Electrocardiogram based cardiovascular disease detection with Ensemble Learning Classifier

Cardiovascular diseases are the main causes of death in today's world. This has driven the need for new technology to be employed to help detect any heart diseases faster and more conveniently. The goal of this study is to develop a novel Electrocardiogram based cardiovascular disease detection method using an Ensemble classifier. A dataset of the various heart diseases, including normal heartbeat, was collected for this study. Many popular machine learning algorithms like K Nearest Neighbor, Logistic Regression, and Support Vector Machine were first executed for the collected dataset. To improve the performance, a new Ensemble classifier has been developed, which gives better performance than the other machine learning algorithms listed before. Then its performance was compared with the other machine learning algorithms. It is seen that the Ensemble classifier can provide an overall accuracy of 93% and outperform all the other ML algorithms. A comparison of the F1-Score and Accuracy of these algorithms for the cases of normal, abnormal heartbeat and myocardial infarction is given in the form of a graph. A comparison table of the accuracy of the existing algorithms is also presented.