Nature Inspired Optimization in Context-Aware-Based Coronary Artery Disease Prediction: A Novel Hybrid Harris Hawks Approach

Coronary Artery Disease (CAD) imposes a significant global health burden, profoundly impacting morbidity and mortality rates worldwide. Accurate prediction of CAD is paramount for efficient management and prevention of associated complications. This study introduces a novel Hybrid Harris Hawks Optimization (H-HHO) approach, incorporating three noteworthy enhancements to augment classifier efficacy in CAD prediction compared to the conventional HHO algorithm. The advanced methodology was deployed for hyperparameter tuning of standard classification algorithms, including Support Vector Machine (SVM), Logistic Regression (LR), K-Nearest Neighbors (KNN), and Random Forest (RF). Moreover, a Context-Aware based Model (CAM) was employed to discern critical features (e.g., thallium and chest pain type) for CAD prediction, with subsequent comparison of their outcomes. The UCI heart disease dataset served as the basis for evaluating the efficiency of HHO and H-HHO algorithms, where H-HHO demonstrated superior performance, achieving an accuracy of 94.74% with LR and SVM, compared to the highest accuracy of 82.46% among classifiers using the HHO approach. The proposed H-HHO methodology for hyperparameter tuning in machine learning algorithms presents a promising framework, showcasing its effectiveness in CAD prediction. Future research endeavors may further explore H-HHO's application across diverse medical prediction tasks and its integration into other meta-heuristic algorithms to advance healthcare applications.