Optimizing Machine Learning Classifiers for Enhanced Cardiovascular Disease Prediction

A key challenge in developing Machine Learning (ML) models for predicting or diagnosing Cardiovascular Disease (CVD), is selecting suitable algorithms and fine-tuning their parameters. In this study, we employed three ML techniques, namely Auto-WEKA, Decision Table/Naive Bayes (DTNB), and Multiobjective Evolutionary (MOE) fuzzy classifier to create diagnostic models using the Heart Disease Dataset from IEEE Dataport. Auto-WEKA generated a highly accurate model with a 100% success rate through optimal classifier selection and hyperparameter configuration. The DTNB classifier yielded a satisfactory 85.63% prediction accuracy concerning patients' risk levels. Further refinements, though, could help reduce possible misclassifications. Finally, the MOE fuzzy classifier achieved approximately 81.6% accuracy, indicating the potential for enhancing precision and recall values by adjusting classifier settings. Our findings underscore the promise of ML tools in CVD diagnosis and suggest further optimization of classifier parameters for superior performance.