Integrated statistical modeling and machine learning techniques with SHAP for epidemiological data analysis

Epidemiological studies increasingly rely on advanced analytics to uncover complex relationships in health data. This study employs an innovative SHAP (SHapley Additive exPlanations)-driven framework to enhance the interpretability of machine learning models applied to a dataset of 1200 patients. Key features, including demographic, anthropometric, lifestyle, and clinical parameters, were analyzed using a Random Forest classifier integrated with SHAP values. Health outcomes, specifically the presence of chronic diseases such as diabetes, were predicted with high accuracy (85 %) and AUC (0.89), outperforming logistic regression (accuracy = 79 %, AUC = 0.84). SHAP values further highlighted influential predictors such as BMI and age, offering insights into individual contributions to health outcomes. By bridging traditional epidemiological analysis and modern machine learning techniques, this study offers a transparent and interpretable model for healthcare decision-making.