Advanced Heart Disease Classification Through Feature-Tuned Explainable Support Vector Machine with Hyperparameter Tuning and Cross-Validation

Heart disease is the leading cause of death globally, claiming approximately 17.9 million lives annually, with a disproportionate impact in low- and middle-income countries. Despite advances in artificial intelligence (AI), current heart disease prediction models often lack accuracy, generalizability, and explainability. To address these challenges, this study presents a novel approach that integrates data from three diverse heart disease datasets, creating a comprehensive and representative dataset. Advanced preprocessing ensures high-quality inputs, while the proposed Feature-Tuned Explainable Support Vector Machine (FTX-SVM) employs manual hyperparameter tuning to enhance performance. The model achieves a remarkable 97.46% accuracy through robust validation using 10-fold cross-validation, demonstrating its reliability and effectiveness. Beyond accurate predictions, the FTX-SVM leverages explainable AI techniques to interpret model behavior and identify key factors contributing to heart disease. Comparative analysis highlights the superior performance of the proposed method over existing models, providing a reliable, interpretable, and actionable tool for early detection and management of heart disease, ultimately contributing to better patient outcomes.