A New Automatic Approach to Distinguish Myocardial Infarction Based on LSTM

Abstract

Myocardial Infarction (MI) has the characteristics of rapid development and poor prognosis. Early intervention is of great significance in relieving pain and preventing death. For reducing the misdiagnosis rate of MI, a novel classification approach of MI based on a long short term memory (LSTM) is proposed in this paper. Firstly, the original electrocardiogram (ECG) signal is preprocessed, and then it is divided into a heartbeat sequence. Then the heartbeat sequence is input into the deep neural network model for training and learning. Finally, the validity of the method is verified on the Physikalisch-Technische Bundesanstalt (PTB) ECG database. The accuracy of the method is 99.91%. The experimental results show that the classification accuracy of the proposed method is superior to the other methods.