Perspective of Generalizing Deep Boltzmann Machine for ECG Signal Classification

Abstract

Deep learning is a highly efficient technique for handling large volume big data and decision-making processes. The above objective is achieved by incorporating healthcare service provider knowledge as a rule to meticulously classify physiological signal like Electrocardiogram (ECG) signal from the extracted large amount of big data from the wearable sensor. In this study, we generalized a novel deep Boltzmann machine (DBM) to diagnose arrhythmia in real-time ECG signals. The performance of proposed DBM was evaluated by accuracy, specificity, sensitivity and mean field interference of training algorithms. From the obtained results, we claim that the proposed deep learning model has high performance in the classification of most of the arrhythmia ECG and has high accuracy of 96.8% and sensitivity of 98.83% in classification. Moreover, the significance of proposed DBM model performs consistently on various real-time ECG dataset.