A Comparative Analysis of LIME and SHAP Interpreters With Explainable ML Based Diabetes Predictions

The escalating prevalence of diabetes necessitates accurate and interpretable predictive models to support early diagnosis and personalized treatment. This study presents a comparative analysis of the Local Interpretable Model-agnostic Explanations (LIME) interpreter integrated with machine learning algorithms for diabetes prediction. The existing system employs Logistic Regression (LR) combined with Explainable Artificial Intelligence (XAI) techniques to provide transparent predictions. In contrast, the proposed system leverages Random Forest and Decision Tree algorithms, integrated with XAI, to enhance predictive performance and interpretability. The study evaluates these models on a diabetes dataset, assessing metrics such as accuracy, precision, recall, F1-score, and the quality of LIME-generated explanations. Random Forest and Decision Tree models are hypothesized to outperform LR due to their ability to capture non-linear relationships and complex feature interactions. LIME facilitates interpretability by providing local explanations for individual predictions, enabling clinicians to understand model decisions. The comparative analysis highlights the trade-offs between model complexity, predictive accuracy, and interpretability, offering insights into the suitability of ensemble and tree-based models for diabetes prediction. The findings aim to guide the development of robust, transparent, and clinically viable predictive systems for diabetes management. Keywords: Diabetes Prediction, LIME, Explainable AI (XAI), Logistic Regression, Random Forest, Decision Tree, Machine Learning, Interpretability, Predictive Modeling, Healthcare