Classification of ECG Arrhythmia Using a Convolution Neural Network

Cardiovascular disease (CVD) is the leading cause of mortality at the moment. The current approach to illness diagnosis involves analyzing the electrocardiogram (ECG), a medical monitoring instrument that records heart activity. Analyzing a lot of ECG data would take too much medical equipment. Machine learning-based methods for recognizing ECG features are becoming more and more common. These well-liked techniques have several limitations, namely the need for manual feature recognition, intricate models, and lengthy training cycles. The ECG data is obtained from the MIT-BIH database and classified using a convolutional neural network in this study. The outcomes show that the model used in this research performs better in terms of accuracy, precision-recall, sensitivity and f1-score. The five distinct heartbeat types reflected by the time domain signals of the ECG in the initial stage were the atrial premature contraction beat, premature ventricular contraction beat, normal beat, left bundle branch block beat and right bundle branch beat. In this work, the total classification accuracy rating for SVM is 94%. Less accurate than linear SVM is RBF SVM. The total classification accuracy rate of a neural network is 95%.