Int. J. Pharm. Sci. Rev. Res., 81(1), July – August 2023; Article No.

27, Pages: 159-164 ISSN 0976 – 044X

Review Article

Applications of Artificial Intelligence in Pharmaceutical Industries

Hansa Choudhary*1, Anta Sharma1, Dr. Dharmendra Ahuja2

1Student, Department of Pharmacology, Jayoti Vidyapeeth Women’s University, Jaipur, Rajasthan, 303122, India.
2Dean, Department of Pharmacology, Jayoti Vidyapeeth Women’s University, Jaipur, Rajasthan, 303122, India.
*Corresponding author’s E-mail: [email protected]

Received: 19-04-2023; Revised: 24-06-2023; Accepted: 02-07-2023; Published on: 15-07-2023.

ABSTRACT
Artificial intelligence (AI) is used more and more in different societal fields, particularly in the pharmaceutical area. AI is advantageous
from an economic and professional standpoint. By imitating human intelligence and combining it with cutting-edge technology to
provide the best results, it minimizes the need for human work. This review examines the useful application of AI in the
pharmaceutical industry at each stage from drug discovery through drug development as well as in several departments. In addition
to requiring less human labor, it also enhances clinical research effectiveness and pharmaceutical productivity while decreasing the
possibility of error. The review also looks at the current difficulties in implementing AI in the pharmaceutical industry. Additionally, it
assesses current difficulties and suggests solutions.
Keywords: Artificial Intelligence, Drug development, Machine Learning, Pharmaceutical Industry.

since it can manage massive data sets with improved

QUICK RESPONSE CODE →
automation. A modern technology-based system called
artificial intelligence (AI) uses advanced tools, networking,
DOI: and technologies to simulate human intelligence. It
10.47583/ijpsrr.2023.v81i01.027 consists of hardware and software which can draw
conclusions from the data provided and acts
DOI link: http://dx.doi.org/10.47583/ijpsrr.2023.v81i01.027
independently.
Computers are used in artificial intelligence to simulate
INTRODUCTION intelligent decision-making. The most essential subsets of
AI named as machine learning (ML), is commonly confused

T he use of AI has grown globally across many

industries. A review of its considerable impact on the
pharmaceutical business is pending. Artificial
intelligence (AI) is being used in a large numbers of
pharmaceutical disciplines like medication research and
with AI. The term "machine learning" refers to the capacity
to continuously statistically understand data without any
explicit programming3. The use of AI could significantly
change the speed and scope of drug discovery today. In
general, the incorporation of AI in the pharmaceutical area
development, medicine repurposing, accelerating can bring advancements such as:
pharmaceutical manufacturing, and clinical studies, which
minimizes the amount of labor needed from humans while • AI-based drug development doesn't rely on pre-
also advancing goals more quickly1. The difficulty of established targets. As a result, during the medication
gathering, examining, and applying that knowledge to development process, personal biases and previous
challenging healthcare scenarios has increased with experience are not taken into account.
greater digitalization2. This article examines the • AI develops innovative algorithms for drug discovery by
application of AI to several area of the pharmaceutical combining the most recent advances in biology and
industry, from discovery of drugs to medication computation. Due to the rapid increase in processing
development. It includes utilizing AI to conduct clinical power and decrease in processing costs, AI is having all
studies, repurposes drugs, increase pharmaceutical what it takes to level up the levels of competition in
output, and more. drug research. AI has a higher capacity for defining
Artificial Intelligence pertinent interactions in pharmaceutical screens.
Therefore, the likelihood of false positives can be
The usage of digitalization in the pharmaceutical industry reduced by carefully selecting the aforementioned test
has undergone significant development. The need for parameters.
examination and trustworthy knowledge to address issues
related to its acquisition and application in the field may • The capacity of AI to transform drug screening into a
be one of the hurdles that change always bring, virtual lab, where promising targets can be reduced
nevertheless. These difficulties drive the inclusion of AI without demanding a lot of experimental input or

International Journal of Pharmaceutical Sciences Review and Research

Available online at www.globalresearchonline.net 159
©Copyright protected. Unauthorised republication, reproduction, distribution, dissemination and copying of this document in whole or in part is strictly prohibited.
Int. J. Pharm. Sci. Rev. Res., 81(1), July – August 2023; Article No. 27, Pages: 159-164 ISSN 0976 – 044X

manpower hours, is perhaps its most important Logic

capability.
It can be used for expressing knowledge and solving
• The collection and storage of large amounts of digital problems, but it can also be used for other issues. For
patient data have become easier because of recent examples:
technological advancements. To help speed up the
• The planning process in the SatPlan algorithm
creation of novel medications and improve their
chances of success in the early stages of development, • Probabilistic approaches to reasoning under
AI platforms may mine the massive libraries of genomic uncertainty
information, health records, medical imaging, and
other patient data4. • Case-based reasoning,

Machine reasoning (MR) and AI • AI research

Machine reasoning (MR) is a fundamental subcategory of • Logic of propositions

AI. MR makes use of learned information to navigate • uses truth-related operations like "or" and "not"
through a variety of options and get the optimal answer.
To address some problems, deep subject knowledge is • Primate logic
required, and MR is perfectly suited for this. A machine • Predicates,
reasoner needs explicit human capture of all prior
information to function on fresh input. MR is an excellent • Fuzzy,
supplement to ML since it may analyze potential causes • Default Logics,
and improvement options in addition to extending the
discoveries provided by ML5. • Non-Monotonic Logics,
Management of networks using AI and ML • Propositional Logic,
Many challenges in network administration call for an • Adds Quantifiers,
immediate response. To solve these urgent issues, AI is
• Fuzzy,
being applied in four different areas of network
administration. As follows: • Non-Monotonic Logics,
• Traffic control, • Circumscription
• Performance evaluation, • Non-monotonic logics,
• Capacity planning • Default reasoning,
• Security surveillance • The frame problem,
Of all the AI techniques, ML is proven to be the most • Circumscription,
helpful in solving issues in the majority of these
situations. AI is sophisticated enough to consider a wide • The closed world assumption,
range of potential fixes for several problems 5,6. Examples • Abduction
include:
• The logic of description
• Random optimization,
• Displaying classes and relations7
• Beam search,
Networks of artificial neurons
• Simulated annealing,
It was inspired from the structure of neuron in the human
• Blind hill climbing, and brain. Simple "neuron" N gets information via different
• Optimization searches "neurons," every one of which casts a weighted vote for or
against whether neuron N should activate itself while
• Calculations based on evolution active (or "fired"). A learning algorithm must be used to
• Gene expression programming, alter these weights in accordance with the training data7.
Modern neural networks search for patterns in data and
• Genetic programming, represent intricate relationships between inputs and
results. They are capable of learning both digital logical
• Genetic algorithms
processes and continuous functions. Since neural networks
Particle swarm optimization was inspired by the behavior employ gradient descent to obtain access to a multiple
of flocks of animals7,8,9 dimensions topology that was constructed during training,
they may be seen as a type of mathematical optimisation.

International Journal of Pharmaceutical Sciences Review and Research

Available online at www.globalresearchonline.net 160
©Copyright protected. Unauthorised republication, reproduction, distribution, dissemination and copying of this document in whole or in part is strictly prohibited.
Int. J. Pharm. Sci. Rev. Res., 81(1), July – August 2023; Article No. 27, Pages: 159-164 ISSN 0976 – 044X

The backpropagation algorithm is the most widely used b. Screening of compounds and lead optimization
training method.
To select drug candidates, compound screening and lead
In-depth learning optimisation procedures combine virtual screening, high
throughput screening, and combinatorial chemistry. To
The different layers can gradually pul upgraded
build the chemical database for AI-based virtual screening,
information from the initial input. Lower layers in image
significant amounts of compounds are retrieved from
processing, for instance, may recognize borders, whereas
openly accessible chemogenomics libraries, which include
higher layers may recognize items that are significant to
millions of compounds with structural information.
humans, such as numbers, characters, and
faces connecting the network's inputs and outputs using c. Preclinical research
an array of layers of neurons 9,10.
Preclinical studies, sometimes referred to as non-clinical
AI Application in the Pharmaceutical Industry studies, are laboratory tests conducted in vitro and in vivo
on novel medicinal substances to ascertain their safety and
AI in Discovery of Drug
efficacy profile. To quickly collect relevant huge amounts
Drug development and research is a costly and laborious of biological data, clustering-based ML methods assess
endeavor that necessitates balancing numerous features RNA sequencing technologies to discover a molecular
of molecules to achieve desirable clinical efficacy and mechanism of action. It also exposes several previously
safety. Among these characteristics, the drugs' unknown relationships between different stimuli and the
performance primarily depends on their pharmacokinetics cytokines they affect.
/pharmacodynamics (PK/PD) characteristics, nonclinical
d. Clinical Research
safety profiles, and absorption, distribution, metabolism,
and elimination (ADME) characteristics. The discovery of For recognizing patient conditions, finding gene targets,
drug and the process of its development may benefit from forecasting the outcome of a created drug, and on and off-
recent advancements in AI/ML techniques since they could targets, a generation of AI tools for clinical trials would
increase the likelihood of success and speed up decision- prove excellent. In Phase II clinical studies, one AI mobile
making. Huge datasets that contain information about application boosted medicine adherence by 25%
chemical structures and the associated experimental compared to conventional direct observation therapy 12.
readouts are used to train the AI/ML system, which
AI IN DRUG SCREENING
subsequently clarifies the underlying ideas, biological
reactions, and interactions between chemical structures. The use of AI in drug screening includes target
The AI/ML models can then be used to predict the identification; molecular simulations; drug property
anticipated biological responses of a molecule that has not predictions; de novo drug creation, synthesis of route
yet been tested and, in some circumstances, hasn't even generation, and candidate drug prioritization12.
been synthesized, only based on its chemical structure. To
AI-based drug discovery tools and resources
choose tiny compounds with the most potential ADME
profiles for synthesis, testing, and process development, Data resources are one of the fundamental elements of
discovery scientists in the ADME sector employ these using AI for drug discovery and evaluation. Examples of
AI/ML models. These compounds are chosen from data resources are ChEMBL, ChemDB, COCONUT, DGIdb,
hundreds of thousands of theoretically possible options11. DrugBank, DTC, INPUT, PubChem, and SIDER or STITCH.
Common AI methods include Delta Vina, neural graph
There are several uses of AI in drug discovery, and they can
fingerprint, alpha fold, and computer and additional
be categorized as follows:
techniques include clustering, classification, and
a. Target validation and choice regression analysis techniques13,14.
Discovering the role of putative drug targets AI in drug molecule design
(genes/proteins of a small molecule) and their importance
Traditionally, designing a drug molecule requires a lot of
in illness requires understanding how these molecules
effort and entails multiple failed attempts. For example, a
work. The selection and validation methods must take into
synthesis plan can have hundreds of distinct steps, many
account functioning genomics, structure genomics,
of which would produce unfavorable byproducts or side
proteomics, in-vitro cell-based assays, and in-vivo animal
effects or will just not work. AI, on the other hand, is now
research assays. AI evaluates the Drug Information Bank to
starting to enhance the efficiency of design, and synthesis,
ascertain the beneficial effects of drug candidates, genetic
speeding up, streamlining, and lowering the cost of
expressions, interactions between proteins, and clinical
company operations while also reducing chemical waste15.
data records from a public library. To ascertain a drug's
In recent years, theoretical studies and research on AI (AI)
efficacy target, it is required to understand the function of
have given way to actual implementations. Thanks to
potential molecular targets and their significance in illness.
advancements in AI and ML techniques like deep learning
(DL), machines may now imitate the intelligence of humans
by learning from data. As a result, thanks to AI and ML

International Journal of Pharmaceutical Sciences Review and Research

Available online at www.globalresearchonline.net 161
©Copyright protected. Unauthorised republication, reproduction, distribution, dissemination and copying of this document in whole or in part is strictly prohibited.
Int. J. Pharm. Sci. Rev. Res., 81(1), July – August 2023; Article No. 27, Pages: 159-164 ISSN 0976 – 044X

approaches, molecular design may see a Process design and scalability: AI can quickly identify the
revolution. Before AI, planning tests and characterizing, processing parameters, reducing waste and development
evaluating, and analyzing the findings of such time.
investigations were almost completely the responsibility of
Advance Process Control (ACP): When paired with real-
specialists working in experimental laboratories that
time sensor data, AI can make it simpler to create process
developed new substances. This bottleneck has slowed the
controls and, as a result, foresee how a process will
creation of new materials, which could affect several
develop. AI tools can recognize different maintenance
sectors including electronics, medicinal development, and
approaches and equipment adjustments. This reduces
energy storage16. As a result of advancements in the field
downtime.
of computer science and the use of ML technology, many
tools are now available. The possibility for considerably Trend detection: AI can be used to examine variations in
fewer computationally demanding simulations with customer feedback and complaints18.
sufficient precision for chemical applications is shown in
Although AI in pharmaceutical manufacturing can speed
quantum chemistry simulation utilizing ML techniques. ML
up and improve process efficiency, it's also important to
applications include calculating stable structural
consider how it will impact the equipment and
conformations and producing coarse-grained models for
infrastructure that are already in place.
molecular dynamics simulations. Inspiring research results
have come from using ML to predict molecular AI for quality assurance and control
characteristics. ML algorithms are now routinely used to
Pharmaceutical quality assurance and control systems are
predict features including the protein's reactivity, toxicity,
expected to benefit from the use of AI technologies. Bands
and ability to bind to receptors from its electronic
are frequently used in the pharmaceutical industry to
structure and amino acid backbone. This approach can be
ensure that product quality requirements can be satisfied,
applied to chemistry to mimic the genesis of novel
although they aren't always as precise as they may be. The
synthetic architectures. The development of synthetic
applications of AI in the pharmaceutical industry can
molecular design with the aid of this technology could
predict even little variations that could have an impact on
considerably boost chemical research16.
the functionality and safety of the products.
AI is being used to develop new pharmaceutical products, Pharmaceutical quality control and assurance using AI can
including the following: improve product quality and process accuracy19. However,
the goods must go through manual quality control checks
• Disease Identification/Diagnosis
to maintain batch-to-batch homogeneity. Integration of AI
• Personalized treatment with digital therapeutics could not be a workable strategy at this time. The
and behavioral modification Electronic Lab Notebook, a platform for automated data
entry, along with cutting-edge AI algorithms, can assure
• Drug Development and Production product quality. It is feasible to adopt new technologies for
• Prognostic forecasting intelligent quality control by using overall quality
management expert systems, data mining, and other
• Medical Tests knowledge discovery techniques17.
AI FOR PHARMACEUTICAL PRODUCTION AI used in designing of Clinical Trials
Manufacturing complexity has increased from prior years Clinical trials are carried out to evaluate the efficacy and
as a result of recent improvements in pharmaceutical safety of drug items in humans for a particular disease
technology and distinctive developments. For the complex condition. It could take 6-7 years or longer to complete this
procedure, a proven product's excellence and laborious process, and a sizable sum of money will also be
effectiveness are also necessary. Modern pharmaceutical required. Furthermore, the success rates are unacceptably
production systems are striving to merge human expertise low, which puts a significant burden on the industry.
into machines to enhance manufacturing processes. The Failures may have a flawed patient selection strategy, poor
employment of AI in the production process has the infrastructure, or a lack of technical requirements as their
potential to expand the pharmaceutical industry. underlying cause. Integration of AI can reduce this errors17.
Numerous pharmaceutical activities will be mechanized AI can carry out all stages of clinical trial design, including
with the aid of AI tools. In addition to addressing problems study preparation and study execution. This may increase
that may occur during the manufacturing process, this can trial success rates while easing the strain of
aid in numerical stimulation17. AI can be included in the pharmaceutical R&D. To obtain ethical and benchmarked
manufacturing of medications to achieve the right level of results, the inclusion of AI in clinical trials must first be
outcomes by utilizing other cutting-edge manufacturing reviewed. The right strategy can reduce trial costs and
technologies. According to a discussion paper released by failure rates while also speeding up the process 20. AI can
the U.S. FDA, the following potential applications of AI help in the selection of the population with specific illness
were seen in pharmaceutical production processes. conditions for Phase II and Phase III studies by using the
patient-specific genome exposome profile analysis. The

International Journal of Pharmaceutical Sciences Review and Research

Available online at www.globalresearchonline.net 162
©Copyright protected. Unauthorised republication, reproduction, distribution, dissemination and copying of this document in whole or in part is strictly prohibited.
Int. J. Pharm. Sci. Rev. Res., 81(1), July – August 2023; Article No. 27, Pages: 159-164 ISSN 0976 – 044X

preclinical discovery of compounds and the identification 3. What is Artificial Intelligence in Networking?
of the lead medications before to the initiation of clinical [Internet] 2023 [Cited 2023 June 4] Available from:
trials can both benefit from predictive ML, among other AI https://www.cisco.com/c/en/us/solutions/artificial-
technologies.17. intelligence/artificial-intelligence-machine-learning-
in-networking.html#~q-a
Pharmaceutical marketing
4. Larranga P, Calvo B, Santana R, Bielza C, Galdiano J,
E-commerce and other marketing strategies are already
Inza I, Lozano JA, Armanazas R, Santafe G, Perez A,
utilizing AI in their successful marketing campaigns.
Robles V. Machine learning in bioinformatics.
Pharmaceutical companies can utilize comparable tools
Briefings in Bioinformatics. 2006;7(1):86-112.
and techniques to expand and effectively serve their
market. The tactics of AI in pharmaceutical marketing can 5. AI & ML in Network Management. [Internet] 2023
greatly increase overall earnings. [Cited 2023 June 4] Available from:
https://www.comparitech.com/net-admin/ai-ml-in-
Limitations of AI at the present moment
network-management/ (accessed June 3, 2023).
Despite all of the uses for AI in the pharmaceutical
6. Thakur A, Pokhriyal A, Gahlot D, Kukreti D. Human-
industry, there are some restrictions attached to it. To
Level Artificial Intelligence: One of the Future. IJRIT.
achieve the present standards of standard outcomes, AI
2014;2(9):373-9.
must be developed. The demand for thorough training
data for AI, which necessitates human labor and hence can 7. Froding B, Peterson M. Ethics and Information
have chances of inaccuracy, is one of the factors that Technology. Friendly AI. 2021;23:207-14.
indicate its limitations. As a result, AI integration is https://doi.org/10.1007/s10676-020-09556-w
unavoidable. Another drawback of AI is that it can forecast
8. Nilsson NJ. Artificial Intelligence: A New Synthesis,
models or structures via de novo medication creation,
Morgan Kaufmann Publishers 1998.
which is impossible to produce. A multidisciplinary
strategy is needed to prepare healthcare workers for the 9. Mirmozafffari M, Yazdani R, Shadkam E, Khalili SM,
use of AI because AI and ML are IT-based technologies, Mahjoob M, Boskabadi A. An integrated artificial
which may prove hard for the medical industry 21. intelligence model for efficiency assessment in
pharmaceutical companies during the COVID-19
CONCLUSION
pandemic. Sustainable Operations and Computers.
AI is increasingly integrating into our daily lives and 2022; 3:156-67.
becoming smarter with each passing day. This ensures its
10. Mishra V. Artificial Intelligence: the beginning of a
presence in every aspect of our existence. The use of AI in
new era in the pharmacy profession, Asian Journal of
the pharmaceutical industry has the potential to
Pharmaceutics. 2018;12(2):18-23.
revolutionize the industry, but its current state of
development needs improvement. Given the limitations of 11. Why Artificial Intelligence Could Speed Drug
current AI technology and the need for further research, it Discovery. [Internet] 2022 [Cited 2023 June 7]
is still too early to predict if AI will have a significant impact Available from:
on the pharmaceutical industry. The pharmaceutical https://www.morganstanley.com/ideas/ai-drug-
industry has the potential to gain from AI because it has discovery
the potential to speed up and lower the cost of drug
12. How Artificial Intelligence is Revolutionizing Drug
discovery while also improving clinical and pharmaceutical
Discovery. [Internet] 2023 [Cited 2023 June 7]
data through its cutting-edge technology. AI will benefit
Available from:
future generations, and pharmaceutical companies will
https://blog.petrieflom.law.harvard.edu/2023/03/2
also benefit from it. There are many obstacles and
0/how-artificial-intelligence-is-revolutionizing-drug-
challenges related to this technology, hence there are
discovery/
currently no drugs on the market that were created
utilizing AI approaches. Regardless of this, AI is likely to 13. Chen W, Liu X, Zhang S, Chen S. Artificial Intelligence
become a crucial tool for the pharmaceutical industry for drug discovery: Resources, methods, and
soon. applications. Molecular Therapy – Nucleic Acids.
2023;31:691-702.
REFERENCES
https://doi.org/10.1016%2Fj.omtn.2023.02.019
1. Baran S, Henstock P. Pathways for Successful AI
14. Sima J, lee J, Wijaya L, Choi K, Coutlakis A. A literature
Adoption in Drug Development. SSRN.
review on the implementation of CRISPR systems and
https://dx.doi.org/10.2139/ssrn.4305068
other biomedical tools on therapeutic interventions
2. Ramesh AN, Kambhampati C, Monson JR, Drew PJ. against tuberculosis. UC Berekeley Research
Artificial Intelligence in Medicine. Annals of the Royal Publications. 2022
College of Surgeons of England. 2004; 86(5):334-7.

International Journal of Pharmaceutical Sciences Review and Research

Available online at www.globalresearchonline.net 163
©Copyright protected. Unauthorised republication, reproduction, distribution, dissemination and copying of this document in whole or in part is strictly prohibited.
Int. J. Pharm. Sci. Rev. Res., 81(1), July – August 2023; Article No. 27, Pages: 159-164 ISSN 0976 – 044X

15. Venkatraman S, Miah M. Artificial Intelligence in 18. Using Artificial Intelligence & Machine Learning in the
Healthcare: A Potential Fame Changer. Global Journal Development of Drug & Biological Products. U.S.
of Business Disciplines. 2022;6(1):56-9. Food & Drug. 2023
16. Artificial Intelligence in molecular design. [Internet] 19. [Will AI Plan an Important Role in Quality Assurance
2023 [Cited 2023 June 7] Available from: and Quality Control Careers? [Internet] 2017 [Cited
https://insights.globalspec.com/article/19909/artific 2023 June 7] https://www.aaps.ca/blog/will-ai-play-
ial-intelligence-in-molecular- an-important-role-in-quality-assurance-and-quality-
design#:~:text=AI%20can%20be%20used%20in,mol control-careers
ecules%20and%20generate%20manufacturing%20p
20. Henstock PV. Artificial Intelligence for Pharma: Time
rocesses
for Internal Investment. Cell Press. 2019;40(8):543-6.
17. Mak KK, Pichika MR. Artificial intelligence in drug
21. Pareek V, Sharma L, Kumar S, Sharma V. Need for
development: present status and future prospects.
Artificial Intelligence in Pharmaceutical Industry and
Drug Discovery Today. 2019;24(3):773-780.
its Limitations. Journal of the Indian Academy of
https://doi.org/10.1016/j.drudis.2018.11.014
Geriatrics. 2022;18(14):217-20.

Source of Support: The author(s) received no financial support for the research, authorship, and/or publication of this article.
Conflict of Interest: The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or
publication of this article.

For any questions related to this article, please reach us at: [email protected]
New manuscripts for publication can be submitted at: [email protected] and [email protected]

International Journal of Pharmaceutical Sciences Review and Research

Available online at www.globalresearchonline.net 164
©Copyright protected. Unauthorised republication, reproduction, distribution, dissemination and copying of this document in whole or in part is strictly prohibited.