A Multistage Deep Learning Algorithm for Detecting Arrhythmia

Abstract

Deep Belief Networks (DBN) is a deep learning algorithm that has both greedy layer-wise unsupervised and supervised training. Arrhythmia is a cardiac irregularity caused by a problem of the heart. In this study, a multi-stage DBN classification is proposed for achieving the efficiency of the DBN on arrhythmia disorders. Heartbeats from the MIT-BIH Arrhythmia database are classified into five groups which are recommended by AAMI. The Wavelet packet decomposition, higher order statistics, morphology and Discrete Fourier transform techniques were utilized to extract features. The classification performances of the DBN are 94.15\%, 92.64\%, and 93.38\%, for accuracy, sensitivity, and selectivity, respectively.