Prediction of Coronary Artery Disease using Electrocardiography: A Machine Learning Approach

Abstract

Coronary Artery Disease (CAD) is a leading cause of cardiovascular morbidity and mortality globally. There has been an indication of association between Electrocardiography (ECG), a measurement for electrical activity in the heart, and CAD, which makes ECG a promising screening tool. Consequently, Machine Learning techniques can detect patterns of ECG that are able to screen CAD cases. We developed a machine learning tool that extracts RR interval features from ECG signals, and used different statistical learning algorithms to detect CAD based on these features. Our results indicate that patterns in ECG signals and attributes of patients such as age and gender can predict CAD in diverse clinical scenarios in real life with a performance superior to the available screening and diagnostic tests.