Research On ECG Signal Classification Based On Convolutional Neural Network With Attention Mechanism

With the gradual acceleration of the pace of life in modern society,more and more people suffer from chronic sleep deprivation.However,long-term inadequate rest will increase the risk of people suffering from cardiovascular diseases.Professional doctors usually use the important features of the electrocardiogram to diagnose this type of heart disease.Since the convolutional neural network can efficiently and automatically extract the features of the time series,we can ingeniously design the convolutional neural network for ECG classification to assist doctors in making efficient and correct diagnosis of diseases.In this article,the convolutional neural network is used as the backbone,combined with the attention mechanism to extract the key features of the ECG for ECG classification,and carry out ECG signal classification experiments in inter patient and intra patient modes on ICBEB database and MIT-BIH database respectively.Specifically,this paper mainly designs the convolutional neural network MyNet1＿Att and MyNet2＿Att by combining the pyramid pooling,residual connection,attention mechanism and other technologies with the multi-channel feature extraction method as the center.The model MyNet1＿Att’s inter patient ECG nine classification task on ICBEB database achieved an accuracy of 80.04%,and this model outperformed other baseline models by more than 3.78%.The model MyNet2＿Att’s five classification task of patients’ ECG signals on MIT-BIH database achieved an accuracy of 98.31%,which is also better than other baseline models.It should be pointed out that the parameters of the model proposed in this paper are much lower than other baseline models.Experiments demonstrate that the attention-based convolutional neural network proposed in this paper achieves improved performance whether it is applied to ECG signal classification in either the inter-patient mode or the intra-patient mode.