i-RMSD Versus Score

Figure 5 is a scatter plot which displays the relationship between the i-RMSD
(interface-root-mean-square deviation) and the Score. It helps understand how well
the docking poses align with the target’s binding site. Lower i-RMSD values generally
indicate more accurate docking predictions.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 399

Fig. 4 Scatter Plot of fnc Versus Score based on Ligand Clustered Rank Results

Fig. 5 Scatter plot of i-RMSD Versus Score based on ligand clustered rank results
400 V. S. Kaza et al.

Fig. 6 Scatter plot of L-RMSD Versus Score based on ligand clustered rank results

5. Scatter Plot of L-RMSD vs. Score

Figure 6 is a scatter plot which shows the relationship between the L-RMSD (ligand-
root-mean-square deviation) and the Score. It helps assess the accuracy of the docking
poses in predicting the ligand’s conformation. Lower L- RMSD values indicate better
predictions.
6. Bar Plot of Structure vs. Score (Top 10)
Figure 7 depicts a bar plot that displays the top 10 structures ranked by their Scores.
It allows quick identification of the best-performing docking poses based on the
scoring function. Higher scores imply more favorable binding interactions.

7. Box Plot of Score

In Fig. 8, the box plot provides a visual representation of the distribution of the
Score values. It shows the median, quartiles, and possible outliers. It helps to assess
the consistency of the scoring function and identify any extreme scores that might
require further investigation.
8. Pair Plot of Pairwise Relationships Across Columns
In Fig. 9, the pair plot presents pairwise scatter plots of all numerical columns (fnc,
i-RMSD, L-RMSD, and Score). It enables us to observe correlations and patterns
between different variables at once. For example, it can help identify if any of the
docking metrics are correlated with the scoring values.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 401

Fig. 7 Bar plot of Structure Versus Score (top 10) based on ligand clustered rank results

Fig. 8 Box plot of score based on ligand clustered rank results

402 V. S. Kaza et al.

Fig. 9 Pair plot of pairwise relationships across columns based on ligand clustered rank

9. Line Plot of i-RMSD vs. L-RMSD

In Fig. 10, the line plot shows the relationship between i-RMSD and L-RMSD values.
It might help understand if there is any correlation between the RMSD values related
to the target interface and the ligand’s conformational changes.
10. Correlation Matrix
The correlation matrix in Fig. 11 is a tabular representation of the correlation coef-
ficients between different variables (fnc, i-RMSD, L-RMSD, and Score). It allows
a quick overview of the relationships between these metrics. Positive correlations
suggest that an increase in one metric corresponds to an increase in another, while
negative correlations imply an inverse relationship.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 403

Fig. 10 Line plot of i-RMSD Versus L-RMSD based on ligand clustered rank results

Fig. 11 Correlation matrix

404 V. S. Kaza et al.

Fig. 12 Clustering of
ligands based on their
coordinates (from scoring
ranking results)

11. Plot Showing the Clustering of Ligands Based on Their Coordinates

The plot in Fig. 12 shows a clustering analysis of ligands based on their 3D coor-
dinates (docking poses) and Scores. It helps identify groups of ligands with similar
binding modes and potentially similar activities. Clustering can aid in identifying
ligand structural diversity and understanding the ligand-binding landscape. It is based
on the results obtained that are given in Table 4.

7 Enhancements and Future Perspectives

7.1 Hybridization with Other Algorithms

In the context of molecular docking and drug discovery, “Hybridization with Other
Algorithms” refers to the integration or combination of multiple computational
methods or algorithms to enhance the accuracy, efficiency, and reliability of the
docking process [7]. Hybrid approaches often aim to capitalize on the strengths of
individual algorithms while mitigating their limitations. Here are some examples of
hybridization in molecular docking.
1. Ligand-based and Structure-based Methods: Ligand-based methods, such
as pharmacophore modeling and quantitative structure–activity relationship
(QSAR) analysis, rely on known ligand properties and activities to predict new
ligands. Structure-based methods, like molecular docking, use the 3D structure of
the target receptor to predict ligand-receptor interactions [12]. Hybrid approaches
Table 4 Scoring ranking results (Overview)
Swarm Glowworm Coordinates Rec ID Lig ID Luciferin Neigh VR RMSD PDB Scoring
22 112 (−8.625, 5.4, 33.18, 0.97, 0.05, 0 0 42.55054 2 1.12 −1 lightdock_112.pdb 28.735
0.134, 0.197)
37 11 (−6.527, 9.572, 28.154, −0.06, 0 0 41.19191 2 0.4 −1 lightdock_11.pdb 28.152
0.186, -0.935, −0.295)
39 11 (−10.535, 7.845, 29.603, −0.176, 0 0 40.33904 0 5 −1 lightdock_11.pdb 26.893
0.177, −0.901, −0.355)
60 115 (0.12, 9.729, 29.967, −0.252, 0 0 40.21248 0 5 −1 lightdock_115.pdb 26.808
0.073, −0.928, −0.266)
54 167 (−5.93, 10.812, 28.775, 0.892, − 0 0 38.30142 3 1.28 −1 lightdock_167.pdb 25.651
0.383, −0.161, −0.179)
Optimizing Drug Discovery: Molecular Docking with Glow-Worm …
405
406 V. S. Kaza et al.

can combine information from both methods to identify and prioritize potential
ligands for docking, improving the overall success rate.
2. Docking with Molecular Dynamics (MD) Simulations: MD simulations can
provide valuable information about the dynamic behavior of ligand-receptor
complexes. By combining docking with MD simulations, researchers can study
the stability of the docked poses over time and gain insights into the binding
kinetics [11]. This hybrid approach is particularly useful when considering the
flexibility of the receptor and the ligand.
3. Machine Learning and Deep Learning Integration: Machine learning algorithms
can be trained to predict binding affinities or docking scores based on known
ligand-receptor interactions. Integrating machine learning models with docking
methods can help refine the scoring functions and improve the accuracy of the
binding affinity predictions. Deep learning techniques, such as neural networks,
have also shown promise in predicting molecular interactions.
4. Free Energy Calculations: Docking provides a snapshot of the ligand-receptor
interaction, but it doesn’t directly account for the energetic contributions to
binding. Hybrid approaches can incorporate free energy calculations, such as
molecular mechanics/Poisson-Boltzmann surface area (MM/PBSA) or molec-
ular mechanics/generalized Born surface area (MM/GBSA), to estimate binding
free energies and rank ligands accordingly.
5. Solvation Models: Properly accounting for solvent effects is essential in accu-
rately predicting ligand-receptor interactions [10]. Hybrid approaches can
include solvation models, such as implicit solvent models or explicit solvent
molecular dynamics, to improve the accuracy of docking predictions.
6. Enhanced Sampling Techniques: Standard docking algorithms might miss poten-
tial ligand-binding poses due to limited conformational sampling. Hybridization
with enhanced sampling techniques like Monte Carlo methods or genetic algo-
rithms can explore a broader conformational space and increase the chances of
finding the best binding pose.
7. Integration of Experimental Data: Hybrid approaches can incorporate experi-
mental data, such as NMR spectroscopy or site-directed mutagenesis, to validate
and refine docking predictions [13]. This integration helps in identifying key
interactions and validating the predicted binding mode.
Overall, hybridization with other algorithms and techniques is an active area of
research in molecular docking and drug discovery. These integrated approaches hold
the potential to provide more accurate predictions of ligand-receptor interactions and
accelerate the identification of promising drug candidates. However, it is essential to
carefully validate and benchmark the hybrid methods to ensure their reliability and
effectiveness.
Here are some potential ways to combine GSO with other optimization algorithms.
1. GSO with Genetic Algorithms (GA): Genetic Algorithms are well-known opti-
mization techniques inspired by the process of natural selection. They involve the
use of techniques like crossover, mutation, and selection to evolve a population of
candidate solutions [15]. By combining GSO with GA, one can introduce genetic
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 407

operators to manipulate the glow-worm positions, such as crossover between

glow-worms or mutation of glow-worm positions. This combination may lead
to a more diverse exploration of the search space and potentially improve the
quality of solutions.
2. GSO with Particle Swarm Optimization (PSO): Particle Swarm Optimization is
another population-based optimization technique that imitates the social behavior
of birds flocking or fish schooling [5]. In PSO, particles adjust their positions
based on their own experience and the experience of the best-performing particle
in the swarm. Integrating GSO with PSO can lead to a hybrid swarm algorithm
where the glow-worms and particles cooperate, sharing information about the
search space and guiding each other toward promising regions.
3. GSO with Simulated Annealing (SA): Simulated Annealing is a global optimiza-
tion algorithm inspired by the annealing process in metallurgy. It involves a
controlled random search that allows the algorithm to escape local optima [7].
By combining GSO with SA, the glow-worm swarm can initially perform an
intensive search (exploitation) to converge toward promising solutions, and then
gradually transition to a more explorative search to escape potential local optima
and refine the solutions.
4. Hybrid GSO Ensemble: Another approach is to create an ensemble of optimiza-
tion algorithms, including GSO and other algorithms like GA, PSO, SA, or
Differential Evolution (DE). Each algorithm in the ensemble can run indepen-
dently and contribute its solutions. The final solution can be a combination of
solutions from different algorithms, either through voting or weighted averaging.
This ensemble approach can provide more robust and diverse solutions, especially
when the algorithms used have complementary search strategies.
5. Adaptive Hybridization: Instead of using fixed combinations, an adaptive
approach can be designed where the hybridization strategy dynamically changes
during the optimization process. The algorithm can assess the performance of
different combinations and adaptively adjust the contributions of each algorithm
based on their effectiveness in different regions of the search space.
It is important to note that while hybridizing GSO with other optimization algo-
rithms shows potential for enhancing performance, it also introduces additional
complexity. Careful parameter tuning and validation are crucial to ensuring that
the hybrid approach indeed improves the efficiency and effectiveness of the opti-
mization process [22]. Additionally, the choice of the specific hybridization method
should be guided by the problem’s characteristics and the behavior of the algorithms
being combined.
408 V. S. Kaza et al.

7.2 Methods to Parallelize GSO for Faster and More Efficient

Molecular Docking

Parallelization is a powerful technique to speed up the execution of algorithms like

GSO (Glow-worm Swarm Optimization) by taking advantage of multiple processing
units or cores. In the context of molecular docking, parallelizing GSO can signifi-
cantly reduce the optimization time, making it more efficient and practical for large-
scale docking studies [13]. Here are some methods to parallelize GSO for faster and
more efficient molecular docking.
1. Parallel Initialization: One straightforward way to parallelize GSO is by initial-
izing multiple independent glow-worm swarms simultaneously. Each swarm can
start from different positions in the search space. This approach allows the
algorithm to explore multiple regions of the conformational space in parallel.
Researchers can distribute the swarms across multiple processors or threads,
taking advantage of modern multi-core CPUs.
2. Spatial Decomposition: In molecular docking, the search space typically repre-
sents the spatial arrangement of the ligand and receptor molecules [11]. Spatial
decomposition involves dividing the search space into smaller subspaces and
assigning each subspace to a different processor or thread. This way, each
processor/thread optimizes a smaller part of the search space independently,
leading to faster convergence.
3. Master–Slave Configuration: In this approach, one processor (the master)
controls and coordinates the optimization process, while the other processors
(slaves) perform the actual GSO computations. The master receives updates from
all the slaves, exchanges information, and directs the optimization process [2].
This configuration is useful when global information is needed to guide the
search.
4. Island Model: The island model is a parallelization technique commonly used
in evolutionary algorithms. It involves running multiple independent GSO popu-
lations (islands) in parallel. Periodically, some glow-worms from each island
migrate to other islands, allowing information exchange between populations
[14]. This migration helps in sharing good solutions and promoting diversity.
5. GPU Acceleration: Graphics Processing Units (GPUs) are well-suited for parallel
computation due to their thousands of cores. Certain GSO operations, such as the
computation of the objective function, can be parallelized and offloaded to GPUs
for significant speedup. This approach requires specialized GPU programming,
typically using libraries like CUDA or OpenCL.
6. Asynchronous GSO: In traditional GSO, all glow-worms update their positions
synchronously based on the information available at that time [18]. Asynchronous
GSO allows glow-worms to update their positions independently as soon as new
information becomes available. This approach reduces waiting time and can be
naturally parallelized.
7. Hybrid CPU-GPU Approach: To further boost performance, a hybrid approach
can be employed, where CPU and GPU resources are combined. The
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 409

computation-intensive parts of GSO can be parallelized on GPUs, while other

tasks that are more suitable for CPUs can run in parallel on multi-core processors.
When implementing parallel GSO, researchers should carefully consider load
balancing, communication overhead, and the potential for race conditions or conflicts
when multiple threads/processes access shared resources [22]. Additionally, the
performance gains achieved through parallelization might depend on the size and
complexity of the docking problem and the available computational resources. Proper
benchmarking and profiling are necessary to select the most suitable parallelization
strategy for a given molecular docking scenario.

7.3 Enhancing Docking Accuracy with Machine Learning

7.3.1 Introduction to Machine Learning Models for Docking

In the context of molecular docking, machine learning (ML) models can be employed
to enhance the accuracy and efficiency of the docking process. Molecular docking is a
computationally intensive task that aims to predict the optimal binding conformation
and affinity of a ligand with a target receptor [5]. Traditional docking algorithms often
rely on physics-based force fields and scoring functions, which may have limitations
in accurately capturing the complexities of molecular interactions.
Machine learning models, on the other hand, offer a data-driven approach to tackle
docking challenges by learning patterns and relationships from large datasets of
known ligand-receptor interactions. These models can then be used to predict binding
affinities, rank ligand poses, and improve the accuracy of docking predictions. Some
key aspects of using machine learning models for docking include.
1. Training Data: The success of a machine learning model relies heavily on the
quality and diversity of the training data. This data typically consists of exper-
imentally determined ligand-receptor structures along with their corresponding
binding affinities. Generating a comprehensive and well-curated dataset is crucial
for training ML models effectively.
2. Feature Engineering: To represent ligand-receptor interactions as input for the
ML model, relevant features need to be extracted [5]. These features could include
geometric descriptors, physicochemical properties, molecular fingerprints, or
structural information. The selection of informative features significantly impacts
the performance of the ML model.
3. Model Selection: Various ML algorithms can be applied to docking problems,
such as Random Forest, Support Vector Machines, Gradient Boosting, Neural
Networks, and more. The choice of the ML model depends on the complexity of
the problem, size of the dataset, and computational resources available.
410 V. S. Kaza et al.

4. Scoring Function Development: ML models can be used to develop new scoring

functions that better capture the ligand-receptor interactions. These scoring func-
tions can then be employed in docking algorithms to guide the search for optimal
binding conformations.
5. Virtual Screening: ML models can aid in virtual screening, where a large database
of chemical compounds is screened to identify potential ligands for a given target
receptor [5]. ML-based virtual screening approaches can significantly reduce the
number of compounds that need to be experimentally tested.
6. Ensemble Methods: Combining multiple ML models using ensemble methods
can lead to improved docking accuracy. Ensemble techniques, such as averaging
predictions from multiple models or using a weighted combination of models,
can enhance the robustness and generalization of the predictions.
7. Transfer Learning: Transfer learning is another valuable approach where pre-
trained ML models on related tasks are fine-tuned for docking purposes. This
leverages knowledge from existing models and accelerates the training process.
It is important to note that while machine learning can enhance docking accu-
racy, it is not a substitute for physics-based docking algorithms. Rather, ML models
complement traditional methods and can offer valuable insights and predictions,
especially when dealing with complex molecular interactions and large chemical
spaces. As with any ML application, careful validation and testing are essential to
ensure the reliability and generalizability of the ML-based docking predictions.

7.4 Comparing Machine Learning and Classical Docking

Approaches

Comparing machine learning (ML) and classical docking approaches is essential

to understand the strengths and limitations of each method. Both approaches aim
to predict ligand-receptor interactions and binding affinities, but they do so use
different underlying principles [5]. Here is a comparison of ML and classical docking
approaches.
• Classical Docking Approaches
1. Physics-Based Methods: Classical docking methods are typically based on
physics and chemistry principles. They use force fields, scoring functions,
and geometric complementarity to predict ligand binding modes and binding
energies.
2. Rigid Body Docking: Classical docking methods often assume rigid structures
for both ligand and receptor, neglecting conformational changes upon binding.
3. Scoring Functions: These methods utilize scoring functions to evaluate the
fitness of a ligand-receptor complex [21]. The scoring functions estimate
binding affinities based on energy terms, van der Waals forces, hydrogen
bonding, electrostatic interactions, etc.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 411

4. Computationally Efficient: Classical docking methods are generally faster

compared to ML approaches, which is advantageous for high-throughput
virtual screening of large compound libraries.
5. Limited Sampling: Classical docking methods rely on a predefined search
space and discrete sampling of ligand conformations, which might miss some
potential binding modes.

• Machine Learning Approaches

1. Data-Driven Models: ML approaches learn from data and can capture complex
non-linear relationships between ligand and receptor features, making them
more flexible than physics-based methods.
2. Flexibility: ML models can account for receptor flexibility and conformational
changes, allowing better prediction of induced fit and ligand adaptation.
3. Transfer Learning: ML models can benefit from transfer learning, where pre-
trained models on related tasks can be fine-tuned for docking tasks, enhancing
performance, especially with limited data.
4. Feature Engineering: ML requires careful feature engineering to represent
ligand-receptor interactions effectively. Choosing relevant features is crucial
for model performance.
5. Data Requirements: ML approaches require large and diverse datasets to
generalize well, and collecting such datasets can be challenging for specific
protein targets.
6. Interpretability: Classical docking methods provide more interpretable results
since the scoring functions are based on physics and chemistry princi-
ples. In contrast, ML models can be considered “black boxes” with limited
interpretability.

• Hybrid Approaches

Researchers often combine classical docking and ML approaches to leverage the

benefits of both methods. For instance, ML models can be used to rescore docking
results from physics-based methods, improving the accuracy of binding affinity
predictions. Additionally, ML can guide the selection of promising docking poses
for further refinement and experimental testing.
In summary, classical docking approaches excel in speed and interpretability but
have limitations in handling receptor flexibility and accounting for complex ligand-
receptor interactions. On the other hand, ML approaches offer flexibility and potential
for higher accuracy but require large and diverse datasets for training and can be
computationally demanding. The choice between these approaches depends on the
specific research objectives, available data, and computational resources.
412 V. S. Kaza et al.

8 Applications of Molecular Docking Using 3 × 29 Dataset

8.1 Identifying Potential Drug Candidates

The application of molecular docking using the 3 × 29 dataset can be instrumental

in identifying potential drug candidates. The 3 × 29 dataset contains information
about the ligands and their interactions with the receptor, which is a protein of
interest. By employing molecular docking techniques, we can virtually screen a vast
number of ligands and prioritize those with the highest binding affinity and favorable
interactions with the receptor. Here is how the process works:
1. Virtual Screening: The 3 × 29 dataset may consist of thousands to millions of
chemical compounds. Virtual screening involves docking each ligand into the
receptor’s binding site to predict how well they can bind. Scoring functions are
used to estimate the binding affinity, which reflects the strength of ligand-receptor
interactions.
2. Ranking Ligands: After docking and scoring, the ligands are ranked based on
their binding affinity scores. Ligands with higher scores are more likely to have
strong interactions with the receptor and, thus, are potential drug candidates.
3. Filtering and Optimization: The ranked list of ligands can be further filtered
based on certain criteria, such as drug-likeness, pharmacokinetic properties, and
chemical diversity. This step helps in selecting ligands that are more likely to be
successful in subsequent stages of drug development.
4. Experimental Validation: The top-ranked ligands from the docking study can then
be synthesized and tested experimentally to validate their binding affinity and
biological activity. Experimental validation is crucial to confirm the predictions
made by the docking simulations.
5. Lead Optimization: Once potential drug candidates are identified and vali-
dated, medicinal chemists can perform lead optimization to improve the ligands’
potency, selectivity, and other drug-like properties. This iterative process aims to
refine the initial hits into optimized drug candidates.
6. Clinical Trials: Finally, the most promising drug candidates undergo preclin-
ical studies and clinical trials to evaluate their safety and efficacy in humans.
Successful candidates can then be approved for use as pharmaceutical drugs.
Molecular docking using the 3 × 29 dataset can significantly accelerate the drug
discovery process by reducing the number of compounds that need to be experi-
mentally tested. It allows researchers to focus their resources on the most promising
candidates, saving time and costs. However, it’s important to remember that docking
predictions are computational and must be validated experimentally to ensure accu-
racy and reliability. Additionally, molecular docking is just one step in the drug
discovery pipeline, and other computational and experimental techniques play vital
roles in the overall process.
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 413

9 Conclusion

9.1 Significance of the Study

The application of the GSO-based molecular docking approach on the 3 × 29 dataset

has several significant implications:
1. Efficient Exploration of Chemical Space: GSO-based docking allowed us to effi-
ciently explore a vast chemical space by considering multiple ligand conforma-
tions and orientations, enabling a more comprehensive search for potential drug
candidates.
2. Cost-Effective and Time-Efficient: By leveraging computational methods, the
GSO-based approach significantly reduces the cost and time required for initial
drug screening, making it a valuable tool in early-stage drug discovery.
3. Identification of Novel Drug Candidates: The approach helped in identifying
ligands with high binding affinities that may serve as novel drug leads or scaffolds
for further development.
4. Insights into Ligand–Protein Interactions: The analysis of docking results
provided valuable insights into the key interactions between ligands and the
target protein’s binding site, guiding the rational design of potential drugs.

9.2 Future Directions and Potential Research Opportunities

for CS Majors in the Field of Molecular Docking

For Computer Science (CS) majors interested in the field of molecular docking, there
are several exciting future directions and potential research opportunities that can
contribute to the advancement of this area. Here are some key areas to explore:
1. Scoring Function Development: Developing more accurate and reliable scoring
functions is a crucial research area in molecular docking. CS majors can work
on novel scoring functions that better capture the complex interactions between
ligands and target proteins. Machine learning and deep learning techniques can
be utilized to derive scoring functions from large datasets of known ligand–
protein complexes.
2. Machine Learning in Docking: Integrating machine learning approaches into
molecular docking can lead to significant improvements in prediction accu-
racy and efficiency. CS majors can explore the application of various machine
learning models, such as neural networks, support vector machines, and random
forests, to enhance different stages of the docking process.
3. High-Performance Computing (HPC) and Parallelization: Molecular docking
involves computationally intensive tasks. CS majors can focus on devel-
oping parallel algorithms and utilizing high-performance computing resources,
including distributed computing and GPU acceleration, to speed up docking
calculations and handle large-scale docking studies efficiently.
414 V. S. Kaza et al.

4. Data Augmentation and Transfer Learning: Applying data augmentation tech-

niques and transfer learning from one docking dataset to another can help
address the scarcity of experimentally validated binding data. CS majors
can investigate methods to leverage knowledge gained from related docking
experiments to improve predictions for new target proteins.
5. Docking with Flexible Receptors: Most docking studies assume rigid protein
structures, but proteins are dynamic and can undergo conformational changes
upon ligand binding. CS majors can work on algorithms that account for receptor
flexibility, allowing for more accurate predictions of ligand–protein interactions.
6. Virtual Screening and Drug Repurposing: CS majors can explore virtual
screening methods to identify potential drug candidates from large compound
databases. Additionally, drug repurposing, which involves finding new uses for
existing drugs, is an area where computational methods can be instrumental in
predicting novel interactions.
7. Interpretable AI for Docking: As AI-driven approaches become more prevalent,
the ability to interpret and explain the results becomes crucial. CS majors can
investigate methods to make AI-driven docking predictions more interpretable
and transparent to aid decision-making in drug discovery.
8. Cloud Computing for Docking: Leveraging cloud computing resources can
provide scalable and cost-effective solutions for docking studies. CS majors
can develop tools and frameworks for running docking simulations on cloud
platforms, making them accessible to researchers worldwide.
9. Multi-Objective Optimization in Docking: Molecular docking often involves
multiple conflicting objectives, such as maximizing binding affinity while
minimizing ligand size or hydrophobicity. CS majors can work on multi-
objective optimization algorithms to help researchers find a balance between
these objectives.
10. Integration of Structural Biology Data: Integrating docking results with other
structural biology data, such as protein–protein interaction networks, can
provide a systems-level understanding of cellular processes and disease mech-
anisms. CS majors can explore methods for data integration and visualization
to extract meaningful insights.
11. Docking with Metal Ions and Cofactors: Incorporating metal ions and cofac-
tors into docking simulations is challenging due to their unique coordina-
tion geometries. CS majors can investigate ways to handle metal-protein and
cofactor-protein interactions more accurately.
12. Graph-Based Docking: Representing ligands and proteins as graphs and
applying graph-based algorithms for docking can be a promising research direc-
tion. CS majors can explore graph neural networks and graph similarity methods
for this purpose.
By delving into these research areas, CS majors can make valuable contributions
to the field of molecular docking, advancing drug discovery efforts and our under-
standing of ligand–protein interactions. These opportunities lie at the intersection of
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 415

computer science, biology, and chemistry, offering an exciting and impactful field
for interdisciplinary research.
Statements and Declarations Competing Interests The authors declare that there are no
competing interests associated with this research. No financial or non-financial interests have influ-
enced the design, data collection, analysis, interpretation, or reporting of this study. The research
presented in this paper is conducted with complete impartiality and integrity.

Funding This research received no specific grant from any funding agency in the public, commer-
cial, or not-for- profit sectors. The authors conducted this study without any external financial
support, demonstrating their commitment to independent and unbiased research.

Ethics Approval This research adheres to all ethical guidelines and principles. All necessary
ethical approvals were obtained from the appropriate institutional review boards and governing
bodies before commencing this study. Participants were provided with informed consent, and their
privacy and confidentiality were strictly maintained throughout the research process.

Data Availability The data used in this study are available upon reasonable request from the
corresponding author or are available open on Kaggle & GitHub(dataset). The authors are committed
to promoting transparency and accessibility, and they are willing to share the data with other
researchers for scientific purposes.

Author Contributions The author Vijaya Sindhoori Kaza conceived and designed the study,
performed data collection and analysis, and contributed significantly to writing and revising the
manuscript. The research work has been supervised and reviewed by Dr P R Anisha and Dr. C
Kishor Kumar Reddy.

Informed Consent Informed consent was obtained from all participants involved in this study.
Participants were informed about the purpose of the research, their role in the study, and the potential
risks and benefits associated with their participation. Written consent was obtained before data
collection, ensuring full comprehension and voluntary participation.

Copyright and Permissions The authors declare that all material used in this paper does not need
any permissions.

Conflict Resolution In the event of any disputes or disagreements related to this research, the
authors agree to address them through amicable discussions and mutual understanding, with the
aim of reaching a resolution that upholds the principles of scientific integrity and collaboration.

Publication Ethics The authors affirm their adherence to strict publication ethics guidelines. This
research was conducted with the utmost integrity and in compliance with ethical standards in
research and scholarly publishing.

Correspondence For any correspondence related to this research, please contact the corresponding
author, Vijaya Sindhoori Kaza @email([email protected]).
416 V. S. Kaza et al.

References

1. Adelusi, Temitope Isaac, Abdul-Quddus Kehinde Oyedele, Ibrahim Damilare Boyenle, Abdeen
Tunde Ogunlana, Rofiat Oluwabusola Adeyemi, Chiamaka Divine Ukachi, Mukhtar Oluwaseun
Idris et al. (2022) Molecular modeling in drug discovery. Inform Med Unlocked 29: 100880
2. Subbarayudu B, Lalitha Gayatri L, Sai Nidhi P, Ramesh P, Gangadhar Reddy R, Kishor Kumar
Reddy C (2017) Comparative analysis on sorting and searching algorithms. Int J Civ Eng
Technol
3. Bagal A, Borkar T, Ghige T, Kulkarni A, Kumbhar A, Devane G, Rohane S (2022) Molecular
Docking-Useful Tool in Drug Discovery. Asian J Res Chem 15(2):129–132
4. Brian Jiménez-García and others (2018) LightDock: a new multi-scale approach to protein–
protein docking. Bioinform 34(1):49–55
5. Kishor Kumar Reddy C, Anisha PR, Srinivasulu Reddy K, Surender Reddy S (2012) Third
party data protection applied to cloud and XACML implementation in the Hadoop environment
with sparql. IOSR J Comput Eng ISSN
6. Corso G, Stärk H, Jing B, Barzilay R, Jaakkola T (2022) Diffdock: Diffusion steps, twists, and
turns for molecular docking. arXiv preprint arXiv:2210.01776
7. Crampon K, Giorkallos A, Deldossi M, Baud S, Steffenel LA (2022) Machine-learning methods
for ligand–protein molecular docking. Drug Discov Today 27(1):151–164
8. García-Ortegón M, Simm GN, Tripp AJ, Hernández-Lobato JM, Bender A, Bacallado S (2022)
DOCKSTRING: easy molecular docking yields better benchmarks for ligand design. J Chem
Inf Model 62(15):3486–3502
9. Garcia-Ruiz, Miguel, Pedro Cesar Santana-Mancilla, Laura Sanely Gaytan-Lugo, and Adriana
Iniguez-Carrillo (2022) Participatory design of sonification development for learning about
molecular structures in virtual reality. Multimodal Technol Interact 6, 10(2022): 89
10. Kiruba Nesamalar E, Satheesh Kumar J, Amudha T (2022) Efficient DNA-ligand interaction
framework using fuzzy C-means clustering based glowworm swarm optimization (FCMGSO)
method. J Biomol Struct Dyn 1–13
11. Li T, Guo R, Zong Q, Ling G (2022) Application of molecular docking in elaborating molecular
mechanisms and interactions of supramolecular cyclodextrin. Carbohyd Polym 276:118644
12. Narasimha Prasad LV, Shankar Murthy P, Kishor Kumar Reddy C (2013) Analysis of magnitude
for earthquake detection using primary waves and secondary waves, IEEE, 23
13. Meng XY, Zhang HX, Mezei M, Cui M (2011) Molecular docking: a powerful approach for
structure- based drug discovery. Curr Comput Aided Drug Des 7(2):146–157
14. Morozov, Dmitry, Artem Melnikov, Vishal Shete, Michael Perelshtein (2023) Protein-protein
docking using a tensor train black-box optimization method. arXiv preprint arXiv:2302.03410
15. Singh S, Baker QB, Singh DB (2022) Molecular docking and molecular dynamics simulation.
In Bioinformatics (pp 291–304). Academic Press
16. Sunny, Sharon, Gautham Sreekumar (2022) SFLADock: a memetic Protein-Protein docking
algorithm. In 2022 IEEE International Conference on Distributed Computing and Electrical
Circuits and Electronics (ICDCECE), pp. 1–6. IEEE
17. Sunny, Sharon, Jayaraj PB (2022) Protein–protein docking: Past, present, and future. Protein J
1–26
18. Tessaro F, Scapozza L (2020) How ‘Protein-Docking’ translates into the new emerging field
of docking small molecules to nucleic acids? Molecules 25(12):2749
19. Vidal-Limon A, Aguilar-Toalá JE, Liceaga AM (2022) Integration of molecular docking anal-
ysis and molecular dynamics simulations for studying food proteins and bioactive peptides. J
Agric Food Chem 70(4):934–943
Optimizing Drug Discovery: Molecular Docking with Glow-Worm … 417

20. Wong F, Krishnan A, Zheng EJ, Stärk H, Manson AL, Earl AM, Jaakkola T, Collins JJ (2022)
Benchmarking AlphaFold-enabled molecular docking predictions for antibiotic discovery. Mol
Syst Biol 18(9):e11081
21. Xue, Qiao, Xian Liu, Paul Russell, Jin Li, Wenxiao Pan, Jianjie Fu, Aiqian Zhang (2022)
Evaluation of the binding performance of flavonoids to estrogen receptor alpha by Autodock,
Autodock Vina and Surflex-Dock. Ecotoxicol Environ Saf 233(2022): 113323
22. Yang, Chao, Eric Anthony Chen, Yingkai Zhang. Protein–ligand docking in the machine-
learning era. Molecules 27(14): 4568
Bridging the Gap Between Learning
and Security: An Investigation
of Blended Learning Systems
and Ransomware Detection

Vikas Pandey, Aman Jolly, Praveen Kumar Malik, and Shashikant

Abstract Blended learning has been identified as a potentially viable strategy for
enhancing the efficacy and efficiency of online learning through the integration of
conventional instructional approaches. Despite the potential advantages, educational
institutions have exhibited reluctance in embracing this strategy due to a range
of difficulties. A prominent issue of concern pertains to the escalating menace
posed by Ransomware virus assaults, which have the potential to result in substan-
tial financial ramifications and reputational harm. This study aims to examine the
many elements that impact student satisfaction in utilising blended learning systems,
focusing primarily on the modules, channels, and lecturers involved. This evaluation
seeks to improve students’ comprehension of literacy within the context of classroom
discourses. In addition, our study aims to enhance the precision of Ransomware
detection by examining the intricate characteristics of the malicious software, partic-
ularly by analysing assembly language instruction patterns. The N-gram technique
is employed in a two-stage process for feature extraction. This procedure involves
the computation of pattern statistics and subsequent feature selection to effectively
reduce dimensionality. In order to assess the efficacy of the chosen features, we
employ a feature cataloguing methodology with the Random Forest algorithm, util-
ising the lowest out-of-bag (OOB) error and a predetermined number of trees. Our
experiments showcase that this method achieved the highest accuracy, sensitivity
value, false positive rate, and precision. On its whole, our study presents a thorough
methodology for improving student happiness and virus detection accuracy inside
blended learning systems. Through the use of our research outcomes, educational
establishments have the potential to furnish pupils with a learning encounter that is
both captivating and efficacious, while concurrently minimising the vulnerabilities
associated with cyber-attacks.

V. Pandey · Shashikant
Babu Banarasi Das University, Lucknow, UP, India
A. Jolly (B)
KIET Group of Institutions, Ghaziabad, UP, India
e-mail: [email protected]
P. K. Malik
Lovely Professional University, Phagwara, Punjab, India

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 419
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_18
420 V. Pandey et al.

Keywords Blended learning · Ransomware · Energy and industry applications ·

Random forest

1 Introduction

The implementation of the Blended Learning System (BLS) project in educational

institutions has shown great promise. However, due to the unfamiliarity and discom-
fort with this platform, getting institutional consensus has proven difficult A compre-
hensive dialogue between decision makers and employees is essential for success.
The complexity of the session management in BLS requires dedicated technical staff
and ongoing support. Additionally, policy development and tight assessment and
measurement are crucial to support the expected quality of BLS.
One of the challenges is the detection of Ransomware virus accurately, which
can cause significant financial loss as well as enterprise reputation. To address this,
this work focuses on developing more precise features of the virus using N-gram
algorithm and Random Forest algorithm for classification. The initial implementa-
tion of BLS faced challenges in areas such as ownership of intellectual property,
copyright, and workload of each member of the institution. Therefore, it is neces-
sary to address these problems at the policy level and ensure commitment from all
concerned members for the long-term success of BLS. In this paper, we discuss the
material and methodology used, performance, experimental results and discussions,
and conclude the work.
The adoption of blended learning has gained interest in informal educational
institutions, leading decision makers to initiate the implementation of the Blended
Learning System (BLS) project in their learning environments. However, achieving
institutional consensus for BLS adoption may prove challenging due to unfamiliarity
and discomfort with the platform. BLS implementation requires consistency between
the objectives of decision makers and employees, who may find it difficult to inte-
grate BLS with institutional structure and strategy, which traditionally rely on formal
teaching and lecturing. Government regulation policy has also contributed to slow
response to BLS implementation due to a shortage of engagement and fear of tech-
nological impact. Additionally, top-down initiatives may be met with suspicion by
institution members who perceive a negative impact on activity processes, focusing
on quality, intellectual property, and ownership. To successfully implement BLS,
comprehensive dialogue between all involved parties is necessary. Despite previous
attempts, gaps in understanding between executives, management, and faculty have
not been satisfactorily addressed in past online learning initiatives [1].
The complexity involved in session management within the Blended Learning
System (BLS) necessitates the allocation of dedicated technical personnel and
ongoing support. It is crucial for students and their instructors to have access to
a reliable learning process that incorporates an intervention mechanism to monitor
motivation and progress. Furthermore, the extension of the learning environment
to the internet underscores the importance of providing sufficient content resources
Bridging the Gap Between Learning and Security: An Investigation … 421

that are easily accessible. However, the management of content systems, network
connectivity, bandwidth levels, and program releases may prove overwhelming
and hinder the reliability of the system. Therefore, policy development must be
timely and accompanied by rigorous assessments and measurements to ensure the
expected quality of BLS and prevent the emergence of complex resistance. Notably,
Ransomware, a type of malware that encrypts the victim’s files and demands a ransom
for data retrieval, poses a significant challenge in terms of accurate detection. Failure
to accurately detect the virus at the outset can lead to significant financial losses and
damage to the enterprise’s reputation.
The identification and categorisation of malicious software, commonly referred
to as malware, are of utmost importance within the realm of computer security. In
recent years, scholars have directed their attention to the development of techniques
aimed at extracting characteristics from malware files with the objective of effectively
categorising and identifying them. One approach that might be employed entails the
extraction of viral features in the form of operation codes (opcodes) from files that
possess the.exe format.
Prior studies have employed machine learning methods, including Random Forest,
Decision Tree, K-Nearest-Neighbor, and Support Vector Machine, for the purpose
of classifying malware files by analysing their opcodes. One study conducted by
researchers [1] found that the Random Forest algorithm demonstrated the greatest
accuracy rate of 98.7% when utilised for the classification of malware files. In
a separate investigation [2], datasets including both normal and malicious files
were acquired by unpacking files from the Windows Operating System using UPX
Unpacker and IDA Pro. The research conducted successfully identified the primary
opcodes utilised in malware files. These opcodes, including MOV, PUSH, CALL,
CMP, POP, JZ, TEST, JNZ, JMP, ADD, LEA, XOR, RETN, AND, SUB, OR, INC,
MOVZX, DEC, and JB, were found to be prevalent in the analysed malware samples.
Nevertheless, despite the significant progress made in this field, the process of
effectively identifying and classifying malware continues to present considerable
challenges. Hence, the objective of this study is to enhance the precision of viral
characteristics in order to enhance the efficacy of malware detection and classifica-
tion. The primary objective of this study is to investigate the fundamental attributes of
viruses by use of a two-stage methodology that involves the analysis of their assembly
language instructions. The initial phase involves the utilisation of the N-gram tech-
nique to extract the features. These characteristics are subsequently subjected to
sequential pattern searching and statistical analysis in order to ascertain their rele-
vance. Ultimately, a feature selection strategy is implemented in order to decrease
the dimensionality of the data. The subsequent phase entails the validation of the
chosen attributes to assess their efficacy in discerning malicious software.
In summary, the precise identification and categorisation of malicious software
play a crucial role in safeguarding the integrity and protection of computer systems.
The objective of this research is to boost the accuracy of malware detection and
classification by refining the properties of the virus in a more exact manner, thereby
building upon prior studies. By employing sophisticated algorithms and meticulous
422 V. Pandey et al.

feature selection, our aim is to achieve substantial advancements in the domain of

computer security.
The present study employs the powerful N-gram algorithm for extracting low-
level features of the virus, which extracts N-byte combinations in sequence. This
algorithm involves four crucial steps, namely N-gram extraction, sequential pattern
extraction, pattern statistics calculation, and feature reduction, which together help
in obtaining more specific and accurate results related to malicious code. In addition,
the Random Forest algorithm, an ensemble learning method used for cataloging and
regression, is utilised for classifying the experimental dataset. This algorithm is based
on constructing multiple decision trees and using the majority vote to make deci-
sions, ensuring maximum accuracy. Our research is organised into several sections,
including an introduction, material and methodology, performance, experimental
results and discussions, and conclusion.
To successfully implement BLS initiatives, policy areas related to intellectual
property ownership, copyright, and workload allocation for each member of the insti-
tution need to be considered. Such initiatives often require the adjustment of current
organisational procedures or employee agreements to meet the requirements arising
from the adoption of the BLS system. The participation of institutional members is
crucial, and identifying success factors can facilitate their commitment and reduce
their fears in the long term [2].

2 Research Methodology

2.1 Blended Learning to Be Used as Mediated Platform

In essence, knowledge can be considered as acquired externally, as a cognitive state

resulting from a thought process, or as a meaning constructed through social interac-
tion [5]. A study found that students who participate more frequently in forums and
achieve higher levels of academic success tend to express more positive views about
BL, highlighting the importance of interaction and communication in the effective-
ness of this approach [6]. When developing a BL program, it’s important for teachers
and students to be aware that the outcomes should align with organisational expec-
tations, and that a balance between motivation and content is necessary for this new
learning endeavour to be collaborative [7].
However, individual antecedents such as self-attitude, self-behavior, self-
cognitive, and intention to comply appear to be less significant than organisational
antecedents like policy compliance, training programs, and security threats, as well
as environmental antecedents such as peer performance, social pressure, and reli-
gious indicators in encouraging students to recognise their roles and responsibilities
in the higher education system [8]. The adoption of BLS programs in numerous
Bridging the Gap Between Learning and Security: An Investigation … 423

Table 1 Hexa view in ransomware files

Hexa view in ransomware Files
0045C180 02 OO 8B CO OO 8D 40 OO 74 2040 OO FC 21 400
0045C190 70 2540 OO 32 IF 8B CO 32 13 8B CO 52 75 6E74
0045CIAO 6961) 65,206,572 72 6F72 202,020,202,061 74
0045C190 70 2540 OO 32 IF 8B CO 32 13 8B CO 52 75 6E74

organisations has been growing steadily, with annual increases of at least single-
digit percentages. To initiate a BLS program, organisations must establish a frame-
work for centralised and continuous data collection to monitor and evaluate the
program’s effectiveness. To ensure the success and efficiency of BLS implementa-
tion, several factors should be considered. To implement blended learning effectively,
it is necessary to select appropriate technology tools that are tailored to each student’s
abilities, design an instructional program that captures comprehensive student data,
create a curriculum that guides the instructor through the next stages of instruction,
and choose a platform that provides sufficient resources and materials to support
organised teaching and learning. In evaluating the quality of technology for blended
learning, learner interactions were considered from the perspectives of cognitivism,
collaborative learning, and student–teacher interactions. “The quality of technology
was assessed based on its availability (72%), content (71%), intervention (69%), chat
features (69%), resources (68%), Internet reliability (66%), and email exchanges
(63.4%) [9]”.
Various factors, such as the “characteristics of the student population”, “the
mission of the organisation”, “the strategic planning processes”, “faculty respon-
siveness”, “student acceptance”, “community values”, “available resources”, “organ-
isation support mechanisms”, and other components, have helped shape blended
learning in a way that makes sense for a particular organisational context [10]
(Table 1).

2.2 Dataset Representation

The purpose of this study involves the creation of a dataset for experimental purposes.
The dataset consists of normal files extracted from the Windows 10 operating
system’s executable files, while the malware files (specifically, Trojan Ransomware)
are sourced from the VX Heaven Virus Dataset. Portable Executable (PE) formats
are utilised to determine how programs are executed within the Windows operating
system. The dataset files are presented in both Hexadecimal and Assembly views.
Furthermore, normal files are graphically represented as flow graphs to enhance their
comprehension, as exemplified by the dfrgui.exe program in Windows.
The Trojan Ransomware Malware files, on the other hand, are extracted from the
VX Heaven Virus Dataset [6]. Prior to cataloguing the dataset, the assembly language
424 V. Pandey et al.

Fig. 1 Energy transformation

codes in both hexadecimal and flow graph formats are analysed for disassembled
files. In malware files, Portable Executable (PE) formats such as.text and CODE are
commonly found. Figure 1 illustrates a malware file in the Hexa view in IDA Pro
[7]. Prior to cataloguing, it is important to disassemble both the malware and regular
files. The process of disassembly entails the conversion of machine code contained
within files of the.exe type into assembly language. This conversion necessitates the
utilisation of a disassembler tool. This study used Interactive Disassembler (IDA)
Pro to perform the disassembly of both malware and regular files, resulting in the
generation of output files stored in the.asm file format.

2.3 Blended Learning for Energy and Industry Applications

System

Blended learning may be conceptualised as an information system, and its efficacy

can be assessed by examining the Energy and Industry Applications utilised by
its users. The assessment of user Energy and Industry Applications holds signif-
icance in ascertaining the efficacy of expenditures made in information systems.
Throughout history, the utilisation of Energy and Industry Applications by users has
been employed as a metric to gauge the effectiveness and accomplishment of infor-
mation systems. The data collection process involved gathering user input about
the distinct attributes of the information system. Additionally, an assessment was
conducted to analyse several elements that impact user Energy and Industry Appli-
cations, including their views of the information system’s attributes and their indi-
vidual requirements. The user’s text does not contain any information to rewrite in
an academic manner.
Information systems (IS) play a crucial role in the administrative and academic
operations of higher learning institutions in developing nations. In order to achieve
favourable outcomes in Information Systems (IS) initiatives, it is crucial to take into
account the contextual and institutional characteristics that are unique to the project at
hand (12). The assessment of project performance relies heavily on the involvement
and contributions of stakeholders in the Energy and Industry Applications domain.
Bridging the Gap Between Learning and Security: An Investigation … 425

This is because stakeholders play a crucial role in shaping the project’s environment
and contextual factors. The issue of energy transformation is depicted in Fig. 1.
Individuals tend to evaluate energy and industry applications based on their prior
experiences with brands, movies, jobs, political parties, and other factors. These eval-
uations are often gathered through objective data and have been extensively used by
psychologists, policy scientists, and consumer researchers to predict repeat purchases
or election outcomes, and to assess happiness or self-well-being among different
population groups [13]. To understand user perceptions, researchers typically rely
on the difference between expectations and performance evaluations. Therefore, it
is important for researchers to obtain detailed information about customer service
expectations and perceived performance. Thus, performance benchmarks for some
periods are considered equivalent to the norm for controlling goods or services.
Nevertheless, when the deviation from this criterion is sufficiently large, that is, when
the perceived performance falls outside the acceptable range, the brand’s performance
will be deemed different from the norm, leading to significant dissatisfaction with
product evaluation [14].
Extensive government support and guidance are crucial for the success of blended
learning systems (BLS) in higher education. To evaluate the effectiveness of BLS,
student and lecturer Energy and Industry Applications must be considered along with
learning outcomes. Although most users have positive perceptions of BLS, support
is still necessary to cater to different learning styles and cultures. Student adoption of
energy and industry applications in blended learning systems is influenced by various
factors, such as adaptability to e-learning, perceived benefits, teacher response, ease
of use, and application usage. Other important determinants of student learning in
blended learning systems include computer self-efficacy, performance expectations,
system functions, content characteristics, interaction style, and learning climate. As
students invest a lot of time, money, and effort in education, their Energy and Industry
Applications are critical for motivation and success. Meeting or exceeding expecta-
tions not only satisfies students but also encourages them to become advocates for
the organisation, enhancing its reputation within the community [15–18].

2.4 Feature Extraction

The present study utilised a different approach compared to previous research [8]
where manual copying of data into the.txt format was conducted along with manual
separation [9]. For this research, disassembled files from IDA Pro were saved in
the.asm file format and the Python programming language was used to select the
opcode strings. This approach only takes instructions as input and doesn’t consider
other details like memory addresses and values. During the feature extraction process,
instructions found in parameters are printed and the process stops if no instructions
are detected. The study used the N-grams algorithm with a range value of N = 1 to
N = 4.
426 V. Pandey et al.

Result--->Feature
Extraction

Instruction Split

N Initialization

N-Gram
Process

Fig. 2 N-gram process

Bootstrap=True Criterion= Gini P=X Feature

Confusion
Training Testing Matrix
OOB Score
Process

Fig. 3 Random forest cataloging process

During this stage, the opcodes were sorted based on their frequency of occurrence.
Python programming language was used for the extraction process using N-grams
[12]. A flowchart illustrating the N-gram programming process is presented in Figs. 2,
3.

2.5 Random Forest Classifier

The current research utilises the Random Forest (RF) algorithm [16] to provide cata-
loguing outcomes using a voting mechanism, wherein the decision tree with the
highest number of votes is selected. The level of accuracy seen in the cataloguing
results serves as an indicator of the algorithm’s performance. The RF method is
designed with the specific goal of attaining an accuracy rate of 98%. This is accom-
plished by utilising a total of 500 trees, denoted as n tree, which has been determined
to provide optimal outcomes [17].
Bridging the Gap Between Learning and Security: An Investigation … 427

This strategy aims to mitigate the issue of decision trees exhibiting overfitting
tendencies towards their training data, hence resulting in improved performance.
While decision trees are often outperformed by random forests, it is worth noting
that the accuracy of random forests is still inferior to that of gradient enhanced trees.
Nevertheless, the efficacy of the procedure is contingent upon the inherent attributes
of the data under scrutiny.

2.6 Performance Analysis

The evaluation of the suggested methodology is conducted by employing a confusion

matrix, a tool that offers insights into the accurate and inaccurate classification of
test data. This matrix facilitates the computation of the algorithm’s accuracy on the
testing dataset. The user has provided a numerical reference, indicating the presence
of a citation or source[20, 31].

3 Research Methodology

This study aimed to examine factors related to the learning environment in Informa-
tion Systems (IS), and their relationship with Overall Energy and Industry Applica-
tions (OS). The researchers utilised theories such as “assimilation theory” [19], “con-
trast theory” [20, 21], “prospect theory” [22], “the theory of adaptation level” [23],
and “generalised negativity theory” [24] to establish the relationship between OS
and computer self-efficacy, expectation of quality, information timeliness, perceived
utility, software adequacy, and user support [8]. A quantitative research design was
employed, and a survey questionnaire was distributed online to a selected sample of
100 students using stratified sampling who had completed the subject of information
system project management. The questionnaire, presented in Indonesian language,
used 5-point Likert scales for response options. SmartPLS 3 was employed as the
analytical tool to examine the fundamental relations and effects among variables,
which were visualised through model visualisation and path structure [25]. Prior to
data collection, experts evaluated the clarity and simplicity of the question statements,
and revised the word choices accordingly. The questionnaire comprised 45 questions,
categorised into seven variables, namely perceived utility (PU, 5 items), expectation
of quality (EQ, 7 items), information timeliness (IT, 6 items), user support (US, 7
items), software adequacy (SA, 8 items), computer self-efficacy (CS, 6 items), and
overall energy and industry applications (OS, 6 items). These are as follows (Table 2).
Factor analysis was utilised to determine the degree of correlation between an
indicator and other predictors in the model, which is a commonly employed method
for detecting linear or multiple relationships. To accurately evaluate the contribution
of predictors to the model, the variance inflation factor (VIF) was utilised. Higher
VIF values indicate greater difficulty in accurately assessing predictor contribution
428 V. Pandey et al.

Table 2 Variance Inflation Factor (VIF) for Indicators in LM-D

Indicators Statements VIF
PU3 The module uplifting intellectually 1.691
PU4 Academical Pressures as student 1.209
EQ1 Feedback to help on the progress 1.583
EQ2 High workload of module 1.276
EQ3 Clear ex expectation for the future goals 1.187
EQ4 Motivation to self-improve 2.261
IT1 Achieve best Performance 1.736
IT2 Critical thinking kind of anal sis 2.002
IT3 Memorisation-type of test 1.356
IT4 Commitment-based as group member 1.136
US1 Sufficiently for knowing the material 1.285
US2 Involvement of lecturers to reduce difficulty 1.739
US3 Good explanation from lecturers on module 1.584
US4 The alignment of theory and Practical 1.406
SA1 Student interest lies upon the lecturers’ skill 1.503
SA2 Academic goals determine the purpose 1.082
SA3 Cognitive skill is extremely critical 1.753
SA4 Lecturers should comment on the assignment 1.732
CS1 Complex in accessing the system 1.582
CS2 impression towards the features of the system 1.943
CS3 Self-study can be fulfilled with the video content 1.678
CS4 It is necessary to have reference link 1.617
PU3 Quiz is important to assess the understanding 2.448
PU4 The slide should be sophisticated 1.754

to the model. A VIF value of one implies that the predictor is not correlated with
other variables, while values above four or five are generally considered moderate
to high, with values exceeding 10 regarded as very high [27]. The majority of VIF
scores were found to exceed 1, with roughly seven indicators having a VIF exceeding
2, according to the table provided above (Table 2).

4 Results and Analysis

The results of experiments using N-gram implementation show varying frequencies

for different N values and operational code vocabularies. The length of the vocabulary
for the N-gram extraction feature will be compared graphically, as depicted in Fig. 4.
Bridging the Gap Between Learning and Security: An Investigation … 429

Fig. 4 N-gram extraction

This study utilises different N-gram feature extractions, including 1-g, 2-g, 3-g,
and 4-g. When N equals 1, the normal file has a length of 118, and malware has a
length of 129. For N equals 2, the normal file’s length is 1344, and malware has a
length of 1230. Similarly, when N equals 3, the length of the normal file is 5582, and
malware has a length of 5255. Finally, when N equals 4, the length of the normal file
is 14033, and malware has a length of 13,655. Increasing the value of N will increase
the vocabulary in N-grams, and more N-grams will result in more sequential patterns.
The experimental outcomes are presented in Table 3, illustrating the performance
of the confusion matrix across all conducted tests. In Experiment 1, the true positive
(TP) score was 259, the true negative (TN) was 237, there was one false positive
(FP), and three false negatives (FN). In Experiment 2, the true positive (TP) score
was recorded as 207, while the true negative (TN) count was 189. Additionally, there
was one false positive (FP) and three false negatives (FN) observed. In Experiment
3, the true positive (TP) score was 149, the true negative (TN) score was 146, there
was one false positive (FP), and four false negatives (FN). In Experiment 4, the true
positive (TP) score was recorded as 96, while the true negative (TN) score was 101. No
false positives (FP) were seen, resulting in a perfect score for FP. However, three false
negatives (FN) were identified in the experiment. In the conducted study, Experiment
1 exhibited the most elevated sensitivity score, reaching a notable percentage of
98.8%. Conversely, Experiment 4 had the best accuracy score, achieving a perfect
percentage of 100%. In Experiment 4, the false positive rate (FPR) achieved the
greatest score, reaching 0%. Conversely, Experiment 1 obtained the highest “accuracy
score” with a value of 99.2%. In summary, the results of experiment 1 demonstrated
superior performance in terms of sensitivity and accuracy, but experiment 4 exhibited
the highest scores in precision and false positive rate (FPR).
The function of “Energy and Industry Applications” (EIAs) is of great importance
in influencing users’ experiences. Positive disconfirmation takes place when products
or services surpass expectations, resulting in an improved experience for customers.
430 V. Pandey et al.

Table 3 Evaluation of experimental scenarios

Metric Experiment 1 Experiment 2 Experiment 3 Experiment 4
True positive 259 207 149 96
True negative 237 189 146 101
False positive 1 1 1 0
False negative 3 3 4 3
Sensitivity 98.80% 98.50% 97.30% 96.90%
Precision 99.60% 99.50% 99.30% 100%
False positive Rate 0.42% 0.50% 0.60% 0%
Accuracy 99.20% 99% 98.30% 98.50%
OOB error 0.01% 0.01% 0.02% 0.02%

The underlying principle of this concept is based on the idea that consumers’ percep-
tions of a product or service’s performance are in accordance with their preferences
and requirements [16]. The estimation of reliability is often conducted using measures
such as “Cronbach’s alpha” and “Tarkkonen’s rho”. In general, a reliability coeffi-
cient of 0.7 or above is often regarded as satisfactory, whereas values below 0.5
are considered unsuitable. Nevertheless, the present investigation yielded unsatis-
factory reliability estimates for two variables, namely PU and US, as indicated by
their respective “Cronbach’s α” values of 0.305 and 0.598. In contrast, the estimated
values of Rho_A for PU and US were 0.773 and 0.669, respectively. These findings
show that other estimators, such as “Rho_A” based on factor analysis, may be more
suitable since they do not underestimate reliability and might potentially offer more
precise estimates [28].
In addition, it is crucial to consider the measures of composite reliability (CR) and
average variance extracted (AVE) in order to ascertain the validity of the structural
model. Generally, it is recommended that AVE values should be at least 0.5 and CR
values should be at least 0.7 [29]. Nevertheless, it was observed in this study that none
of the variables satisfied the “AVE threshold” of 0.5, while the variables “PU”, “SA”,
and “US” did not match the “CR” criterion. Consequently, the structural model was
unable to demonstrate convergent validity, a critical aspect in assessing the degree
of correlation among various indicators of the same construct that exhibit agreement
[30]. Furthermore, a strong correlation was seen between several pairs of variables
inside the “Latent Variables (LVs)” framework. For instance, the variables “PU”
and “EQ” exhibited a correlation coefficient of 1.028, while the variables “US” and
“CS” showed a correlation coefficient of 1.041. There were eight pairings of variables
that exhibited scores exceeding one, suggesting the presence of model specification
mistakes, estimation difficulties, or suppressor effects. The fundamental characteris-
tics of the “LVs” are closely connected to the characteristics of the indicator variables
employed to define them. These “LVs” have the potential to have both direct and
indirect impacts, which can vary in magnitude from −1 to greater than 1 [31].
Bridging the Gap Between Learning and Security: An Investigation … 431

Based on the aforementioned results, it is crucial to take into account alternative

estimators, such as “Rho_A,” and to tackle concerns pertaining to “AVE” and “CR”
in order to verify the validity and dependability of the structural model. By attending
to these concerns, researchers will be able to enhance the precision of measuring
the degree of association among various indicators pertaining to a certain construct,
hence enhancing the overall validity and reliability of the study’s outcomes (Tables 4,
5).
The present study employs a thorough and complete method to data analysis.
The researchers have employed factor analysis and route analysis techniques on the
structural model, doing one thousand iterations and setting seven as the stopping
criteria. This rigorous approach has been undertaken to assure the stability and reli-
ability of the scores of the latent variables. Moreover, the use of the Partial Least
Squares Structural Equation Modelling (PLS-SEM) technique is notably noteworthy
due to its ability to elucidate the highest degree of association between variables and
provide more consistent latent variable scores throughout the algorithmic process.
Moreover, the researchers have exhibited meticulousness in their investigation, as
they have thoroughly examined the loadings of the indicators. Although the majority
of external loadings failed to exceed the acceptable threshold of 0.7, a few indicators,
including “CS3”, “EQ7”, “IT4”, “OS1”, “OS5”, “OS6”, “PU3”, “SA5”, and “SA7”,

Table 4 Validity & Reliability evaluation

Cronbach’s Rho A Composite Averange
Alpha Reliability Variance
Computer Self-Efficacy (CS) 0,797 0.82 0.807 0.416
Expectation of Quality (EQ) 0.807 0.828 0.809 0.389
Information Timeliness (IT) 0.742 0.773 0.752 0.346
Overall (os) Satisfaction 85.10% 85.60% 84.80% 0.485
Perceived Utility (PU) 30.50% 77.30% 41% 0.359
Software (SA) Adequacy 70.50% 81.80% 63% 0.268
User Support (US) 60% 66.90% 54.60% 0.194

Table 5 Correlation among latent variables

CS IT PU US EQ OS SA
US 1.04 1.21 0.847 1 1.047 0.751 0.963
SA 0.86 0.95 0.782 0.963 0.87 0.741 1
PU 0.69 0.92 1 0.847 1.028 0.673 0.782
OS 0.84 0.82 0.673 0.751 0.645 1 0.741
IT 0.95 1 0.922 1.21 0.99 0.827 0.952
EQ 0.81 0.99 1.028 1.047 1 0.645 0.87
CS 1 0.95 0.691 1.041 0.815 0.842 0.86
432 V. Pandey et al.

exhibited outstanding results. Furthermore, the researchers have taken measures to

assure the inclusion of just the most reliable indicators in their study by excluding
those with borderline values below 0.5.
Furthermore, the researchers have also taken into account the importance of each
indicator, evaluating their values using the process of bootstrapping. Although the
researchers have advised against relying solely on statistical significance, it is worth
noting that just eight indicators had values below 1.96. The authors appropriately
acknowledge that the prevalent utilisation of statistical significance as a means to
assert scientific findings might result in substantial distortions within the scientific
process. The methods and analysis employed in this work exhibit a meticulous and
comprehensive approach to the data, hence assuring the reliability and robustness of
the findings (Table 6).
The objective of the discriminant validity assessment is to ensure that a reflec-
tive construct exhibits stronger relationships with its indicators than with any other
construct in the PLS path model [34]. Consequently, some researchers propose the
use of “Heterotrait-monotrait” ratio (HTMT) of the correlation, in addition to the
“Fornell-Larcker criterion”, due to the latter’s lack of reliability in detecting possible
indiscriminate among latent variables [35]. However, the exact threshold level for
HTMT remains a matter of debate, with some researchers recommending a value of
0.85, while others suggest 0.9 [36, 37]. The table indicates eight variable relation-
ships that require consideration due to HTMT values below the threshold, including
“CS-PU (0.68), CS-SA (0.76), EQ-OS (0.63), EQ-SA (0.75), OS-PU (0.68), OS-SA
(0.61), and” OS-US (0.67).”
The term “saturated model” refers to a structural model where all constructs
are freely correlated, with a standardised root mean square residual (SRMS) of no
more than 0.08 or 0.10 and a normalised fit index (NFI) greater than 0.90. On the
other hand, the estimated model is based on the model structure and considers the
total effect scheme. The values of d_ULS and d_G must be less than the 95% or
99% bootstrap quantile to assess the model fit. The exact model fit test is based on
the discrepancy between the empirical covariance matrix implied by the composite
factor model, and only the bootstrap results of the exact model fit measure allow for
interpretation of results. If the correlation matrix is small enough to be attributed to

Table 6 Validity analysis of HTMT discriminant

CS EQ IT OS PU SA US
CS
EQ 0.81
IT 99 0.97
OS 0.85 0.63 0.84
PU 0.68 1 0.88 0.68
SA 0.76 0.75 0.84 0.61 0.72
US 0.95 0.94 1.12 0.67 0.82 0.97
Bridging the Gap Between Learning and Security: An Investigation … 433

Table 7 Correction value analysis

“ Saturated Estimated Correction
Model Model Value (ci
SRMR 0.10 2.29 CS 2.78
d_ULS 11.26 5,423.20 EQ 2.95
d_G1 6.12 21.41 IT 2.36
d_G2 5.84 19.09 OS 3.12
Chi-Square 2.230.84 n/a PU 2.02
NFI 0.41 n/a SA 2.38
US 1.66”

sampling error, or the discrepancy is not significant (p > 0.05), then the model fit is
considered established [25, 35] (Table 7).
The study employed the PLS path modelling technique to investigate the rela-
tionships among various constructs. The findings suggest that computer self-efficacy
has a negative direct impact on “software adequacy (β = −0.580)” but a positive
direct effect on “expectation quality (β = 3.320)”, which is more significant than the
effect of overall satisfaction with “Energy and Industry Applications (β = 2.294)”.
Additionally, information timeliness was discovered to have a negative direct effect
on “perceived utility (β = −3.052)”, which is more substantial than the effect of
overall satisfaction with “Energy and Industry Applications (β = −2.314)”, while
also having a positive direct impact on “software adequacy (β = 1.503)”, which is
more significant than the impact of “user support (β = 1.210)”. Furthermore, user
support had a negative direct impact on overall satisfaction with “Energy and Industry
Applications (β = −0.740)” and “expectation quality (β = −2.418)”.
It is crucial to acknowledge that these results were achieved through the use of the
Partial Least Squares (PLS) route model and are predicated on the beta coefficients
acquired from the investigation. The presence of a negative sign in the beta coef-
ficients signifies an inverse association between the constructs, whereas a positive
sign suggests a positive association. The study also analysed the effect sizes of the
associations, where the magnitude of the beta coefficients was used to determine
the strength of the links between the constructs. The findings of this study have
significant significance for the comprehension of the interconnections between the
components and their impact on the overall satisfaction with “Energy and Industry
Applications”.

5 Conclusion

In order to improve the structural model designed to assess user satisfaction with
Energy and Industry Applications in blended learning, it is necessary to take into
account many factors. These factors encompass the choice of indicators, sampling
434 V. Pandey et al.

methods, the reflective model, and the desire of respondents to take part in the study.
The rigorous evaluation of these areas is crucial in order to ensure the robustness
of the final model and its correct representation of the aspects that lead to user
satisfaction with Energy and Industry Applications.
The findings of the study indicate that the timeliness of information has a detri-
mental influence on the level of satisfaction in Energy and Industry Applications.
Conversely, user assistance, perceived usefulness, software adequacy, computer self-
efficacy, and expectation quality positively contribute to the overall satisfaction. The
findings of this study highlight the significance of giving priority to these factors
during the creation of blended learning programmes, with the aim of improving user
satisfaction in the context of Energy and Industry Applications.
The selection of the value of N is of utmost importance when utilising N-
gram extraction in a Random Forest classifier, as it has the potential to signifi-
cantly influence the quantity of vocabulary included by the N-grams and sequen-
tial patterns generated. In trial 1, the Random Forest classifier had the highest level
of accuracy in cataloguing, reaching 99.2%. Additionally, the sensitivity value of
98.8% suggests that the algorithm effectively categorised the majority of malware
samples. Furthermore, the false positive rate (FPR), which represents the proportion
of malware samples that were incorrectly categorised as normal, was found to be
0%. Additionally, the precision, which indicates the percentage of accurately iden-
tified malware samples, was determined to be 100%. The out-of-bag (OOB) error
exhibited minimum values, as seen by experiment 1’s low error rate of 0.008. Future
study should aim to conduct experiments using the latest and comprehensive malware
dataset in order to enhance the performance of the model. The findings underscore the
capacity of the Random Forest classifier to effectively categorise malware samples,
hence opening avenues for further investigations in this domain.

References

1. Vyas R, Luo X, Mcfarland N, Justice C (2017) Investigation of Malicious Portable Executable

File Detection on the Network using the Supervised Learning Techniques Symposium on
Integrated Network and Service Management (IM). Thing. 941–946. 2017
2. Oliver M, Trigwell K (2005) Can ‘blended learning’ be redeemed? ELearning 2(2):17–26
3. Huda MQ, Hussin H (2017) Evaluation model of Information technology innovation effective-
ness case of higher education institutions in Indonesia. Proc ICIC
4. Yewale A (2016) Malware Detection Based on Opcode Frequency. International Conference
on Advanced Communication Control and Computing Technologies. 978:646–649
5. Al-rimy MAM, Shaid SZM (2018) Ransomware threat success factors, taxonomy, and
countermeasures: A survey and research directions. Computer Security 74:144–166
6. Raff E et al (2016) An Investigation of Byte N-Gram Features For Malware Classification. A
Comput Virol Hack Tech. 14(1):1–20
7. Rogers E (2003) Diffusion of innovation. 5th edition. An integrated approach to communication
theory and research. Simon and Schuster. ISBN 9780743258234
8. Vaezi R (2013) User Satisfaction with Information Systems: A Comprehensive Model of
Attribute Satisfaction. Ph.D. Dissertation, University of Houston
Bridging the Gap Between Learning and Security: An Investigation … 435

9. Adam F et al (2014) The use of blog as a medium of Islamic da’wah in Malaysia. Int. Journal
of Sustainable Human Development 2(2):74–80
10. Meyer S et al (2014) Rhetoric and reality: Critical perspective on educational technology. In:
Proceedings Ascilite Dunedin, 89–98
11. Alonso F et al (2005) An instructional model for web-based e-learning education with a blended
learning process approach. British J of Educational Technology 36(2):217–235
12. Akkoyunlu B, Soylu MY (2006) A study on students’ views on blended learning environment.
Turkish Online Journal of Distance Education 7(3/3):43–56
13. Lalima KL, Dangwal (2017) Blended learning: an innovative approach. Universal Journal of
Educational Research 5(1): 129–136, HRPUB
14. Ghahari S, Ameri-Golestan A (2013) The effect of blended learning vs. classroom learning
techniques on Iranian EFL learners’ writing. Int. J. of Foreign Language Teaching and Research
1(3)
15. Breiman L (2001) Random Forest. Kluwer Academic Publishers. 45:5–32
16. Heaven V (n.d.) Computer Virus Collection. Retrieved from http://vxheaven.org/vl.php
17. IDA Support: DownloadCenter.(n.d.). Retrievedfrom https://www.hexrays.com/products/ida/
support/download.shtml
18. Richardson R, North M (2017) Ransomware: Evolution, mitigation and prevention. Interna-
tional Management Review. 1, 13, pp. 10–21
19. Kinder H, Veith Jakstab (2008) A static analysis platform for binaries. Springer-Verlag Berlin
Heidelberg. 5123, pp 423–427.
20. Kassim ES et al (2012) Information System Acceptance and User Satisfaction: The Mediating
Role of Trust. Procedia Social and Behavioral Sciences 57:412–418
21. Subiyakto A et al. (2016) The user satisfaction perspective of the information system projects.
Indonesian Journal of Electrical Engineering and Computer Science, 4(1), pp 215–223
22. Levy-Garboua, Montmarquette C (2007) A theory of satisfaction and utility with empirical and
experimental evidence. In: Conference of the French Economic Association, Behavioral and
Experimental Economics
23. Yuksel A, Yuksel F (2008) Consumer satisfaction theories: a critical review. In Book: Tourist
Satisfaction and Complaining Behavior, Publisher: Nova Sciencce
24. Tahar NF et al. (2013) Students’ satisfaction on blended learning: the use of factor analysis.
In: IEEE Conference on e-Learning, e-Management and e-Services
25. Naaj MA et al (2012) Evaluating Student Satisfaction with Blended Learning in a Gender-
Segregated Environment. Journal of Information Technology Education: Research 11:185–200
26. Knox WE et al (1993) Does college make a difference? Long-term changes in activities and
attitudes. Greenwood Press, Westport, CT
27. Chute AG et al (1999) The McGraw-Hill handbook of distance learning. McGraw-Hill, New
York
28. Xie Y, Greenman E (2005) Segmented assimilation theory: a reformulation and empirical test.
Population studies center research report 05–581
29. Hovland C et al. Assimilation and constrast effects in reaction to communication and attitude
change. Journal of Abnormal and Social Psychology, 55(7), pp 244–252
30. Ko D, Afternoon K (2018) Comparative Study on Perturbation Techniques in Privacy Preserving
Data Mining. International Journal of Innovative Computing 8(1):27–32
31. Canfora G, De Lorenzo A (2015) Effectiveness of opcode n grams for detection of multi family
android malware. In: International Conference on Availability Reability and Security. 333–340
32. Sornil O, Liangboonprakong C (2013) Malware classification using N-grams sequential
pattern features. In: International Journal of Information Processing and Management (IJIPM).
4(7):59–67
33. Bilar D (2007) Opcodes as Predictor for Malware. Electronic Security and Digital Forensics
1(2):156–168
436 V. Pandey et al.

34. Belgiu L, Dra (2016) Random forest in remote sensing: a review of applications and future
directions. ISPRS Journal. 114, pp 24–31
35. Lin, Chihta, Wang, Naijian (2015) Feature selection and extraction for malware classification.
Journal of Information Science and Engineering. 992. pp 965–992
36. Dijkstra TK, Henseler J (2015) Consistent and asymptotically normal PLS estimators for linear
structural equations. Comput Stat Data Anal 81:10–23
Ethical Considerations in AI-Based
Cybersecurity

Keshav Kaushik, Aadil Khan, Ankita Kumari, Ishu Sharma,

and Rajat Dubey

Abstract The ethical implications of incorporating artificial intelligence (AI) into

cybersecurity procedures are examined severely in this study article. The study
highlights the significance of addressing ethical questions to guarantee respon-
sible and equitable use of this technology, as artificial intelligence (AI) plays an
increasingly vital role in identifying, preventing, and mitigating cyber risks. The
research work highlights the important applications of AI in cybersecurity, stressing
its capacity to detect abnormalities, evaluate large datasets, and adapt to changing
attack tactics. But as AI-driven cybersecurity solutions become more common,
privacy-related ethical issues become more important. The study investigates the
possible privacy infringements brought about by the comprehensive data collection
and surveillance powers built into AI-powered security systems. It highlights how
important it is to use AI to identify threats while also protecting customers’ personal
information. The study also looks at moral concerns about the use of AI to cyber-
security globally. It addresses how AI systems in one country may have an impact
on systems in other countries, emphasizing the need of maintaining international
harmony, upholding standards, and averting damage or conflict in AI-related situ-
ations. In order to address global concerns, the report emphasizes the significance
of ethical protections, widespread access to technology, and adherence to privacy
and data protection legislation. The research work concludes by recommending
a thorough strategy that gives privacy protection, bias mitigation, openness, and

K. Kaushik
School of Computer Science, University of Petroleum and Energy Studies, Dehradun, India
A. Khan · A. Kumari · I. Sharma (B)
Chitkara University Institute of Engineering and Technology, Chitkara University, Rajpura,
Punjab, India
e-mail: [email protected]
A. Khan
e-mail: [email protected]
A. Kumari
e-mail: [email protected]
R. Dubey
Cybersecurity Expert, Allianz Commercial, Austin, USA

© The Author(s), under exclusive license to Springer Nature Singapore Pte Ltd. 2024 437
K. Kaushik and I. Sharma (eds.), Next-Generation Cybersecurity, Blockchain
Technologies, https://doi.org/10.1007/978-981-97-1249-6_19
438 K. Kaushik et al.

accountability top priority when incorporating AI into cybersecurity. The develop-

ment and use of AI-powered cybersecurity solutions must be guided by continuing
ethical concerns in order to ensure that these technologies uphold moral standards
and enhance global digital security.

Keywords Ethics · Cybersecurity · Technology · Network · Artificial

Intelligence · Blockchain · Privacy

1 Introduction

In order to safeguard digital environments from constantly evolving cyberthreats, a

paradigm change is required. Artificial intelligence (AI)-enhanced cybersecurity is
one indication of this change. Businesses may strengthen their defenses by identi-
fying and evaluating threats more intelligently by using increasingly sophisticated
AI technologies in their operations. These techniques scan network traffic and user
behavior for changes that can indicate security flaws fast. Artificial intelligence (AI)
powers both anomaly detection and behavioral analysis. Experts in cybersecurity can
anticipate and respond to emerging threats by using machine learning models for
predictive analysis. Artificially intelligent response techniques may expedite crisis
management, reduce damage, and swiftly isolate malfunctioning systems. Artificial
intelligence (AI) is utilized to identify people and control access, while biometric
and behavioral detection are used for security [1]. More instances of the several
ways AI has enhanced cybersecurity include big data analytics, automated risk
assessment, and fraud detection. Artificial Intelligence is a dynamic security ally
that adapts to new threats by constantly learning and evolving. This provides enter-
prises with an adaptable means of defending themselves against more sophisticated
cyber-attacks. When integrated properly, AI and cybersecurity strengthen digital
systems and increase their resistance to various cyberthreats. AI improves safety by
streamlining the process of identifying, addressing, and neutralizing hazards [2].
AI plays a critical role in improving defense. Making threats easy to identify is
one of the finest things AI has accomplished. AI systems search for odd patterns in
network traffic and user behavior using enormous datasets and complex algorithms.
These instruments have access to a wealth of data. This gives us a significant edge in
identifying and mitigating emerging risks before they worsen by making it feasible
to identify potential security problems early on. With the use of machine learning
models and prediction analysis, cybersecurity professionals are now able to anticipate
emerging threats and take immediate action against them [3]. AI-driven models
provide a proactive protection that can keep up with a dynamic risk environment by
staying one step ahead. This is accomplished by identifying common patterns among
various cyberattack types.
Ethical Considerations in AI-Based Cybersecurity 439

1.1 How Effective Is AI for Cybersecurity?

Artificial Intelligence (AI) has become a new and potent force in cybersecurity. With
AI, it can detect, prevent, and address an increasing array of cyberthreats. Huge
volumes of data can be handled by it, and it can recognize patterns and quickly
adjust to evolving attack strategies. The strong integration of AI and cybersecurity
has ushered in a new age of defenses that go beyond conventional techniques. As a
consequence, digital settings are now much more reliable.
Artificial intelligence systems are capable of analyzing vast volumes of data,
including network traffic, user activity, and system motions. Because of this, errors
and departures from the established guidelines might be identified as potential indi-
cators of security breaches. Artificial intelligence (AI) that can detect anomalies is
crucial because it may identify complex threats that may evade detection by stan-
dard security protocols [4]. Here’s where AI really excels. Machine learning models
enable prediction analysis, which allows artificial intelligence to identify and fore-
cast patterns associated with various cyber threats. With this proactive strategy, busi-
nesses can remain ahead of emerging threats, which is crucial in the cutthroat and
dynamic industry of cybersecurity. Rather than just reacting to attacks that have
already occurred, cybersecurity professionals can now address possible vulnerabili-
ties and new attack vectors before they arise thanks to AI-driven prediction models.
This approach is much superior than the previous one. Systems that manage user
authentication and access are greatly benefited by artificial intelligence. An efficient
and secure method of confirming that users are who they claim to be is biometric
authentication, which is made feasible by AI-powered recognition systems.
Behavioral biometrics is another AI-powered aspect of security. In order to iden-
tify anomalies, this kind of biometrics examines user behavior patterns. It adds
another degree of protection to access control systems. The use of AI to automatically
scan computer networks and systems is significantly altering the way vulnerabilities
are handled. AI-powered technologies enable enterprises to patch and secure their
systems before malicious actors take advantage of them by identifying and analyzing
vulnerabilities before they are exploited. This is made feasible by the ability of AI-
run tools to identify and assess flaws. This strategic strategy is crucial to remember
in order to reduce the attacks area and strengthen barriers. Scam detection becomes
simpler with AI, which is problematic for defense. Scams and malicious files in
email communications may be identified by AI systems using intricate formulae.
This increases the likelihood of social engineering attacks and provides a crucial
line of defense against a prevalent cyberthreat. AI-powered security analytics trans-
forms large-scale security data analysis. This capability enables proactive protection
strategies by enabling security specialists to go through large data for threats and
take appropriate action. The degree of AI’s flexibility determines how successfully
it defends against cyber dangers. Because AI systems can adapt to changing threat
environments and learn from fresh data, they are effective in combating online threats
[5].
440 K. Kaushik et al.

1.2 What Are Ethical Considerations in Cybersecurity

with AI?

Artificial intelligence (AI) and privacy raise many significant ethical issues. These
queries highlight the significance of designing and using these technologies with
transparent, ethical practices. The fact that protection measures powered by AI are
able to gather and monitor vast volumes of data is one of the main concerns. The possi-
bility of privacy violations resulting from these abilities is one of the key concerns.
In order to promptly identify ethical issues, we must simultaneously safeguard the
privacy of consumers’ personal information. Algorithms never favor or discriminate
against anybody, which is another significant social issue. Biases present in training
data may be reinforced by artificial intelligence systems. Results may become skewed
as a consequence. When it comes to cybersecurity, threat detection might be biased,
which can result in incorrectly identifying some people or unfairly characterizing
others [6]. Ensuring that AI protection solutions are created and utilized fairly is
crucial to upholding ethical norms and reducing the possibility of unjust impacts.
The clarity and rationality of AI algorithms is another societal problem to consider.
Problems arise when the decision-making process of AI systems is unclear. Under-
standing the risk assessment process of AI systems is crucial for user confidence
and safety obligations. Clear system choices are necessary for AI to be employed in
an ethical manner. When errors occur or the system is unable to thwart an attack,
AI-driven cybersecurity makes it difficult to assume accountability and responsi-
bility. Ethics requires accountability from individuals as well as solutions for errors
or unforeseen consequences that may arise from the use of these instruments. Secu-
rity hazards, such as the possibility of being attacked by an adversary, add layers
of ethical complexity. Hostile attacks occur when artificial intelligence (AI) systems
are programmed to behave in unexpected ways.
AI-based protection solutions must be reinforced against these types of threats
due to ethical concerns, preventing their unethical usage. Unethical issues include
AI-driven hacking and unequal access to resources. Increased resource inequality
favoring wealthy companies might result from advanced AI. Ethical concerns need
equitable safeguards and universal access to technology. Online safety and the ethics
of AI are also influenced by global events. AI systems in one nation may have an
impact on systems in other nations because of the global nature of safety. International
cohesion, norms, and preventing harm or conflict in AI-related hacking are among
the ethical issues. Observing the law and the principles of ethics are necessary. As
long as AI technologies adhere to privacy and data protection regulations, they are
employed effectively [7].
Ethical Considerations in AI-Based Cybersecurity 441

1.3 Why It Is Necessary?

It’s critical to consider ethical issues when combining artificial intelligence (AI)
with defense. This will ensure that these technologies are created and used respon-
sibly. The importance of striking the correct balance between identifying hazards and
protecting people’s right to privacy is shown by conflicts involving privacy. Algo-
rithmic bias and fairness are critical to preventing unjust outcomes from surveil-
lance procedures, promoting inclusiveness, and guaranteeing that all individuals are
safeguarded equally [8]. People are able to comprehend how artificial intelligence
systems determine which threats are there as long as there is transparency and explain-
ability. For user responsibility and trust, this is crucial. In order to correct errors and
unanticipated consequences and foster a culture of dependability and accountability,
it is essential to establish procedures for accountability and responsibility in AI-
driven protection. When safety conditions change, having humans oversee critical
choices ensures their morality and prevents users from becoming too dependent on
self-driving cars. Individual oversight ensures that significant decisions adhere to
the law [9]. In order to preserve their integrity, AI-powered security technologies
must be shielded from potential threats such as hostile cyberattacks. It’s critical that
everyone be involved, treated fairly, and that nations collaborate while using AI due
to ethical concerns regarding resource allocation and its consequences on the global
community. For AI systems to operate legally and to uphold moral standards, laws
and regulations must be obeyed. In order to avoid moral dilemmas and make respon-
sible AI decisions, it is crucial to consider the long-term repercussions on society.
Fundamentally, ethical concerns serve as a guide to ensure that the development and
use of AI in defense aligns with the principles of accountability, transparency, equity,
and public health.

2 How to Use AI for Cybersecurity?

There are many useful tools that artificial intelligence (AI) can give in the area
of cybersecurity that should be used to improve digital protections. You can’t say
enough good things about AI’s contribution to threat detection. It carefully looks
through huge datasets to find trends and oddities that could mean a security breach.
This includes desktop security, where AI-powered technologies carefully watch
each device, finding dangerous actions right away and stopping them before they
can do any damage. Phishing is a common type of online attack that faces a big
problem from artificial intelligence (AI). Artificial intelligence (AI) can expertly
search communication channels for signs of fake efforts, keeping users from falling
for scams.
Artificial intelligence is powerful not only because it can react, but also because it
can stop bad things from happening [10]. AI programs can find normal behaviors in
users and systems and report them right away if they are different, which could mean
442 K. Kaushik et al.

there has been a security breach. Behavioral research lets you figure out what normal
user and system behavior looks like so you can report any problems. Additionally,
AI makes a big difference in incident reaction by automating tasks, which speeds up
the process of finding, limiting, and reducing the effects of security events. In vulner-
ability management, AI does regular checks on systems and networks to see if they
have any holes that could let hackers in. Artificial intelligence (AI)-powered tools are
very helpful for security analytics because they give more detailed information about
security data through trend analysis, pattern recognition, and prediction analytics.
Artificial intelligence (AI) helps to tighten access controls by making user identity
better. It does this by using cutting-edge methods like biometrics and behavioral
analysis [11]. One area where artificial intelligence is used is in security automation.
This makes routine tasks easier and gives cybersecurity experts more time to work on
more difficult issues. Furthermore, machine learning improves safety by allowing for
more advanced malware detection and the exact identification of new threats. Addi-
tionally, artificial intelligence is a useful tool for cybersecurity, but it may work better
when it is part of a full security plan along with other safety measures. When it comes
to safety, AI tools are great at finding threats, keeping endpoints safe, and stopping
scams. Their skills include behavioral analysis, incident response automation, and
risk management, all of which make digital security stronger. AI’s important part
in security analytics, improving user identification, and finding advanced malware
makes its place in complete cybersecurity strategies even stronger. The Components
of Artificial Intelligence for Cybersecurity are depicted in Fig. 1.

2.1 Machine Learning

Machine learning makes security better by letting computers look through huge
datasets, find security risks, and predict them in real time. Its ability to change
makes it better at finding anomalies by reducing false hits and raising total accuracy.
Machine learning improves our understanding of normal trends in user behavior data,
making it easier to spot changes that might mean there has been a security breach.
Additionally, machine learning methods improve in the area of malware identification
as risks change, making security measures more adaptable. When this technology
is used in risk assessment, it can adapt quickly to changing cyber settings [12].
This makes cybersecurity stronger against a wide range of threats that are always
changing.

2.2 Deep Learning

Deep learning makes safety much better by changing how threats are found and how
they are dealt with. Because their neural networks are so complex, deep learning
algorithms are very good at finding complicated patterns and outliers in big datasets.
Ethical Considerations in AI-Based Cybersecurity 443

Fig. 1 Components of artificial intelligence for cybersecurity

This makes it easier to find possible security holes before they happen. This lets
systems naturally deal with new dangers, lowering risks in real time. Deep learning,
in particular, makes intruder detection systems, behavioral analysis, and malware
identification better. This adds another layer of protection against complex cyberat-
tacks. Deep learning models can keep learning and getting better from their mistakes
[13]. This makes sure that the cybersecurity framework is dynamic and flexible,
which is very important for keeping private data and systems safe in a dangerous
environment that is always changing.

2.3 Classification

Classification algorithms make security better by accurately describing the difference

between normal and dangerous behavior. By looking at trends in huge databases,
these programs quickly find things that don’t make sense and could be signs of
security holes. By observing patterns, they help make attack detection systems better,
which makes protections against new cyber threats stronger. Also, to find phishing,
classification algorithms look at communication channels and tell the difference
between real and malicious data. This lowers the risk of scams. These algorithms are
444 K. Kaushik et al.

very important for user identification because they sort physical and behavioral traits
into groups, which makes access controls stronger. Basically, using classification
algorithms lets cybersecurity systems make decisions in real time based on trends
they’ve learned. This makes digital platforms more resistant to a wide range of
complex cyberattacks. Here are some categorization algorithms that are regularly
used in cybersecurity.
Random Forest algorithms make security better by making it easier to find threats
in a way that is both reliable and flexible. Random Forest is a machine learning
ensemble method that is very good at looking at big datasets, finding trends, and
telling the difference between behaviors that are beneficial and detrimental. This
method is very important in cybersecurity for finding and predicting potential threats
correctly, without giving false positives or rejections [14]. Random Forest’s strength
is that it can handle a lot of different and complicated data, which lets it find new and
complex computer risks. Its group nature, which includes many decision trees, makes
it less likely that the model will be overfit, which improves its total performance.
Random Forest algorithms can be used in cybersecurity systems to find and stop new
threats before they happen. This protects digital platforms against a wide range of
constantly changing cyber threats.
Support Vector Machine (SVM) methods are very useful for improving safety
because they are exceptionally effective at what they’re supposed to do. When it
comes to threat identification, SVMs are great at sorting data points, which lets you
find trends that are linked to adverse behavior. This lets network data be watched
in real time, so any cyberattacks can be found and stopped. SVMs are great at
finding anomalies, which are changes from how a system usually works. This is very
important for stopping threats early on. By arranging network events into groups,
SVMs also help breach detection systems find malicious activity. Because they can
handle complex, high-dimensional data, SVMs are good at detecting cyber risks that
are minor and change over time. This makes total cybersecurity protection stronger
with a proactive and adaptable approach.
Using the k-Nearest Neighbour’s (KNN) method in security makes finding threats
and responding to them much better. The machine learning method KNN is very good
at finding data points by how similar they are to examples that are already known.
When it comes to cybersecurity, KNN can look at trends of network behavior to
quickly find outliers and possible security holes [15]. It can find new data points by
comparing them to close neighbors, which helps with assessing risk in real time.
KNN helps breach detection systems find trends and actions that don’t seem right,
so they can move quickly and stop the attack. This algorithm is very versatile as well
as effective at dealing with changing cyber threats. It is an important tool for making
digital environments stronger, and it can be used in addition to normal security
measures as an extra line of resistance. KNN gives people who work in safety a
broader and more flexible way to deal with new dangers.
Ethical Considerations in AI-Based Cybersecurity 445

2.4 Multi-class Classification

Multi-class classification helps improve cybersecurity by separating different cyber

threats into separate classes. This makes it easier to find and stop threats. With this
method, AI models can tell the distinction between different types of threats, like
malware, scams, and hacks. This helps faster and more targeted defenses be used.
The detailed labeling also helps put threats in order of importance based on how bad
they are, which makes it easier to use resources to respond to incidents. Multi-class
classification enables defense systems to respond quickly to changing threat settings,
making it easier for them to find and stop new threats. This purposeful use of AI not
only makes digital settings more secure, but it also creates a proactive defense against
a wide range of cyber risks that are always changing.

2.5 Pattern Recognition

Pattern recognition is an important part of enhanced cybersecurity because it protects

against novel threats in a preventative and dynamic way. By using strong algorithms,
artificial intelligence systems can quickly find and study trends in very large datasets.
This lets us find things that don’t seem right, which could mean that there have been
security breaches. This ability also includes behavioral analysis, which lets computers
tell the difference between normal human behavior and strange behavior that could
be a sign of bad behavior. In the field of cybersecurity, pattern recognition also helps
find current malware signs quickly and guess how attacks might happen in the future
[16]. It helps security experts stay ahead of cyber risks which speeds up response
times and makes the system more resistant to a threat environment that is always
changing. Deep knowledge in pattern recognition works like a force booster, making
digital systems stronger and reducing risks.

2.6 Federated Learning

Federated Learning, which spreads out model training, looks like a game-changing
way to make safety better. With this new way, technology can learn from each
other without sharing raw data. Federated Learning can help organizations make
their danger identification models effective in the field of cybersecurity. By mixing
information from different sources without compromising the protection of indi-
vidual data, the system gets better at spotting new risks [17]. This autonomous
learning makes it less likely that centralized data will be stolen and also makes
sure that everyone works together to make the network safer. Federated Learning
makes protection measures stronger by letting people work together and protect
their privacy. This helps people respond quickly and effectively to new cyber threats.
446 K. Kaushik et al.

3 Real-World Cybersecurity with AI Application Areas

Artificial intelligence (AI) is used in many areas of real-world safety and is very
important. Through clever Intrusion Detection Systems (IDS), which use machine
learning to find strange trends in network behavior, AI changes the way threats
are found and analyzed. It includes computer security, using behavioral analysis
and anomaly recognition to quickly find and stop bad behavior. Using AI to power
automated incident reaction speeds up the process of reducing security incidents by
making routine jobs easier to do [18].
The use of natural language processing (NLP) to find scam emails and harmful
links is made easier by AI systems that look at the text and trends of communi-
cation. AI-driven automatic scanning improves the finding and ranking of system
flaws, making proactive risk reduction easier in vulnerability management. Through
behavioral biometrics, AI can also affect identity and access management. This adds
an extra layer of security by using user behavior to verify identities.
AI-driven solutions are very important for improving and changing cybersecurity
measures in areas like threat intelligence, user behavior analytics, mobile security,
and cloud security. All together, these apps give companies the tools they need to deal
with cyber risks that are always changing, quickly and smartly [19]. This makes sure
that their digital defenses are strong even when they are up against smart attackers.
Table 1 summarizes the comparative analysis of available cybersecurity tools.

3.1 Datasets for Research Work

When doing research in the field of cybersecurity, having access to a variety of

datasets of a high quality is very necessary in order to create effective protection
mechanisms against constantly changing threats. A wide variety of organizations,
ranging from educational establishments to research institutes specializing in cyber-
security, play an essential part in the process of gathering datasets from controlled
lab conditions and real-world situations.
Universities all throughout the world, such as the Canadian Institute for Cyberse-
curity, the United States Military Academy at West Point, and the University of New
South Wales in Sydney, are actively involved in research and collect databases that
serve as useful resources for scholars. These databases make it easier to find new
dangers and build effective solutions, both of which are important goals.
In addition, public platforms like Kaggle, GitHub, Medium, and the UCI Machine
Learning Repository provide researchers easy access to routes for discovering
and retrieving datasets. Because they are collaborative in nature, these platforms
encourage the exchange of information and speed up the progression of research
in the field of cybersecurity. Researchers may use these datasets to their advantage
by analyzing attack patterns, modeling threat behaviors, and coming up with novel
ways to better cybersecurity measures.
Ethical Considerations in AI-Based Cybersecurity 447

Table 1 Comparative analysis of cybersecurity tools

Functionality Tools Supported Integration Key features
platforms capabilities
Threat detection Darktrace, Windows, Linux, Limited Advanced threat
and analysis CrowdStrike, macOS (integration APIs detection,
Symantec available) real-time
monitoring,
automated
response
Malware Cylance, Windows, macOS Extensive Dynamic defense
detection Symantec, (integrations against evolving
McAfee with various threats,
tools) signature-less
protection
Endpoint CrowdStrike, Windows, macOS Limited Behavioral
security Carbon Black, (integration APIs analysis, threat
SentinelOne available) hunting, endpoint
detection and
response (EDR)
phishing Proofpoint, Windows, macOS Extensive Email threat
detection and Barracuda (integration APIs detection,
email security Sentinel, available) real-time
Ironscales protection,
incident response
automation
Behavioral Exabeam, Windows, Linux, Limited Anomaly
analytics Securonix, macOS (integration APIs detection, user
Varonis available) risk scoring,
automated
incident response
SIEM (Security Splunk, Windows, Linux, Extensive Log management,
Information and ArcSight, Elastic macOS (integration with real-time
Event SIEM various SIEM monitoring, threat
Management) solutions) intelligence
integration
Vulnerability Qualys, Tenable, Windows, Linux, Extensive Continuous
management Rapid7 macOS (integration with scanning, risk
various prioritization,
vulnerability compliance
scanners) management
Incident Demisto, Windows, Linux, Extensive Automated
response Phantom, IBM macOS (integration with playbooks, case
automation Resilient various security management,
tools) threat intelligence
integration
(continued)
448 K. Kaushik et al.

Table 1 (continued)
Functionality Tools Supported Integration Key features
platforms capabilities
UEBA (User and Gurucul, Rapid7 Windows, Linux, Extensive Insider threat
Entity Behavior InsightIDR, macOS (integration with detection,
Analytics) Splunk various identity risk-based
management analytics,
tools) anomaly detection
Network Vectra AI, Windows, Linux, Extensive Real-time threat
security Darktrace macOS (integration with detection,
Enterprise various network automated
Immune System, security response, network
Cisco solutions) anomaly analysis
Stealthwatch

A robust ecosystem of cybersecurity datasets has been contributed to by the

combined efforts of academic institutions, research institutes, and open platforms.
Researchers now have the ability to investigate new threats, test ideas, and eventu-
ally strengthen digital landscapes to withstand an ever-evolving cyber threat scenario
because of the quantity of data at their disposal. The fact that so many different
datasets are readily available helps to guarantee that the cybersecurity community
will continue to be at the forefront of innovation and preparation in the face of more
complex forms of online threats. Table 2 gives insights of available cybersecurity
datasets for researchers.

4 Use Cases of AI in Cybersecurity

This section discusses the use cases of Artificial Intelligence (AI) in Cybersecurity
area in detail. Figure 2 gives an overview of different use cases of AI in Cybersecurity.

4.1 Healthcare

Cybersecurity is critical to the healthcare industry because it ensures the availability,

accuracy, and privacy of sensitive patient data as well as critical medical systems.
Ensuring the security of vast quantities of patient data, including electronic health
records (EHRs) and personal information, is one of the most critical issues facing
healthcare companies today [35]. To address this issue, many security measures have
been implemented. Encrypting data, restricting who may access it, and routinely
upgrading security are some of the measures used to prevent data breaches and
unauthorized access. The increasing prevalence of connected medical equipment
and video services raises new issues. Safety procedures have been implemented
Table 2 Available cybersecurity datasets for researchers
S. No. Dataset Research Center/ Attacks Format Area of attack Description
Universities
1 CICEV202 Canadian Institute Ddos CSV EV Industry [20] A DDoS attack on any industry includes sending a huge
DDoS for Cybersecurity amount of data to its online systems, which can interrupt
services and put important operations at risk, posing a serious
threat to their usefulness and dependability
2 CICAndMa Canadian Institute Adware, Pcap Android [21] Adware shows annoying ads, ransomware locks files to get
2017 for Cybersecurity Ransomware, money, scareware scares people with fake threats, and SMS
Scareware, SMS malware gets into phones through text messages
Malware
3 ISCX Canadian Institute URL Pcap Website [22] URL attacks confuse users or exploit weaknesses by changing
URL2016 for Cybersecurity or hiding site addresses, typically resulting in unauthorized
access or dangerous information. URL verification is essential
to prevent phishing and malware
4 CICDarkne Canadian Institute VPN, Tor Parquet Audio-Stream, Attacks on VPNs and Tor take advantage of flaws to
Ethical Considerations in AI-Based Cybersecurity

2020 for Cybersecurity Browsing, Chat, compromise users’ privacy and get illegal access to private
Email, P2P, data. This shows how important strong encryption and security
Transfer, measures are for these privacy-focused technologies
Video-Stream and
VOIP [23]
5 UNSW UNSW Sydney Fuzzers, Analysis, Pcap IoT [24] Fuzzers test software for vulnerabilities, analysis examines
NB15 Backdoors, DoS, files, system weaknesses, backdoors allow unauthorized access,
Exploits, Generic, BRO DoS floods servers to disrupt, exploits exploit vulnerabilities,
Reconnaissance, files, generics cover various attacks, reconnaissance gathers intel,
Shellcode and Argus shellcode executes malicious operations, and worms
Worms files, self-replicate to spread rapidly in cyber attacks
CSV
files
(continued)
449
Table 2 (continued)
450

S. No. Dataset Research Center/ Attacks Format Area of attack Description

Universities
6 Bot-IoT UNSW Sydney DDoS, DoS, OS Pcap, IoT [25] DDoS floods systems with traffic, DoS blocks access, OS and
and Service Scan, Csv Service Scan looks for holes in the system, Keylogging records
Keylogging and keystrokes, and Data exfiltration leaks private data in
Data exfiltration cyberattacks
7 Honeypots Data Driven AWS honeypots Csv AWS [26] AWS honeypots act like security holes to attract and study
Security cyber risks. This helps us understand possible attacks and
makes cloud security better
8 2007 TREC 2007 Public Spam Csv Email [27] A spam attack sends a lot of unwanted and often useless
TRECPubli Corpus messages through communication channels to mess them up,
SpCorpus trick them, or spread harmful content
9 ICML-09 Detecting URL Matlab Website [28] A URL attack changes the names of websites to trick people or
Malicious URLs use holes in security. For your own safety, check URLs
carefully because they can lead to scams, malware, or people
breaking the law
10 CTU-13 CTU University Botnet traffic Pcap Web [29] As a network of hacked computers floods a target with
malicious requests, overloading systems and disrupting
services, botnet traffic attacks demonstrate the necessity for
strong defenses against coordinated and automated cyber
attacks
11 CTU-IoT Aposemat project Mirai Pcap IoT malware [30] The Mirai attack took advantage of IoT devices that weren’t
Malware safe, turning them into a huge botnet that slowed down online
Capture-341 services by flooding targets with too much traffic. This showed
how vulnerable connected devices are to cyber threats
12 Android StratosphereLab Remote Access Pcap Android [31] A Remote Access Trojan (RAT) attack enables an attacker to
Mischief Trojan (RAT) remotely manipulate a device, compromising security and
Dataset privacy. To reduce RAT risks, use robust security measures
(continued)
K. Kaushik et al.
Table 2 (continued)
S. No. Dataset Research Center/ Attacks Format Area of attack Description
Universities
13 IoT devices Alto University Networkflow CsV, Iot devices [32] Using network flow attacks to look into or change the trends of
captures Pcap network traffic data could put private information at risk. This
shows how important it is to have good tracking and breach
detection systems
14 KDD Cup University of Dos, R2L, U2R Pcap IP address [33] DDoS stops services from working, R2L uses security holes to
1999 Data California get in illegally, and U2R gives users more rights, so specific
security measures are needed
15 Domain USMA network DNS Pcap CDX network [34] DNS attacks use DNS flaws to stop domain names from being
Name resolved to IP addresses and send users to hostile websites.
Service Logs This shows how important it is to have good DNS security
Ethical Considerations in AI-Based Cybersecurity
451
452 K. Kaushik et al.

Fig. 2 Use cases of AI in cybersecurity

in order to safeguard these technologies and maintain the security of patient inter-
actions [36]. [Protective measures for the internet] [Cybersecurity measures] are
designed to safeguard patient communications and maintain the security of these
instruments [37]. Pharmaceutical firms that engage in research and development
also have cybersecurity issues. Advanced systems that can detect attacks and prevent
data loss are necessary for these organizations to safeguard their intellectual property
and important research data. As more individuals utilize linked technology, such as
video services and medical devices, new issues begin to surface. To safeguard these
technologies and maintain patient contacts’ privacy, security precautions have been
taken. [Care should be used to safeguard online] The privacy of these technologies and
the exchanges between patients are being safeguarded by [cybersecurity] measures.
Hacking is another issue that pharmaceutical businesses that engage in research and
development may face [38]. These organizations need sophisticated systems that can
identify potential attacks and prevent data loss in order to safeguard their priceless
research records and intellectual property. Aside from emergency response systems,
other crucial topics include hospital networks and staff awareness training programs.
Ethical Considerations in AI-Based Cybersecurity 453

Through the consideration of these use cases, healthcare institutions may construct
a robust cybersecurity framework that guarantees the confidentiality of patient data
while simultaneously ensuring the continuous availability of critical systems and
services. This enhances both the quality of treatment provided to patients and their
general health.

4.2 Finance

In the banking industry, cybersecurity is crucial since several threats have the poten-
tial to compromise confidential financial information and halt significant transac-
tions. Ensuring the security of financial transactions is among the most significant
challenges. In order to prevent theft and unwanted access, it is crucial to use trans-
action monitoring software, multi-factor authentication, and robust encryption in
this scenario. Numerous personal details about its clients are handed to financial
organizations. In order to prevent unauthorized access or data breaches, they use
sophisticated encryption, restrict who may access it, and conduct frequent security
audits. The use of cards and automated teller machines (ATMs) presents new cyber-
security issues. Installing secure hardware, updating software often, and maintaining
tight monitoring are necessary to prevent card skimming and other kinds of fraud
[39]. There are now greater security concerns in this new arena as mobile banking
gains traction. Making secure mobile applications, routinely checking security, and
teaching clients the best practices for using their banks are all necessary to address
these issues. To prevent financial information from being stolen online and to main-
tain the integrity of the market, online trading platforms must be secure. Real-time
tracking, security testing, and safe coding approaches are required for this to func-
tion. Strict guidelines regarding who may access what, employee training programs,
and constant monitoring of user behavior to identify and address any unusual activity
are all necessary to guard against insider threats. Respecting the guidelines estab-
lished by authorities such as PCI DSS and GDPR is also crucial. For this reason,
extensive safety regulations and routine inspections were created. It’s more difficult
to monitor your cryptocurrency wallet securely, so you should use multi-signature
authentication and other cutting-edge security techniques to keep your bitcoin safe
from theft [1]. Supply chain security, crisis response plans, and recovery plans are
all components of a comprehensive cybersecurity strategy that protects the banking
sector from emerging internet threats.

4.3 Supply Chain

Cybersecurity becomes more important as supply chains become more complex in

order to ensure the integrity of data, protect critical operations, and facilitate the
flow of information. Technology used in shipping makes it more difficult to protect
454 K. Kaushik et al.

Internet of Things (IoT) devices from hackers. Ensuring that these devices are not
accessed by unauthorized individuals requires robust security measures. Several of
them include systems that scan for breaches, encryption, and frequent modifications.
Because inventory management systems are a crucial component of the supply chain,
they must be shielded against intrusions and unauthorized access. Important possibil-
ities include access controls, security audits, and open inventory management using
blockchain technology. Companies must establish secure channels of communication
and provide frequent hacking training to staff members who interact with suppliers
in order to ensure that confidential information is sent and received by suppliers
in a secure manner. Warehouse management systems need to be shielded from the
dangers associated with security breaches since they preserve correct product data
[40]. These systems may be kept secure via data consistency checks, blockchain
technology for unchangeable record storage, and other security measures. Strong
security measures must be implemented immediately since transportation systems,
particularly vehicles connected to distribution platforms, are vulnerable to hacking.
By ensuring that connected automobiles are safe, establishing secure communication
channels, and routinely assessing the security of delivery platforms, vulnerabilities
in the transportation system may be minimized. Anti-counterfeiting technologies,
blockchain monitoring, and routine inspections are required to prevent counterfeit
items from entering the supply chain. It needed cross-border training and a global
cybersecurity strategy in order to manage cybersecurity across international supply
chains with many partners and legal contexts. Businesses must prepare for supply
chain resilience if they want to continue operating in the face of interruptions and
hacks [41]. The significance of cybersecurity for the stability and dependability of
the supply chain is shown by these many examples.

4.4 Digital Media

The world of digital media is constantly evolving. In order to maintain the safety
of the intricate interaction between content providers, users, and websites, cyber-
security has become crucial. One of the most important issues is keeping digital
assets secure. Use digital rights management (DRM), encryption, and watermarking
to prevent unauthorized users from stealing and distributing digital material. Since
companies retain so much personal data about their users, streaming services struggle
to protect their privacy. Strict adherence to data protection regulations, encryption
techniques, and explicit privacy rules are necessary to safeguard the security and
privacy of user information [42]. Although online advertising networks play a signif-
icant role in digital media business plans, issues like virus placement and ad frauds
harm them. Finding and preventing frauds depends heavily on using ad verifica-
tion tools, maintaining the security of ad networks, and doing routine audits. Secure
payment methods, robust login procedures, and frequent security audits are necessary
for digital media organizations to prevent unauthorized users from accessing paid
material, which is often how they generate the majority of their revenue. Social media
Ethical Considerations in AI-Based Cybersecurity 455

platforms, which are crucial for digital communication, are struggling with issues
including user account takeovers by hackers and the dissemination of misleading
information. Social networking platforms may be made safer for users with the use
of information programs, content screening technologies, and multi-factor authenti-
cation. Content makers utilize copyright protection, digital watermarking, and legal
procedures to prevent piracy. Digital media websites need robust network architec-
ture, content delivery networks (CDNs), and security techniques against denial-of-
service (DDoS) attacks [43]. The legitimacy of digital media content is under danger
due to deepfake technology; thus, money has to be invested on systems that can detect
it, consumers need to be made aware of the risks, and cooperation between all parties
is required. Digital products retailers on the internet face the danger of fraudulent
transactions and data breaches. Robust payment mechanisms, secure e-commerce
platforms, and frequent security audits are essential for cybersecurity. Finally, user-
generated content websites must handle abuse and unauthorized file submissions.
User reports, content control systems, and community guidelines all contribute to a
secure and entertaining online environment.

4.5 Research Institutes

Research organizations are primarily in charge of a number of significant respon-

sibilities, including advancing human understanding, fostering the development of
fresh concepts, and solving difficult challenges. These institutions serve as centers
for innovative research. To explore uncharted ground, they bring together groups of
academics, researchers, and businesspeople from many sectors. Research institutes
provide significant contributions to scientific discoveries, ground-breaking discov-
eries, and answers to pressing global issues via meticulous testing, in-depth anal-
ysis, and collaboration with other professional groups [44]. They often work in a
wide range of fields, including technology, natural science, medicine, and the social
sciences. In addition to doing fundamental research, these centers play a critical role
in educating the next generation of scientists, disseminating knowledge, and facil-
itating collaboration between government, business, and academic organizations.
Research institutes have a significant influence on the globe in addition to their posi-
tion in education [45]. They have the power to influence public policy, grow the
economy, and direct human development.

4.6 SCADA

Supervisory control and data acquisition, or SCADA, systems are utilized in many
different business domains and are essential for monitoring and managing complex
operations. SCADA is used in the energy industry to monitor and control the produc-
tion, transmission, and distribution of electricity. This contributes to ensuring the
456 K. Kaushik et al.

effective and dependable transmission of electricity. SCADA enables real-time moni-

toring of manufacturing lines in the industrial sector. Additionally, it enhances the
overall utility of processes and the pace at which work is done. Water and wastewater
treatment facilities employ SCADA to monitor water quality, automate processes to
maximize resource use, and regulate water flow [46]. SCADA systems are useful in
the transportation industry because they facilitate the management of many aspects
such as airport operations, rail systems, and traffic signals. Additionally, SCADA
plays a major role in ensuring the safety and problem-solving capabilities of critical
equipment. It does this by providing control, real-time knowledge, and the capacity
to act. Reducing the impact of any delays and improving system reliability are the
main objectives. SCADA is a fundamental technology for enhancing the safety and
efficiency of complex industrial processes because of its wide range of applications.

4.7 Big Data and Analytics

Big Data and analytics are a revolutionary force in today’s world, revolutionizing
the way organizations get insights from vast quantities of complex data. This is just
one example of the numerous ways that big data and analytics are transforming the
world. “Big data” refers to the ability to manage and analyze enormous volumes of
ordered and disorganized data—often more than what typical systems can handle.
Conversely, analytics involves a meticulous examination of this data in order to
identify potentially valuable patterns, relationships, and trends [47]. When combined,
these elements aid in improving decision-making, streamlining processes, and giving
businesses a competitive advantage. The use of big data and analytics is widespread
and serves a variety of purposes, including enhancing consumer satisfaction, fore-
casting market trends, enhancing healthcare outcomes, and streamlining supply chain
operations. In our data-driven world, collaboration between big data and analytics
is more important than ever. It provides you with more opportunities than any other
element to be resourceful, frugal, and make well-informed judgments.

4.8 Internet of Thing

The Internet of Things (IoT) has already been shown to have several significant
applications. Numerous commercial sectors are changing as a result of these usage,
and many aspects of daily life are improving. The Internet of Things allows for remote
patient monitoring in the medical field [48]. This provides current information on
long-term ailments and vital signs to medical workers. Smart cities are using IoT
(Internet of Things) to link infrastructure, enhance public services, and improve
traffic flow in order to improve urban living. Internet of Things monitors provide
agriculturalists with information on crop growth, weather patterns, and the condition
of the land. They can utilize better, more sustainable agricultural techniques as a
Ethical Considerations in AI-Based Cybersecurity 457

result. Predictive maintenance in industrial settings is made feasible by the Internet

of Things, which monitors the condition of the machinery. This reduces the amount
of time and associated expenditures when the equipment is offline [49]. IoT gadgets
in the house, such as security systems and smart heaters, provide each individual
greater control over their level of safety and comfort [50]. These many application
examples demonstrate how the Internet of Things has transformed society by linking
settings that improve the productivity, sustainability, and creativity of a wide range
of business sectors.

4.9 Wireless Sensor Network

Wireless sensor networks, or WSNs, are being employed in a wide range of appli-
cations due to their ability to collect data remotely and follow objects in real time.
In the field of environmental tracking, weblogs make it simpler to monitor vari-
ables like tree health, weather patterns, and air and water quality [51]. WSNs are
helpful in the business sector because they enable you to do preventive mainte-
nance by continuously monitoring the state of equipment and tools. This reduces
unscheduled downtime and improves the efficiency of operations. In the medical
field, Wi-Fi sensor networks (WSNs) are highly helpful for monitoring patients,
recording their vital signs, and assisting physicians in identifying health issues early
on. “Smart agriculture” refers to the use of wireless sensor networks (WSNs) for
precision farming, optimal drainage management, and land monitoring to enhance
crop growth. Monitoring the condition of vital infrastructure, such as highways and
dams, is another crucial application required to ensure the safety and soundness of
significant facilities [52]. These examples demonstrate how versatile and essential
wireless sensor networks are for creating rapid and intelligent solutions across a wide
range of industries.

4.10 Defense Services

Modern technology is used by the military forces in a variety of operational envi-

ronments to protect national security and achieve a tactical edge. Unmanned aerial
vehicles, or UAVs, are flying machines equipped with sophisticated cameras and
monitoring systems that allow them to collect data, monitor their surroundings, and
gaze about in real time. This improves people’s overall understanding of the problem.
The importance of cyber security in protecting military networks and assets against
threats and cyberattacks cannot be overstated. Military operations are made simpler
by satellite technology, which provide secure means of communication, navigation,
and position determination worldwide [53]. Drones designed with military specifica-
tions and fitted with precise armaments are often used for precise strikes in hazardous
458 K. Kaushik et al.

environments. People are less likely to be injured as a result of this. Biometric tech-
nologies are used by people to verify their identity and prevent unauthorized access
to secure locations. This makes some locations inaccessible to anybody other than
authorized staff [54]. The armed forces also utilize data analytics to examine vast
volumes of data. This aids in their decision-making, threat assessment, and mission
planning. You can see how crucial cutting-edge technology is for enhancing military
capabilities and maintaining national security from these examples of uses.

4.11 Robotics

Robotics is being used in many distinct and revolutionary ways in several sectors. As
a result, processes have evolved and become more effective. When robots are used in
production, they automatically perform precise, repetitive jobs. This reduces errors
while increasing manufacturing speed. Medical treatments may be performed with
little harm to patients thanks to robotic surgical equipment, which improves patient
outcomes and precision. Autonomous vehicles, a kind of robotics, are transforming
the transportation sector in a number of ways, including self-driving cars, delivery
robots, and drones [55]. Robotic technology has applications in agriculture that are
beneficial to the industry. As an example, both precision farming and automated
reaping increase food output. Robots are an essential component of the response to
natural disasters since they may be used for search and rescue operations as well
as in hazardous environments. By assisting with activities like picking, packaging,
shipping, and tracking of items in the logistics and storage industries, robots are
increasing the efficiency of the supply chain [56]. The many applications of robots
here demonstrate how adaptable and extensive the technology is in transforming
several industries and resolving challenging issues.

4.12 Augmented Reality/Virtual Reality

The way that people engage with augmented reality (AR) and virtual reality (VR)
has completely altered as these technologies have proliferated in various businesses.
With the use of virtual reality (VR), educators may build realistic models that allow
students to learn by doing. Virtual reality field excursions and interactive anatomy
lessons are two instances of VR in the classroom. Virtual reality (VR) is being
utilized in the medical industry for therapeutic reasons, such as exposure treatment
for anxiety disorders [57]. Augmented reality (AR) is used in the medical industry
for educational purposes. Shoppers can virtually try on clothing thanks to retailers
using augmented reality (AR), which improves online buying in general. Meanwhile,
realistic product presentations and virtual marketplaces are made feasible by virtual
reality (VR) [58]. AR overlays, which provide in-the-moment employment advice,
and VR models, which replicate potentially hazardous surroundings, are helpful for
Ethical Considerations in AI-Based Cybersecurity 459

workplace training. Virtual reality (VR) allows users to take comprehensive tours
of virtual locations, while augmented reality (AR) allows users to see buildings
in their natural environments. Virtual reality (VR) is utilized in the entertainment
industry to provide greater realism in video games, while augmented reality (AR)
adds captivating features to live events to make them better [59]. new applications of
virtual and augmented reality demonstrate how new technologies are revolutionizing
a wide range of industries by improving participation, experiences, and instruction.

4.13 Smart City

Smart cities employ state-of-the-art technology to improve urban living and increase
sustainability and efficiency in many areas. Smart parking options, ingenious
traffic control systems, and real-time public transportation monitoring are all being
employed as part of smart city initiatives to improve mobility by making the city
less congested and more functional. Energy-efficient infrastructure contributes to
resource management and energy consumption that is non-destructive [60]. These
include web-connected devices and services, such as smart grids. Smart city tech-
nology is used in public safety to monitor situations and act swiftly in case of an
emergency via the use of surveillance cameras, sensor networks, and data analytics,
among other tools. Digital platforms benefit urban governance by facilitating more
participation, data-driven decision-making, and transparent service delivery [61].
Cities may become healthier and more sustainable by doing things like monitoring the
environment, cleaning up after litter, and conserving water. A connected ecosystem
may be established by combining data processing with Internet of Things (IoT)
devices [62]. This improves people’s lives generally and enables municipalities to
react swiftly to emerging issues.

4.14 Gaming Server

In order to ensure that players can connect with one another, that games run smoothly,
and that a variety of games may be played, gaming servers are required. In massively
multiplayer online games, or MMOs, computers manage a huge number of partic-
ipants and enable real-time communication, collaboration, and competition [63].
First-person shooter (FPS) games need computers for low-latency exchanges. This
guarantees that gameplay is swift and equitable for all players. Role-playing games,
or RPGs, allow users to connect to servers where they may explore endless worlds,
advance their characters, and share narratives with other players. When many players’
movements are coordinated and intricate game states are monitored by a server,
strategy games become even more entertaining. E-sports tournaments also need game
servers, which must be dependable and responsive to ensure fair competition. Having
a robust computer network is even more crucial in light of the popularity of cloud
460 K. Kaushik et al.

gaming. This is due to the fact that excellent games may be watched without requiring
strong equipment to be close by [64]. This makes powerful computers even more
crucial nowadays. An essential component of the technology that enables internet
gaming is the game server. They also set the broad rules for how the game is played.

4.15 Power Grid/Nuclear Plant

Modern technology is essential to power grids and nuclear power plants in order to
ensure that energy is generated, transferred, and delivered in the most dependable
and safe manner possible. Monitoring and managing the electrical grid’s operation
using Supervisory Control and Data Acquisition (SCADA) systems is critical. These
tools enable workers to manage energy flow, react to power outages, and improve
grid performance in real time [65]. Power firms may also use predictive analytics
to schedule repairs, increase system resilience, and estimate demand. Automation
and sensors are employed in nuclear power plants’ safety systems to monitor the
reactor, search for issues, and initiate safety procedures as required. The use of
robotics and other remote-operated technology facilitates the restoration of hazardous
environments with less human exposure. Power lines and nuclear power plants are
made more stable, secure, and long-lasting by these technical applications. Therefore,
they ensure that there is a reliable and secure supply of electricity to fulfill the rising
demand for energy.

5 How to Achieve Trustworthy Cybersecurity with AI?

Artificial Intelligence requires a comprehensive strategy including scientific, ethical,

and pragmatic components to ensure human safety. This method’s premise is that
data security will be maintained. Strong security measures, such as encryption proto-
cols, secure storage techniques, and stringent access restrictions, are necessary for
businesses to protect sensitive data from unauthorized access and potential breaches.
Adhering to data protection regulations is crucial if you want to ensure that your
handling of personal and private information is appropriate. It must be evident for
humans to trust AI to make judgments. When AI models explain why users made
the decisions they did, users tend to trust them. Techniques for interpretable machine
learning models, or explainable artificial intelligence (XAI), aid in the understanding
of algorithmic processes and decision-making. Working with professionals in AI and
human protection requires transparency and honesty, which fosters trust [66]. When
deploying AI for protection, ethics are crucial. Ethics must guide the development
and use of AI. In order to ensure equity and diversity, organizations must combat
prejudiced AI systems. Regular tests for bias in AI systems encourage ethical AI.
Developers, consumers, and other stakeholders must be informed about initiatives
Ethical Considerations in AI-Based Cybersecurity 461

that support responsible AI behavior and must continue learning about AI in order
for it to be utilized in an ethical manner.
An effective protection strategy must be updated and monitored often. AI-driven
real-time threat identification and adaptable responses improve cyber risk detection
and response. AI models and algorithms need to be updated often to stay up to date
with new cyber threats. To avoid and resolve security issues, comprehensive incident
response procedures that seamlessly integrate AI should be used. To provide trust-
worthy AI-powered defense solutions, safe AI development is required. AI model
developers must build secure code, monitor for security flaws, and maintain AI tools
up to date in order to address them [67]. Artificial intelligence applications may
be protected against hacking and unauthorized usage using preventive protection.
Adhering to the rules is a crucial component in attacking AI trust. It’s critical to stay
current on the laws and regulations pertaining to hacking. AI applications should
abide by security and privacy regulations to avoid legal issues and to perform at the
highest level in the industry.

6 Challenges and Limitations of AI in Cyber Security

AI in defense has many drawbacks and limitations, despite the possibility of signif-
icant advancements. One major issue is the dynamic nature of internet threats. AI
systems can be vulnerable to attacks since adversaries are constantly altering their
intentions. Additionally, asymmetrical attacks are made feasible by the intrinsic
complexity of AI algorithm construction. These occur when crafty attackers attempt
to evade detection by manipulating the AI’s decision-making processes. A different
problem is that there aren’t enough designated good quality datasets [68]. This poses
a challenge to the development of trustworthy AI models. Furthermore, a lot of AI
algorithms lack total clarity, which might make them more difficult to operate and
comprehend.
As a result, it might be challenging for those in the safety field to completely
comprehend and have faith in the decisions made by AI systems. This makes it far
more difficult to tackle the issues than it would be otherwise since there aren’t many
qualified safety specialists who are proficient with AI-powered systems. It is crucial
to ensure that artificial intelligence advances rather than replaces human expertise in
order to strike a balance between the control automation grants people over processes
and the control automation grants humans over those processes. In order to use AI’s
potential for defense and build a robust and secure digital environment, it is imperative
to address these issues and constraints [69]. Owing to these issues and constraints,
doing both at once is not feasible.
462 K. Kaushik et al.

7 Need of Cybersecurity AI Frameworks for Ethical

Considerations

The cybersecurity landscape is dynamic, and artificial intelligence (AI) is playing a

bigger role in fortifying defenses against more sophisticated attacks. However, as AI
technologies are becoming more and more crucial for identifying dangers, handling
emergencies, and reaching judgments, the societal challenges that arise from their
application have received a lot of attention [70]. This is because artificial intelligence
(AI) technologies are playing an increasingly significant role in these processes. AI
systems with comprehensive protection and moral foundations are becoming more
crucial. Some of the most significant issues, such as the possibility of bias in AI
systems, are addressed by ethics models [71]. Due to the fact that these algorithms
are trained on historical data, there is a chance that they can unintentionally amplify
preexisting biases in the data. Fairness, transparency, and accountability need to be
crucial components of the ethical AI defense architecture.
As a consequence, the likelihood of distorted outcomes from the AI will decrease.
One approach that businesses may take to ensure that AI tools function accurately and
impartially is to establish standards for assessing and mitigating bias in AI models.
You may find these recommendations here [72]. Another crucial component of moral
AI protection systems is transparency. Some refer to sophisticated AI models—
particularly deep learning models—as “black boxes” since it’s difficult to understand
how they arrive at conclusions. Open-minded moral theories provide credence to
explainable AI (XAI) models. These models increase trust and a feeling of obligation
by assisting those who work in the safety field in understanding judgments made by
AI [73].

8 Role of Explainable AI in Cybersecurity

Explainable Artificial Intelligence (XAI) is becoming a fundamental component of

open and trustworthy defensive systems. Safety options must be clear. XAI is crucial
because it shows how AI models decide. By making AI findings understandable, XAI
solves the “black box” problem [74]. In cybersecurity, this includes understanding
how AI systems identify and manage threats. XAI helps security specialists compre-
hend system decisions—leading to better-educated and well-planned responses to
security flaws or odd behavior.
XAI simplifies AI-powered protection solutions, boosting user confidence. Secu-
rity professionals and decision-makers may verify AI reports and proposals. Open-
ness is crucial to ensure that AI systems are moral, bias-free, and accountable
[75].
Ethical Considerations in AI-Based Cybersecurity 463

9 Responsibility and Accountability of AI-Based

Cybersecurity Systems

The evolving landscape of digital defense raises many concerns, chief among them
the obligation and accountability of AI-driven defensive systems. Regretfully, these
state-of-the-art technologies excel at identifying risks, stopping them, and lessening
their effects. But in order for them to function effectively, roles and duties must be
established. Businesses that use AI in the hacking space are accountable for ensuring
that the systems are secure and that they are utilizing the data they get honestly
[76]. Strong governance frameworks must be established in order to achieve this.
These systems need to include moral norms, compliance, and supervision. Ensuring
the accuracy and dependability of AI choices is the responsibility of cybersecurity
specialists monitoring AI systems. Artificial intelligence programs are more respon-
sible since their output is explicable and their algorithms are transparent [77]. This
facilitates comprehension and evaluation of the justifications for taking protective
action by all parties. Additionally, engineers and programmers play a critical role in
ensuring that AI models are robust, safe, and impartial. This supports the promotion
of duty throughout the system’s lifetime. Everyone must feel that they have a shared
obligation in order to reduce risks, maintain moral behavior, and maintain account-
ability when it comes to safeguarding digital environments. Coders and end users
are included in this [78]. This is even more crucial now that artificial intelligence is
being used in defense on a growing basis.

10 Ethical Issues with Large Language Models

Late last year, conversational AI chatbots, also known as generative AI (GAI),

produced stunning text, images, code, music, and video depending on user requests.
Many AI applications gained stakeholder and media interest because of their promise,
flexibility, simplicity, and availability. Because of this, they had numerous users and
varied uses. Conversational AI chatbot ChatGPT can answer general and specific
questions and write effectively on any topic. The response may be refreshed following
user input. TikTok’s nine-month milestone was exceeded by 1 million users in four
days and 100 million in two months. GAI tools were popular and expensive to create.
Personal and business apps utilize GAI. OpenAI’s ChatGPT API lets developers add
ChatGPT to programs. Microsoft is bringing it to Power and Dynamics. SAP and
others incorporate ChatGPT into applications [79]. Thus, many businesses are using
GAI more. GAI is promising, popular, and praised yet has major constraints. Users are
bewildered by ChatGPT’s factual errors, fabrications, and wrong answers. AI content
producers raise ethical challenges that developers, users, and politicians must address.
Unexpected impacts might harm society, corporations, and the economy [80].
A big moral problem is bias in AI content producers, especially text engines like
ChatGPT. The big language models (LLMs) that run content makers are trained on
464 K. Kaushik et al.

a lot of information, pictures, and data from many sources, including the Web. This
means that any flaws in the training data will show up in the model output. This
could lead to answers that are cruel, biased, wrong, or narrowly focused, as well as
race or gender abuse [81]. GAI is biased because of how it was built and trained.
LLM answers will be affected by biased facts and information, even if they are not
on purpose. So, to get a good result and lessen bias, you need a sample that is fair,
varied, and representative. AI may be less biased if data is curated, filtered, and
training data is chosen. These steps look harder to put into action because there is so
much training data. But software companies like OpenAI are making chatbots less
biased and letting users change how they behave.
Misusing and taking advantage of AI content creators is another moral problem.
Text makers like ChatGPT might spread lies, racist, sexist, or otherwise offensive
messages. This tool could also be used to make harmful content that encourages
violence or pretends to be someone else. Since users can ask any question, bad
people could use the bot to act in a rude way [82]. They could learn how to make
bombs, shoplift, or cheat from chatbots. So, safeguards are needed to stop and punish
people who abuse technology. People who use tools like ChatGPT may be more
likely to copy other people’s work in ways that are hard to spot. ChatZero and AI
Text Classifier are two tools that try to tell the difference between text written by AI
and text written by a person. Copying is hard to do without AI tools, so this is done
to stop it.
Hackers could easily use AI content producers to make personalized junk
messages and images that hide harmful code. The people you want to reach might
be persuaded by these words and images. There may be a lot more hacks because of
this, which could hurt a lot of people. Chatbot users may also give private personal or
business information to them, which could lead to the creator abusing the data users
give them. The creator could decide to keep and look over the data that people give
them. Think about these two use cases as models that will help you learn [83]. An
important person in the company recently copied and pasted the 2023 strategy paper
into a robot and asked it to make presentation PowerPoint slides. In a different case, a
doctor used ChatGPT to enter the name of his patient and the patient’s medical situ-
ation. He then asked the software to write a letter to the patient’s insurance company
on his behalf. There are social concerns, problems with data protection, and security
risks and threats that come with AI content creators that can be seen in these use
cases.
When AI creates content, copyright issues come up. For example, who owns
the rights to a story, song, or piece of art that AI created? Who is responsible for
teaching the chatbot? The people who gave the data that was used to teach it, or
the people who came up with the questions and answers that the AI should give? In
the context of this conversation, you should know that the US Copyright Office has
said that pictures made by Midjourney and other AI text-to-image technologies are
not protected by US copyright law because they were not created by a person and
cannot be considered works of art [84]. A class action lawsuit has been filed against
companies that sell AI-generated art. The case questions whether it is legal to use
Ethical Considerations in AI-Based Cybersecurity 465

data to train an AI without the permission of the people who gave the data. Artists
brought the case to court.

11 Future Research Directions—IOTA, Federated

Learning Blockchain, and Explainable AI

Exciting research opportunities arise from the combination of IOTA, blockchain,

explainable AI, and shared learning. IOTA’s Tangle DAG ledger and shared learning
make sense when combined. How data privacy and security in decentralized systems
might be improved by the Tangle by providing a clear and secure communication
channel for participants in shared learning. Safe and cooperative machine learning
models may be achieved using blockchain-based open cooperative learning. In order
to reduce latency and resource consumption without sacrificing the independence of
the system, researchers may focus on enhancing the scalability, efficiency, and agree-
ment techniques of blockchain-based federated learning in the future. Explainable
AI is critical to cooperative learning because transparency fosters trust. Scholars can
concentrate on developing interpretable models that demonstrate decision-making
processes while safeguarding contributor privacy. The goal of federated learning
and explainable AI is to create a mechanism to understand models in decentralized
settings while maintaining the privacy of data. The primary area of study is IOTA-
distributed learning techniques that protect privacy. Autonomous model training and
the protection of private information may be facilitated by homomorphic encryption
and secure multi-party computing. Before federated learning is extensively used in
the real world, privacy concerns must be resolved. Decentralized systems struggle to
expand and operate quickly. This research might investigate methods for managing
large gatherings of individuals more effectively, cutting down on communication
expenses, and expediting the decision-making process. Scalability issues must be
resolved for shared learning on blockchain networks to be feasible. Decentralized
systems struggle to expand and operate quickly. This research might investigate
methods for managing large gatherings of individuals more effectively, cutting down
on communication expenses, and expediting the decision-making process. Scala-
bility issues must be resolved for shared learning on blockchain networks to be
feasible. Further investigation is required to ensure the safety and dependability of
autonomous systems using shared learning, blockchain, and IOTA. To safeguard
decentralized federated learning models, vulnerabilities must be identified, secure
smart contracts must be created, and barriers against hostile attacks must be erected.
466 K. Kaushik et al.

References

1. Ahmed F (2022) Ethical aspects of artificial intelligence in banking. J Res Econ Finance Manag
1:55–63. https://doi.org/10.56596/jrefm.v1i2.7
2. Johnson A, Grumbling E (eds) (2019) Implications of artificial intelligence for cybersecurity.
National Academies Press, Washington, DC. https://doi.org/10.17226/25488
3. Martinho A, Herber N, Kroesen M, Chorus C (2021) Ethical issues in focus by the autonomous
vehicles industry. Transp Rev 41:556–577. https://doi.org/10.1080/01441647.2020.1862355
4. Mirbabaie M, Hofeditz L, Frick NRJ, Stieglitz S (2022) Artificial intelligence in hospitals:
providing a status quo of ethical considerations in academia to guide future research. AI Soc
37:1361–1382. https://doi.org/10.1007/s00146-021-01239-4
5. Dash B, Ansari MF, Sharma P, Ali A (2022) Threats and opportunities with AI-based cyber
security intrusion detection: a review. Int J Softw Eng Appl 13:13–21. https://doi.org/10.5121/
ijsea.2022.13502
6. Naik N, Hameed BMZ, Shetty DK, Swain D, Shah M, Paul R, Aggarwal K, Ibrahim S, Patil
V, Smriti K, Shetty S, Rai BP, Chlosta P, Somani BK (2022) Legal and ethical consideration
in artificial intelligence in healthcare: who takes responsibility? Front Surg 9. https://doi.org/
10.3389/fsurg.2022.862322
7. Helkala K, Cook J, Lucas G, Pasquale F, Reichberg G, Syse H (2023) AI in cyber operations:
ethical and legal considerations for end-users. In: Tuomo S, Kokkonen T (eds) Artificial intel-
ligence and cybersecurity: theory and applications. Springer International Publishing, Cham,
pp 185–206. https://doi.org/10.1007/978-3-031-15030-2_9
8. Rodrigues R (2020) Legal and human rights issues of AI: gaps, challenges and vulnerabilities.
J Responsib Technol 4:100005. https://doi.org/10.1016/j.jrt.2020.100005
9. Al-Mansoori S, Salem MB (2023) The role of artificial intelligence and machine learning in
shaping the future of cybersecurity: trends, applications, and ethical considerations. Int J Soc
Anal 8:1–16
10. Wirtz BW, Weyerer JC, Geyer C (2019) Artificial intelligence and the public sector—applica-
tions and challenges. Int J Public Adm 42:596–615. https://doi.org/10.1080/01900692.2018.
1498103
11. Walshe R, Koene A, Baumann S, Panella M, Maglaras L, Medeiros F (2021) Artificial intel-
ligence as enabler for sustainable development. In: 2021 IEEE international conference on
engineering, technology and innovation (ICE/ITMC), pp 1–7. https://doi.org/10.1109/ICE/
ITMC52061.2021.9570215
12. Chaudhary H, Detroja A, Prajapati P, Shah P (2020) A review of various challenges in
cybersecurity using artificial intelligence. In: 2020 3rd international conference on intelligent
sustainable systems (ICISS), pp 829–836. https://doi.org/10.1109/ICISS49785.2020.9316003
13. Shaukat K, Luo S, Varadharajan V, Hameed IA, Xu M (2020) A survey on machine learning
techniques for cyber security in the last decade. IEEE Access 8:222310–222354. https://doi.
org/10.1109/ACCESS.2020.3041951
14. Timmers P (2019) Ethics of AI and cybersecurity when sovereignty is at stake. Minds Mach
(Dordr) 29:635–645. https://doi.org/10.1007/s11023-019-09508-4
15. Romancheva NI (2021) Duality of artificial intelligence technologies in assessing cyber security
risk. IOP Conf Ser Mater Sci Eng 1069:12004. https://doi.org/10.1088/1757-899X/1069/1/
012004
16. Ferreyra NED, Aimeur E, Hage H, Heisel M, van Hoogstraten CG (2020) Persuasion meets
AI: ethical considerations for the design of social engineering countermeasures. CoRR. abs/
2009.12853
17. Christodoulou E, Iordanou K (2021) Democracy under attack: challenges of addressing ethical
issues of AI and big data for more democratic digital media and societies. Front Polit Sci 3.
https://doi.org/10.3389/fpos.2021.682945
18. Abdulllah SM (2019) Artificial intelligence (AI) and its associated ethical issues. ICR J 10:124–
126. https://doi.org/10.52282/icr.v10i1.78
Ethical Considerations in AI-Based Cybersecurity 467

19. Zhang Z, Ning H, Shi F, Farha F, Xu Y, Xu J, Zhang F, Choo K-KR (2022) Artificial intelligence
in cyber security: research advances, challenges, and opportunities. Artif Intell Rev 55:1029–
1053. https://doi.org/10.1007/s10462-021-09976-0
20. Kim Y, Hakak S, Ghorbani A (2023) DDoS attack dataset (CICEV2023) against EV authen-
tication in charging infrastructure. In: 2023 20th annual international conference on privacy,
security and trust (PST). IEEE Computer Society, Los Alamitos, CA, pp 1–9. https://doi.org/
10.1109/PST58708.2023.10320202
21. Lashkari AH, Kadir AFA, Taheri L, Ghorbani AA (2018) Toward developing a systematic
approach to generate benchmark android malware datasets and classification. In: 2018 interna-
tional Carnahan conference on security technology (ICCST), pp 1–7. https://doi.org/10.1109/
CCST.2018.8585560
22. Moustafa N, Creech G, Slay J (2018) Anomaly detection system using beta mixture models
and outlier detection. In: Kumar PP, Rautaray SS (eds) Progress in computing, analytics and
networking. Springer Singapore, Singapore, pp 125–135
23. Habibi Lashkari A, Kaur G, Rahali A (2021) DIDarknet: a contemporary approach to detect
and characterize the darknet traffic using deep image learning. In: Proceedings of the 2020 10th
international conference on communication and network security. Association for Computing
Machinery, New York, NY, pp 1–13. https://doi.org/10.1145/3442520.3442521
24. Moustafa N, Slay J (2015) UNSW-NB15: a comprehensive data set for network intrusion
detection systems (UNSWNB15 network data set). In: 2015 military communications and
information systems conference (MilCIS), pp 1–6. https://doi.org/10.1109/MilCIS.2015.734
8942
25. Koroniotis N, Moustafa N, Sitnikova E, Turnbull BP (2018) Towards the development of
realistic botnet dataset in the internet of things for network forensic analytics: Bot-IoT dataset.
CoRR. abs/1811.00701
26. Valeros V, Garcia S (2022) Hornet 40: network dataset of geographically placed honeypots.
Data Brief 40:107795. https://doi.org/10.1016/j.dib.2022.107795
27. Moustafa N, Creech G, Slay J (2018) Flow aggregator module for analysing network traffic.
In: Kumar PP, Rautaray SS (eds) Progress in computing, analytics and networking. Springer
Singapore, Singapore, pp 19–29
28. Ma J, Kulesza A, Dredze M, Crammer K, Saul L, Pereira F (2010) Exploiting feature covariance
in high-dimensional online learning. In: Teh YW, Titterington M (eds) Proceedings of the
thirteenth international conference on artificial intelligence and statistics. PMLR, Chia Laguna
Resort, Sardinia, pp 493–500
29. Ma J, Saul LK, Savage S, Voelker GM (2009) Identifying suspicious URLs: an application
of large-scale online learning. In: Proceedings of the 26th annual international conference on
machine learning. Association for Computing Machinery, New York, NY, pp 681–688. https://
doi.org/10.1145/1553374.1553462
30. Miettinen M, Marchal S, Hafeez I, Frassetto T, Asokan N, Sadeghi A-R, Tarkoma S (2017)
IoT sentinel demo: automated device-type identification for security enforcement in IoT. In:
Lee K, Liu L (eds) 2017 IEEE 37th international conference on distributed computing systems
(ICDCS). IEEE, pp 2511–2514. https://doi.org/10.1109/ICDCS.2017.284
31. Koroniotis N, Moustafa N (2020) Enhancing network forensics with particle swarm and deep
learning: the particle deep framework. CoRR. abs/2005.00722
32. Koroniotis N, Moustafa N, Sitnikova E (2020) A new network forensic framework based on
deep learning for internet of things networks: a particle deep framework. Future Gener Comput
Syst 110:91–106. https://doi.org/10.1016/j.future.2020.03.042
33. Moustafa N, Slay J (2016) The evaluation of network anomaly detection systems: statistical
analysis of the UNSW-NB15 data set and the comparison with the KDD99 data set. Inf Secur
J: Glob Perspect 25:18–31. https://doi.org/10.1080/19393555.2015.1125974
34. Moustafa N, Creech G, Slay J (2017) Big data analytics for intrusion detection system: statistical
decision-making using finite Dirichlet mixture models. In: Iván PC, Kalutarage HK (eds)
Data analytics and decision support for cybersecurity: trends, methodologies and applications.
Springer International Publishing, Cham, pp 127–156. https://doi.org/10.1007/978-3-319-594
39-2_5
468 K. Kaushik et al.

35. He Y, Luo C, Camacho RS, Wang K, Zhang H (2020) AI-based security attack pathway for
cardiac medical diagnosis systems (CMDS). In: 2020 computing in cardiology, pp 1–4. https://
doi.org/10.22489/CinC.2020.439
36. Lorenzo P, Stefano F, Ferreira A, Carolina P (2022) Artificial intelligence and cybersecurity:
technology, governance and policy challenges. Centre for European Policy Studies (CEPS)
37. Gerke S, Minssen T, Cohen G (2020) Chapter 12—Ethical and legal challenges of artifi-
cial intelligence-driven healthcare. In: Bohr A, Memarzadeh K (eds) Artificial intelligence
in healthcare. Academic Press, pp 295–336. https://doi.org/10.1016/B978-0-12-818438-7.000
12-5
38. He Y, Efpraxia DZ, Yevseyeva I, Luo C (2023) Artificial intelligence-based ethical hacking for
health information systems: simulation study. J Med Internet Res 25:e41748. https://doi.org/
10.2196/41748
39. Cobianchi L, Verde JM, Loftus TJ, Piccolo D, Dal Mas F, Mascagni P, Garcia Vazquez A,
Ansaloni L, Marseglia GR, Massaro M, Gallix B, Padoy N, Peter A, Kaafarani HM (2022)
Artificial intelligence and surgery: ethical dilemmas and open issues. J Am Coll Surg 235
40. Sarker IH, Janicke H, Mohammad N, Watters P, Nepal S (2023) AI potentiality and awareness:
a position paper from the perspective of human-AI teaming in cybersecurity
41. Flechais I, Chalhoub G (2023) Practical cybersecurity ethics: mapping CyBOK to ethical
concerns
42. Jackson D, Matei SA, Bertino E (2023) Artificial intelligence ethics education in cybersecurity:
challenges and opportunities: a focus group report
43. Ramya P, Babu SV, Venkatesan G (2023) Advancing cybersecurity with explainable artificial
intelligence: a review of the latest research. In: 2023 5th international conference on inventive
research in computing applications (ICIRCA), pp 1351–1357. https://doi.org/10.1109/ICIRCA
57980.2023.10220797
44. Mohamed N (2023) Current trends in AI and ML for cybersecurity: a state-of-the-art survey.
Cogent Eng 10:2272358. https://doi.org/10.1080/23311916.2023.2272358
45. Kumar S, Gupta U, Singh AK, Singh AK (2023) Artificial intelligence: revolutionizing cyber
security in the digital era. J Comput Mech Manag 2:31–42. https://doi.org/10.57159/gadl.jcmm.
2.3.23064
46. Santosh KC, Wall C (2022) AI and ethical issues. In: AI, ethical issues and explainability—
applied biometrics. Springer Nature Singapore, Singapore, pp 1–20. https://doi.org/10.1007/
978-981-19-3935-8_1
47. Alawida M, Mejri S, Mehmood A, Chikhaoui B, Isaac Abiodun O (2023) A comprehensive
study of ChatGPT: advancements, limitations, and ethical considerations in natural language
processing and cybersecurity. Information 14. https://doi.org/10.3390/info14080462
48. Kuzlu M, Fair C, Guler O (2021) Role of artificial intelligence in the internet of things (IoT)
cybersecurity. Discov Internet Things 1:7. https://doi.org/10.1007/s43926-020-00001-4
49. Parul I, Thakur S (2021) Ethics and artificial intelligence: the pandora’s box. In: Parul I, Thakur
S (eds) Artificial intelligence and ophthalmology: perks, perils and pitfalls. Springer Singapore,
Singapore, pp 145–150. https://doi.org/10.1007/978-981-16-0634-2_11
50. Ashraf J, Keshk M, Moustafa N, Abdel-Basset M, Khurshid H, Bakhshi AD, Mostafa RR (2021)
IoTBoT-IDS: a novel statistical learning-enabled botnet detection framework for protecting
networks of smart cities. Sustain Cities Soc 72:103041. https://doi.org/10.1016/j.scs.2021.
103041
51. Kalla D, Kuraku S (2023) Advantages, disadvantages and risks associated with ChatGPT and
AI on cybersecurity. JETIR
52. Li F, Ruijs N, Lu Y (2023) Ethics & AI: a systematic review on ethical concerns and related
strategies for designing with AI in healthcare. AI 4:28–53. https://doi.org/10.3390/ai4010003
53. Shasha Y, Carroll F (2021) Implications of AI in national security: understanding the security
issues and ethical challenges. In: Reza M, Jahankhani H (eds) Artificial intelligence in cyber
security: impact and implications: security challenges, technical and ethical issues, forensic
investigative challenges. Springer International Publishing, Cham, pp 157–175. https://doi.org/
10.1007/978-3-030-88040-8_6
Ethical Considerations in AI-Based Cybersecurity 469

54. Vakkuri V, Kemell KK, Abrahamsson P (2019) Implementing ethics in AI: initial results of an
industrial multiple case study. In: Xavier F, Männistö T (eds) Product-focused software process
improvement. Springer International Publishing, Cham, pp. 331–338
55. Dasawat SS, Sharma S (2023) Cyber security integration with smart new age sustainable startup
business, risk management, automation and scaling system for entrepreneurs: an artificial intel-
ligence approach. In: 2023 7th international conference on intelligent computing and control
systems (ICICCS), pp 1357–1363. https://doi.org/10.1109/ICICCS56967.2023.10142779
56. Kaloudi N, Li J (2020) The AI-based cyber threat landscape: a survey. ACM Comput Surv 53.
https://doi.org/10.1145/3372823
57. Du S, Xie C (2021) Paradoxes of artificial intelligence in consumer markets: ethical challenges
and opportunities. J Bus Res 129:961–974. https://doi.org/10.1016/j.jbusres.2020.08.024
58. Sarker IH, Furhad MH, Nowrozy R (2021) AI-driven cybersecurity: an overview, security
intelligence modeling and research directions. SN Comput Sci 2:173. https://doi.org/10.1007/
s42979-021-00557-0
59. Yildirim M (2021) Artificial intelligence-based solutions for cyber security problems. In:
Luhach AK, Elçi A (eds) Artificial intelligence paradigms for smart cyber-physical systems.
IGI Global, Hershey, PA, pp 68–86. https://doi.org/10.4018/978-1-7998-5101-1.ch004
60. Buzzanell PM (2023) Risk, resilience, and ethical considerations in artificial intelligence.
Emerg Media 1:30–39. https://doi.org/10.1177/27523543231188274
61. Koroniotis N, Moustafa N, Schiliro F, Gauravaram P, Janicke H (2020) A holistic review of
cybersecurity and reliability perspectives in smart airports. IEEE Access 8:209802–209834.
https://doi.org/10.1109/ACCESS.2020.3036728
62. Guleria A, Krishan K, Sharma V, Kanchan T. ChatGPT: forensic, legal, and ethical issues. Med
Sci Law 00258024231191829. https://doi.org/10.1177/00258024231191829
63. Morovat K, Panda B (2020) A survey of artificial intelligence in cyber-security. In: 2020
International conference on computational science and computational intelligence (CSCI), pp
109–115. https://doi.org/10.1109/CSCI51800.2020.00026
64. Garcia AB, Babiceanu RF, Seker R (2021) Artificial intelligence and machine learning
approaches for aviation cybersecurity: an overview. In: 2021 integrated communications navi-
gation and surveillance conference (ICNS), pp 1–8. https://doi.org/10.1109/ICNS52807.2021.
9441594
65. Sharma I (2021) Evolution of unmanned aerial vehicles (UAVs) with machine learning. In: 2021
international conference on advances in technology, management & education (ICATME), pp
25–30. https://doi.org/10.1109/ICATME50232.2021.9732774
66. Ahmad OF, Stoyanov D, Lovat LB (2020) Barriers and pitfalls for artificial intelligence in
gastroenterology: ethical and regulatory issues. Tech Innov Gastrointest Endosc 22:80–84.
https://doi.org/10.1016/j.tgie.2019.150636
67. Stahl BC (2021) Ethical issues of AI. In: artificial intelligence for a better future: an ecosystem
perspective on the ethics of AI and emerging digital technologies. Springer International
Publishing, Cham, pp 35–53. https://doi.org/10.1007/978-3-030-69978-9_4
68. Soni VD. Challenges and solution for artificial intelligence in cybersecurity of the USA. Artif
Intell Cybersecur USA 2
69. Kent AD. Cyber security data sources for dynamic network research. In: Dynamic networks
and cyber-security, pp 37–65. https://doi.org/10.1142/9781786340757_0002
70. Sarhan M, Layeghy S, Moustafa N, Portmann M (2020) NetFlow datasets for machine learning-
based network intrusion detection systems. CoRR. abs/2011.09144
71. Sharma I, Kaushik K. Chhabra G (2023) Augmenting transparency and reliability for national
health insurance scheme with distributed ledger. In: 2023 4th international conference on
electronics and sustainable communication systems (ICESC), pp 1399–1405. https://doi.org/
10.1109/ICESC57686.2023.10193127
72. Nour M, Slay J (2018) A network forensic scheme using correntropy-variation for attack
detection. In: Gilbert P, Shenoi S (eds) Advances in digital forensics XIV. Springer International
Publishing, Cham, pp 225–239
470 K. Kaushik et al.

73. Al-Hawawreh M, Moustafa N, Garg S, Hossain MS (2021) Deep learning-enabled threat intel-
ligence scheme in the internet of things networks. IEEE Trans Netw Sci Eng 8:2968–2981.
https://doi.org/10.1109/TNSE.2020.3032415
74. Miettinen M, Marchal S, Hafeez I, Asokan N, Sadeghi A-R, Tarkoma S (2017) IoT SENTINEL:
automated device-type identification for security enforcement in IoT. In: Lee K, Liu L (eds)
2017 IEEE 37th international conference on distributed computing systems (ICDCS). IEEE,
pp 2177–2184. https://doi.org/10.1109/ICDCS.2017.283
75. Keshk M, Moustafa N, Sitnikova E, Creech G (2017) Privacy preservation intrusion detection
technique for SCADA systems. In: 2017 military communications and information systems
conference (MilCIS), pp 1–6. https://doi.org/10.1109/MilCIS.2017.8190422
76. Sharma I, Ramkumar KR (2017) A survey on ACO based multipath routing algorithms for ad
hoc networks. Int J Pervasive Comput Commun 13:370–385. https://doi.org/10.1108/IJPCC-
D-17-00015
77. Mamun MS, Rathore MA, Lashkari AH, Stakhanova N, Ghorbani AA (2016) Detecting mali-
cious URLs using lexical analysis. In: Chen J, Piuri V (eds) Network and system security.
Springer International Publishing, Cham, pp 467–482
78. Moustafa N, Misra G, Slay J (2021) Generalized outlier Gaussian mixture technique based
on automated association features for simulating and detecting web application attacks. IEEE
Trans Sustain Comput 6:245–256. https://doi.org/10.1109/TSUSC.2018.2808430
79. Ashraf J, Bakhshi AD, Moustafa N, Khurshid H, Javed A, Beheshti A (2021) Novel deep
learning-enabled LSTM autoencoder architecture for discovering anomalous events from intel-
ligent transportation systems. IEEE Trans Intell Transp Syst 22:4507–4518. https://doi.org/10.
1109/TITS.2020.3017882
80. Koroniotis N, Moustafa N, Sitnikova E, Slay J (2018) Towards developing network forensic
mechanism for botnet activities in the IoT based on machine learning techniques. In: Hu J,
Khalil I (eds) Mobile networks and management. Springer International Publishing, Cham, pp
30–44
81. Moustafa N, Keshk M, Choo K-KR, Lynar T, Camtepe S, Whitty M (2021) DAD: a distributed
anomaly detection system using ensemble one-class statistical learning in edge networks.
Future Gener Comput Syst 118:240–251. https://doi.org/10.1016/j.future.2021.01.011
82. García S, Grill M, Stiborek J, Zunino A (2014) An empirical comparison of botnet detection
methods. Comput Secur 45:100–123. https://doi.org/10.1016/j.cose.2014.05.011
83. Haider W, Moustafa N, Keshk M, Fernandez A, Choo K-KR, Wahab A (2020) FGMC-HADS:
fuzzy Gaussian mixture-based correntropy models for detecting zero-day attacks from Linux
systems. Comput Secur 96:101906. https://doi.org/10.1016/j.cose.2020.101906
84. Weinger B, Kim J, Sim A, Nakashima M, Moustafa N, Wu KJ (2020) Enhancing IoT anomaly
detection performance for federated learning. In: 2020 16th international conference on
mobility, sensing and networking (MSN), pp 206–213. https://doi.org/10.1109/MSN50589.
2020.00045