Nature-Inspired Hybrid Harris Hawks Algorithm for Context-Aware Prediction of Coronary Artery Disease

Abstract

Predictive models that are both accurate and efficient are needed for the early detection and prevention of coronary artery disease (CAD), which is still a major cause of death worldwide. Using the Heart Disease Prediction dataset—which include both original features and data from the Context-Aware Model (CAM)—this study examines how CAD prediction accuracy is affected by the use of advanced feature selection approaches in conjunction with machine learning algorithms. In order to optimize feature subsets and enhance model performance, a new hybrid feature selection method is suggested. This method combines Harris Hawks Optimization (HHO) and Grey Wolf Optimization (GWO). The research looks at many ML methods, but it primarily looks at a Voting Classifier ensemble that uses Decision Tree and Random Forest models. The Voting Classifier got 79% accuracy on the CAM dataset. The original dataset achieved a flawless classification accuracy of 100% when HHO-based feature selection was applied. The hybrid HHO-GWO feature selection method outperformed baseline models significantly, with an accuracy of 89%. When it comes to improving feature selection for CAD prediction models, the outcomes highlight the effectiveness of optimization methods inspired by nature, specifically HHO and GWO. These results show how these technologies can be used to improve CAD early diagnostic tools and help preventive healthcare programs work better.