Multi-label Anomaly Classification Based on Electrocardiogram

Abstract

Under the background of 5G and AI, it is particularly important to use cloud computing, Internet of things and big data technology to analyze massive physiological signals of patients in real time. Arrhythmia can cause some major diseases, such as heart failure, atrial fibrillation and so on. It’s difficult to analysis them quickly. In this paper, a deep learning model of multi-label classification based on optimized temporal convolution network is proposed to detect abnormal electrocardiogram. The experimental results show that the accuracy of the model is 0.960, and the Micro F1 score is 0.87.