Research on Arrhythmia of College Students Based on Convolutional Neural Network

Lei Qi, Fuchun Zhang, Baiyang Wang, Meng Li

Abstract: Arrhythmia is a group of common diseases related to irregular heart rate. Accurate classification of electrocardiogram is very important for college students to detect heart disease. Experts will spend a major expenditure of time and effort in the examination and analysis of college students' ECG waveform. In order to solve this problem, this paper proposes a method of college students' arrhythmia detection based on deep learning, which classifies different types of arrhythmias. Firstly, 1-D ECG signals are converted into 2-D ECG images. Then the signals of different channels are fused and finally input into the network. In this work, we do experiments on the basis of residual network. For verifying the feasibility of this method, it is trained and verified in the public data set MIT-BIH, and the classification accuracy reaches 99.00%. The experimental results proved that it can effectively and reliably identify ECG signals, and has potential clinical application value. It can assist doctors to screen college students for arrhythmias.