Early Prediction of Type 2 Debites Using Non-invasive Lifestyle Factors and Machine Learning

Type 2 diabetes mellitus (T2DM) remains one of the most pressing global health challenges, primarily influenced by sedentary behavior, poor dietary habits, and other modifiable lifestyle factors. Early detection of individuals at high risk is vital for timely intervention and effective prevention strategies. In this study, a machine learning–based framework is proposed to predict the likelihood of developing T2DM using only non-invasive, lifestyle-related features such as body mass index (BMI), physical activity, diet, stress levels, sleep duration, hydration, and other behavioral indicators. Unlike conventional approaches that depend on invasive biomarkers such as blood glucose or HbA1c, the proposed model leverages easily obtainable data, making it suitable for large-scale, cost-efficient screening. Multiple algorithms were evaluated, including Linear Regression, Decision Tree, Random Forest, and Gradient Boosting. Among these, the Gradient Boosting model demonstrated superior performance, achieving a mean squared error (MSE) of 11.02 and an R² score of 0.94. Feature importance analysis further revealed that BMI, medical adherence, and physical activity were the most significant contributors to diabetes risk prediction. The findings suggest that integrating non-invasive lifestyle data with advanced machine learning models can serve as an effective approach for predicting T2DM risk. This framework shows potential for deployment within digital health platforms to enhance preventive care and promote early intervention.