Machine Learning based Screening for Diabetes Risk Prediction

Diabetes Mellitus (DM) poses a major global health issue, impacting millions and significantly affecting quality of life. Early detection and risk prediction are crucial, but conventional diagnostic methods are often costly and limited by the availability of clinical markers. Machine Learning (ML) approaches, however, have shown higher predictive accuracy, reaching 77.9%, compared to traditional methods. This study examines ML techniques, such as Support Vector Machines (SVM) and Logistic Regression (LR), applied to datasets encompassing lifestyle, clinical, and demographic factors. Additionally, it explores the potential of integrating ML with wearables and electronic health records (EHR) to enable personalized, real-time risk assessments. These ML models consistently demonstrate improved accuracy, sensitivity, and specificity in predicting diabetes risk, with an accuracy rate of 77.9%, exceeding that of conventional diagnostic tools.