A Comparative Study of Machine Learning Approaches for Diabetes Risk Prediction: Insights from SHAP and Feature Importance

This study leverages machine learning models, including Logistic Regression, XGBoost, and Random Forest, to predict diabetes using BRFSS data. XGBoost emerged as the topperforming model, achieving an AUC of 0.83. Feature importance was analyzed using SHAP values, highlighting general health, high blood pressure, and BMI as key predictors. This research demonstrates the potential of machine learning models in identifying individuals at risk of diabetes, aiding in early intervention and improved public health outcomes.