Comparative Study of Feature Selection Algorithms for Cardiovascular Disease Prediction with Artificial Neural Networks

Predicting cardiovascular diseases through advanced machine learning models is contingent upon the judicious selection of relevant features. This study engages in a comprehensive comparative analysis of eight feature selection algorithms, aiming to discern the optimal methodology for enhancing the predictive performance of an Artificial Neural Network (ANN) model. Leveraging the extensive dataset from the Framingham Heart Study, our investigation navigates through the intricacies of feature selection, evaluating algorithms ranging from correlation-based techniques to ensemble learning approaches. Each algorithm undergoes meticulous empirical evaluation, shedding light on its efficacy in distilling the dataset’s complex features into a concise yet informative set. The foundational section of our study provides a detailed exploration of the Framingham dataset, outlining key attributes and their interpretations. Emphasizing the critical role of feature selection in predictive modeling, we elucidate the significance of refining datasets for optimal model performance. Our findings hold implications not only for the specific domain of cardiovascular health but also contribute to the broader landscape of predictive modeling in healthcare. The discernment of an optimal feature selection algorithm, tailored for ANNs, paves the way for more accurate, efficient, and interpretable models. This research offers valuable insights for healthcare professionals and researchers, augmenting the toolkit for advancing predictive capabilities in the realm of cardiovascular disease management.