Performance Comparison of Machine Learning Classifiers and Cardio-Sense Ensemble for Heart Disease Detection

Heart disease remains a leading cause of global mortality; consequently, accurate, reliable, and interpretable predictive models are needed for early diagnosis. This study was aimed at developing a robust hybrid ensemble learning framework that improves heart disease predictive accuracy while preserving clinical interpretability. The Cardio-Sense Ensemble Framework (CSEF) for heart disease prediction was developed by using the publicly available Behavioral Risk Factor Surveillance System dataset. Laplacian binary optimization was used for optimized feature selection, to eliminate redundancy and enhance discriminative information. The refined feature set was used to train multiple baseline classifiers, including logistic regression, random forest, K-nearest neighbors, and XGBoost. Their outputs were combined with Light Gradient Boosting Machine as a meta-classifier. Model performance was assessed with accuracy, sensitivity, specificity, and the area under the receiver operating characteristic curve (AUC). The CSEF achieved an accuracy of 0.98 and an AUC of 0.98, thus outperforming conventional machine learning models while demonstrating improved sensitivity and specificity. Our hybrid ensemble framework offers superior predictive performance and interpretability; therefore, it is suitable for early cardiovascular risk screening and clinical decision-support integration.