Efficient Automated Cardiovascular Disease Detection Using Machine Learning

Abstract

Cardiovascular diseases pose a significant threat to global public health. Among diagnostic approaches, heart sound detection techniques offer non-invasive means for predicting cardiovascular conditions. While electrocardiogram (ECG) signals are commonly utilized for heart disease diagnosis, their limited spatial resolution necessitates alternative methods. Phonocardiogram (PCG) signals and sound processing techniques present viable alternatives. This paper explores the extraction of diverse features from PCG signals to classify patients using artificial intelligence algorithms. Simulation results demonstrate the superior performance of the XGBoost algorithm in cardiovascular patient detection compared to other methods. Specificity, sensitivity, and accuracy values were recorded at 99±1.93%, 98±2.76%, and 99±1.78%, respectively. Implementing the proposed method promises accurate and swift diagnosis of cardiovascular patients, offering significant support to healthcare professionals in patient screening. Future research avenues include extending this methodology to diagnose heart valve diseases, further enhancing its clinical utility.