Multiple Disease Prediction System Using Machine Learning: A Comparative Study for Early Healthcare Decision Support
Multiple Disease Prediction System Using Machine Learning: A Comparative Study for Early Healthcare Decision Support
Authors:
Aradhy Umesh1
Supervised by - Dr. Megha Gupta (Associate Professor)2
1,2Department of Computer Science and Engineering Dr. Akhilesh Das Gupta Institute of Professional Studies
Delhi, India
Email: 1aradhyumesh@email.com,
Abstract—Timely identification of chronic and neurological conditions remains a persistent challenge in overburdened health-care systems, where specialist access is often limited. We describe the design and evaluation of a web-based Multiple Disease Prediction System (MDPS) capable of screening three clinically distinct conditions simultaneously: Diabetes Mellitus, Coronary Heart Disease, and Parkinson’s Disease. Rather than committing to one classifier upfront, we benchmark ten machine learning algorithms — spanning ensemble learners, kernel methods, and a feedforward neural network — against each disease’s unique feature space and class distribution. The highest-performing model per disease was then embedded into a Flask web ap-plication equipped with role-based access control and automated PDF report generation. Experimental results show that Random Forest, SVM with RBF kernel, and an Artificial Neural Network respectively yield the best trade-off between sensitivity and specificity for the three diseases. The work demonstrates that model selection must be disease-specific, and that production-grade deployment of such systems requires explicit attention to class imbalance, demographic bias, and clinical oversight protocols.
Keywords—Machine Learning, Disease Prediction, Healthcare Analytics, Decision Support System, Random Forest, Support Vector Machine, ROC-AUC, Cross-Validation, Neural Network, Comparative Analysis