Arrhythmia Classification of Merged Features Method Based on SENet and BiLSTM

Arrhythmia is a common cardiovascular disease. Using ECG to automatically classify arrhythmia episodes has important application value. To improve the accuracy of automatic classification, a deep learning method (SEN-BiLSTM) based on the feature fusion of single heartbeat and continuous heartbeat interval was proposed. The combination of SENet and BiLSTM was used in the extraction of morphological features of single heartbeat. In the MIT-BIH arrhythmia dataset, the five-fold cross-validation of the proposed method obtained 99.68% accuracy, 94.46% sensitivity, 99.43% specificity and 94.98% positive value. The results show that compared with other methods, the model can effectively extract signal features and has higher adaptability and accuracy. Therefore, the model can be used as an auxiliary tool to help clinicians diagnose arrhythmia.