EFFECTIVE MACHINE LEARNING FEATURE SELECTION FOR CARDIOVASCULAR DISEASE DETECTION

Abstract

The cause of death for many people worldwide is cardiovascular disease (CVD). According to the World Health Organization (WHO), cardiovascular disease (CVD) claimed 19.1 million lives in 2022, accounting for 33% of all fatalities worldwide. With standard machine learning, ECG is frequently utilized for automated CVD diagnosis; however, selecting the appropriate characteristics can be challenging. A scalable ML-based architecture for early CVD detection that makes use of the optimal feature selection can address this issue. By promptly diagnosing and treating CVD, this design aims to transform healthcare and reduce the number of fatalities. ML classifiers are used to gather, store, and evaluate data in order to predict heart problems. After removing ECG characteristics, FCBF, MrMr, alleviation, and PSO-optimization choose the best features. The Random Forest and trait-trained Extra Tree classifiers have 100% accuracy. On both small and big datasets, we also evaluate the suggested approach against the best one. The suggested approach might enhance quality of life by altering the way patients are treated and significantly reducing the number of deaths from CVD. In summary, this design improves healthcare systems and offers a practical CVD solution