Predicting Diabetes Using Machine Learning: A Comparative Study of Logistic Regression, k-NN, and SVM

Abstract

Diabetes mellitus is a common and persistent chronic metabolic disease. Accurate diagnosis is vital to prevent or slow severe complications such as renal failure and cardiovascular diseases. This study applies modern machine learning techniques to predict diabetes using the Pima Indians Diabetes Database. A comparative analysis of three classification algorithms—Logistic Regression, k-Nearest Neighbors, and Support Vector Machines—is conducted to evaluate their predictive performance. Each model is assessed in terms of accuracy, recall, and AUC-ROC metrics. Results show that all three algorithms perform well and hold promise for clinical application. Among them, Logistic Regression achieves the best performance, with 78% accuracy, a recall score of 0.82, and an AUC-ROC value of 0.84, indicating its strong ability to identify true positive cases while minimizing false negatives. These findings demonstrate the potential of machine learning to enhance diagnostic precision and support clinical decision-making in diabetes management. The study highlights the broader role of artificial intelligence in healthcare, emphasizing its capacity to reduce system burdens, optimize resources, and improve patient outcomes. By leveraging such technologies, healthcare systems can move toward more efficient and timely interventions, ultimately transforming patient care.