Explainable Artificial Intelligence for Heart Disease Prediction: A SHAP and LIME-Based Approach — A Review

Cardiovascular disease continues to be one of the leading causes of mortality worldwide, making early and accurate risk prediction essential for timely clinical intervention. Machine learning (ML) models such as Random Forest, Support Vector Machine and XGBoost have achieved high predictive accuracy for heart disease, yet their black-box nature limits clinician trust and hinders real-world adoption. This paper reviews recent research on Explainable Artificial Intelligence (XAI) applied to heart disease prediction, with a focus on two widely used model-agnostic interpretability techniques — SHAP (SHapley Additive exPlanations) and LIME (Local Interpretable Model-agnostic Explanations). It presents the theoretical foundations of both techniques, examines recent studies in terms of dataset, classifier, XAI method and reported performance, identifies open research gaps, and proposes an eight-phase methodology and system architecture for an XAI-based Clinical Decision Support System (CDSS) that predicts heart disease risk while generating transparent, feature-level explanations at both the global (population) and local (patient) level. Challenges, limitations and future research directions are also discussed. The review concludes that combining high-performing ensemble classifiers with SHAP and LIME can substantially improve the transparency, accountability and clinical usability of AI-assisted cardiovascular diagnosis.