An Explainable Hybrid WCBA–BPN–Fuzzy Framework for Heart Disease Prediction: A Comparative Performance Analysis

Abstract

Heart disease remains one of the leading causes of mortality worldwide, necessitating the development of accurate and interpretable prediction systems for early diagnosis. Traditional machine learning approaches often achieve satisfactory predictive performance but lack transparency and the ability to effectively manage uncertainty present in clinical data. To address these limitations, this study proposes an explainable hybrid framework that integrates Weighted Classification Based Association (WCBA), Back Propagation Neural Network (BPNN), and Fuzzy Logic for heart disease prediction. The Cleveland Heart Disease dataset obtained from the UCI Machine Learning Repository was used for experimental evaluation. The proposed framework employs weighted association rule mining to identify clinically significant features, a neural network to model complex nonlinear relationships among risk factors, and fuzzy reasoning to handle uncertainty and provide interpretable diagnostic outcomes. A weighted fusion mechanism combines the outputs of the three modules to generate the final prediction. Experimental results obtained through stratified 10-fold cross-validation demonstrate that the proposed framework achieves superior performance compared with conventional machine learning models. The model attained an accuracy of 92.08%, precision of 0.91, recall of 0.93, F1-score of 0.92, and ROC-AUC of 0.95. The integration of feature-weighted learning and fuzzy inference significantly improves both predictive capability and interpretability. The proposed framework can serve as a reliable clinical decision-support system for the early detection of cardiovascular diseases and may be extended to other healthcare prediction applications.