Evaluating the Efficacy of Feature Selection Methods in Cardiovascular Disease Prediction with Machine Learning

Cardiovascular disease (CVD) remains a primary contributor to global mortality, necessitating refined predictive models. Machine learning (ML) has emerged as a pivotal tool in augmenting clinical decisions, particularly in CVD prediction. Nonetheless, the performance of ML classifiers is substantially influenced by the judicious selection of relevant features. This manuscript presents a focused comparative analysis of feature selection methodologies in CVD prediction through ML classifiers. Utilizing a clinical dataset, the study assesses the efficacy of various classifiers based on distinct feature subsets. Empirical results indicate that the Random Forest classifier, with a mere five selected features, attains a remarkable 99.70% accuracy, thus qualifying as an optimal choice for clinical settings. The research underscores the necessity of scrupulously evaluating diverse combinations of feature selection and learning algorithms to derive the most effective model. Collectively, this work provides pivotal insights into optimal feature selection for precise CVD prediction, thereby contributing to the development of more efficient predictive strategies.