Hybrid shallow and deep learned feature mixture model for arrhythmia classification

Abstract

For acquiring the abnormality of the signal, the Electrocardiography (ECG) signals are classified into two classes. Usually, ECG is acquired by placing an electrode on the skin and recording the ECG signals. Then, these signals are used by the physician to analyze the situation of the patients and report their health. In this paper, four features are extracted from the ECG signal in order to classify the signal. The MIT-BIH arrhythmia dataset, which contains an enormous number of signals encompassing the label of normal and arrhythmia signal have been used to acquire the ECG signal. The classification of the signal into arrhythmia and sinus rhythm is done using Deep Neural Network (DNN). The result obtained clearly depicted that the DNN has a higher classification than the other classifiers.