Explainable AI-powered hybrid metaheuristic optimization for heart disease prediction in connected healthcare systems

Abstract

Coronary heart disease continues to be a major cause of death worldwide, and insufficient ongoing health monitoring makes it difficult to make precise predictions. Real-time patient data collection is made possible by modern IoT devices, but their massive data volumes necessitate efficient machine learning analysis. In order to optimize Random Forest hyperparameters for better CHD prediction, this study suggests a hybrid Genetic-Red Kite Optimization Algorithm. Explainable AI techniques are integrated to enhance model transparency, and feature-level interpretability. Experiments conducted on Framingham, Z-Alizadeh Sani, and a comprehensive combined dataset demonstrate accuracies of 91.5%, 97.7%, and 95.2%, outperforming state-of-the-art methods across multiple evaluation metrics under diverse clinical scenarios.