Automatic classification of ECG heartbeats using deep neural networks

Abstract

Automatic ECG classification systems are a valuable tool for assisting doctors and su- pervising patients. Deep neural networks have been widely used as an alternative to the existing classification systems based on distinct feature extraction and classification phases. In this study, we propose an automatic system that classifies ECG heartbeats based on 1D convolutional neural network. To evaluate the proposed model, we perform tests on the MIT-BIH arrhythmia database and we considers four classes