Classification of Electrocardiogram Signal Using Deep Learning Models

Cardiovascular disease is a disease caused by impaired heart and blood vessels function, and is the leading cause of premature death globally. One type of heart disease is the occurrence of symptoms of arrhythmia, which is a condition in which the heartbeat rate is too fast, too slow, or irregular. In recent years, researchers have developed various machine learning (ML) and deep learning (DL) techniques to classify various types of arrhythmias via electrocardiogram (ECG) signals. This monitoring aims to be an early treatment for various types of arrhythmia disorders, so that the proper detection of ECG deviations will assist doctors and other health professionals in taking appropriate medical action quickly and accurately. The classification proposed in this study is to classify 5 types of arrhythmia disorders from the MIT-BIH database based on AAMI rules using several CNN architectural learning models such as AlexNet, ResNet-50, InceptionNet, and VGG-16 by utilizing resampling techniques and Gaussian mixed models. The results show that the classification accuracy of each proposed model is 97.99% for AlexNet, 97.60% for ResNet-50, 97.59% for InceptionNet, and 97.55% for VGG-16.