Optimizing Heart Disease Prediction Using a Hybrid Dynamic Swarm Evolution Approach

Early identification of Cardiovascular issues is crucial due to the high mortality rate associated with heart disease. This research assesses machine learning algorithms for heart disease prediction using tabular datasets. Previous researchers suggested Tree-based models like Random Forest, XGBoost, and Decision Trees and various hyper-parameter optimization methods including Grid Search, Random Search, Swarm and Evolutionary Algorithms which excel in accuracy and robustness. However, they are computationally inefficient and less effective in dynamic settings. A novel Hybrid Swarm Evolution Optimization (HySEOpt) is introduced, which adjusts mutation rates based on performance curves and utilizes parallel processing for faster optimization achieving 98.01% accuracy. HySEOpt enhances model’s quality and robustness, addressing limitations of existing methods and contributes to hyper-parameter optimization in predictive healthcare modeling.