COMPARATIVE ANALYSIS OF LIME AND SHAP FOR EXPLAINABLE DIABETES PREDICTION

Diabetes mellitus is one of the most rapidly growing chronic diseases worldwide, affecting millions of individuals and creating a major burden on healthcare systems. Early detection and accurate risk prediction are essential to reduce complications such as cardiovascular disease, kidney failure, neuropathy, and vision loss. In recent years, machine learning techniques have significantly improved the accuracy of disease prediction models. However, many machine learning models operate as black boxes, making it difficult for medical professionals to understand how predictions are generated. This lack of interpretability reduces trust and limits the practical adoption of these models in clinical environments. To address this issue, explainable artificial intelligence (XAI) techniques have been introduced to provide transparency in model decisions. This study presents a comparative analysis of two widely used explanation methods, Local Interpretable Model-Agnostic Explanations (LIME) and SHapley Additive exPlanations (SHAP), applied to diabetes prediction. The proposed system uses the Pima Indians Diabetes Dataset and employs the Light Gradient Boosting Machine (LightGBM) algorithm for accurate prediction. Data preprocessing techniques such as handling missing values, normalization, and feature selection are applied to improve model performance. LIME and SHAP are integrated to provide both local and global explanations for predictions, highlighting how clinical factors such as glucose level, body mass index, age, and insulin influence outcomes. The system is implemented as a full-stack web application with a FastAPI backend and a Reactbased frontend to provide an interactive dashboard for healthcare professionals. Experimental results demonstrate that the model achieves high predictive performance while maintaining interpretability. The comparative analysis shows that SHAP provides more consistent global explanations, while LIME offers intuitive local explanations for individual predictions.