Healthcare Analytics xxx (xxxx) xxx

Contents lists available at ScienceDirect

Healthcare Analytics
journal homepage: www.elsevier.com/locate/health

Artificial intelligence and diagnostic healthcare using computer vision and medical imaging

The application of artificial intelligence (AI) in healthcare has The third paper by Talukdar et al. titled “A comparative assessment
garnered substantial attention due to its potential to transform various of most widely used machine learning classifiers for analyzing and
aspects of medical practice. Computer vision, a subfield of AI, enables classifying autism spectrum disorder in toddlers and adolescents,” pro­
computers to interpret and process visual information from the world, vides a thorough evaluation of various machine learning (ML) algo­
similar to how humans do. When combined with medical imaging, rithms for the diagnosis of autism spectrum disorder (ASD). ASD is a
computer vision algorithms can analyze these images to detect anoma­ developmental disorder that affects communication and behavior, with
lies, diagnose diseases, and even accurately predict patient outcomes. early diagnosis being crucial for effective intervention and support. Each
This technological synergy has opened new avenues for early diagnosis, classifier is assessed based on accuracy, precision, recall, F1 score, and
personalized treatment plans, and enhanced patient care, especially in computational efficiency. The authors employ rigorous cross-validation
areas with limited access to specialized medical professionals. This techniques to ensure the robustness and reliability of their results. The
editorial provides an overview of the latest research contributions in this findings reveal that while all classifiers offer valuable insights, specific
field, focusing on innovative methodologies and applications of AI in models, such as Random Forest and Gradient Boosting Machines,
diagnostic healthcare. The selected papers demonstrate the diverse ways demonstrate superior performance in terms of accuracy and robustness.
AI is being integrated into medical practice, showcasing its potential to Neural Networks also show promise due to their ability to capture
revolutionize the healthcare landscape. complex patterns in the data, although they require more computational
The first paper by Kiran et al. titled “A Gradient Boosted Decision resources.
Tree with Binary Spotted Hyena Optimizer for cardiovascular disease The fourth paper by Wankhade et al. titled “A novel hybrid deep
detection and classification,” presents an innovative approach to learning method for early detection of lung cancer using neural net­
improving the accuracy and efficiency of cardiovascular disease di­ works,” introduces an innovative approach that leverages the power of
agnostics. The authors propose a hybrid model that combines the hybrid deep learning techniques to improve the early detection of lung
Gradient Boosted Decision Tree (GBDT) with the Binary Spotted Hyena cancer. Early detection is crucial for improving lung cancer patient’s
Optimizer (BSHO), leveraging the strengths of both techniques to prognosis and survival rates, and this research focuses on enhancing
enhance performance. The study demonstrates that when applied to diagnostic accuracy through advanced neural network architectures.
cardiovascular disease datasets, the GBDT-BSHO model outperforms The study demonstrates that this hybrid model significantly outperforms
traditional GBDT and other standard classifiers regarding accuracy, traditional methods and standalone neural network models. By inte­
precision, recall, and F1-score. This hybrid approach improves diag­ grating deep learning, the model benefits from the strengths of both
nostic accuracy and reduces computational time, making it a viable tool architectures, leading to improved accuracy, sensitivity, and specificity
for real-world medical applications. The authors highlight the potential in detecting early-stage lung cancer. The authors validate their approach
of their model to assist healthcare professionals in making more accurate using a comprehensive dataset of lung images, employing rigorous
and timely decisions, ultimately improving patient outcomes. cross-validation techniques to ensure the robustness of their findings.
The second paper by Phridviraj et al. titled “A bi-directional Long The fifth paper by Gudadhe et al. titled “A novel machine learning-
Short-Term Memory-based Diabetic Retinopathy detection model using based feature extraction method for classifying intracranial hemor­
retinal fundus images,” introduces a cutting-edge method for detecting rhage computed tomography images,” presents an advanced approach
diabetic retinopathy (DR) leveraging the capabilities of bi-directional to improving the accuracy of intracranial hemorrhage (ICH) detection
Long Short-Term Memory (Bi-LSTM) networks. Diabetic retinopathy is and classification using computed tomography (CT) images. Intracranial
a severe complication of diabetes that affects the eyes and can lead to hemorrhage is a critical condition requiring rapid diagnosis to admin­
blindness if not detected and treated early. Accurate and early detection ister timely and potentially life-saving treatments. The authors employ
is crucial for preventing vision loss and managing the condition effec­ advanced image processing techniques and ML algorithms to extract
tively. The research demonstrates that the Bi-LSTM model outperforms features that capture the essential characteristics of different types of
traditional machine and deep learning models in detecting diabetic hemorrhages. These features include texture, shape, intensity, and
retinopathy. The model’s superior performance is attributed to its ability spatial relationships within the CT images. By leveraging these
to effectively handle the complex and sequential nature of the retinal comprehensive and discriminative features, the ML models can better
fundus images. The authors validate their approach using a compre­ distinguish between different ICH types. This innovative feature
hensive dataset, showing significant improvements in accuracy, sensi­ extraction method enhances the performance of existing classifiers and
tivity, and specificity metrics. offers a robust framework that can be adapted to other medical imaging

https://doi.org/10.1016/j.health.2024.100352

Available online 25 June 2024

2772-4425/© 2024 The Author(s). Published by Elsevier Inc. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).

Please cite this article as: Gaurav Dhiman et al., Healthcare Analytics, https://doi.org/10.1016/j.health.2024.100352
G. Dhiman et al. Healthcare Analytics xxx (xxxx) xxx

applications. The study’s findings suggest that incorporating sophisti­ intelligent system integrates cutting-edge technologies, prominently
cated ML-based feature extraction techniques can lead to more reliable featuring a fuzzy-based inference engine designed to enhance decision-
and precise diagnostic tools, aiding radiologists in making quicker and making within the surgical environment. The system incorporates a tele-
more accurate decisions. robotic surgical platform that facilitates seamless interaction between
The sixth paper by Koyyada et al. titled “An explainable artificial surgeons and robotic instruments. This platform ensures that surgical
intelligence model for identifying local indicators and detecting lung actions performed remotely accurately reflect the surgeon’s intent,
disease from chest X-ray images,” explores the development of an AI maintaining crucial levels of precision and control throughout the pro­
model that achieves high diagnostic accuracy and provides explain­ cedure. The fuzzy-based inference engine plays a pivotal role by
ability for its decisions. Explainability is crucial in medical AI applica­ employing fuzzy logic to process and interpret many inputs, including
tions, as it builds trust among healthcare professionals and ensures that surgeon commands, patient data, and real-time feedback from the sur­
the AI’s recommendations are transparent and understandable. The gical site. This approach allows the system to make nuanced and
research involves training the model on a large dataset of labeled chest context-aware decisions, adapting dynamically to the complexities
X-ray images, followed by rigorous testing to evaluate its performance. inherent in surgical interventions.
The results indicate that the model achieves high accuracy, sensitivity, In conclusion, the integration of AI with diagnostic healthcare is not
and specificity in detecting lung diseases. More importantly, the just a technological advancement but a paradigm shift that has the po­
explainability features provided by Grad-CAM help validate the AI’s tential to save countless lives. The research presented in this issue of
decisions, making it easier for healthcare professionals to trust and Healthcare Analytics is a testament to the remarkable progress being
adopt the technology in clinical practice. made in this field. We extend our deepest gratitude to Prof. Madjid
The seventh paper by Shinde et al. titled “A Multi-Classifier-Based Tavana, the Editor-in-Chief, for his unwavering dedication and support
Recommender System for Early Autism Spectrum Disorder Detection in bringing these pioneering studies to light.
using Machine Learning,” presents an innovative approach to enhancing
the early detection of Autism Spectrum Disorder (ASD). Early diagnosis Gaurav Dhiman*
of ASD is crucial for timely intervention and support, significantly Centre of Research Impact and Outcome, Chitkara University Institute of
improving the developmental outcomes for affected children. The Engineering and Technology, Chitkara University, Punjab, India
recommender system operates by aggregating the predictions from each MEU Research Unit, Middle East University, Amman, Jordan
classifier and using a voting mechanism or ensemble learning technique Department of Electrical and Computer Engineering, Lebanese American
to make the final decision. This multi-classifier approach ensures that University, Byblos, Lebanon
the system can capture a wide range of patterns and indicators of ASD,
Wattana Viriyasitavat
thereby enhancing the robustness of the diagnostic process. One of the
Business Information Technology Division, Department of Statistics, Faculty
key advantages of Shinde et al.’s approach is its ability to recommend
of Commerce and Accountancy, Chulalongkorn University, Bangkok,
further actions based on the classification results. For instance, if a high
Thailand
risk of ASD is detected, the system can suggest specific follow-up as­
sessments or interventions. This feature makes the system a diagnostic Atulya K. Nagar
tool and a valuable support mechanism for healthcare professionals and Liverpool Hope University, Liverpool, UK
parents.
Oscar Castillo
In the eighth paper, titled “An intelligent system with fuzzy-based
Tijuana Institute Technology, Tijuana, Mexico
inference engine for secured tele-robotic surgery,” Patil et al. present a
pioneering approach to advancing the precision and security of tele- *
robotic surgeries. Tele-robotic surgery, which enables remote surgical Corresponding author.
procedures using robotic systems, demands meticulous control and E-mail address: [email protected] (G. Dhiman).
safety measures to ensure optimal patient outcomes. Patil et al.’s