Deep Learning Based Supraventricular Arrhythmia Detection

Deep learning is used in various applications like health care, virtual assistants, chatbots. The heart’s electrical activity is captured by an electrocardiogram (ECG). Irregular heartbeat or dysrhythmia is known as arrhythmia. An improved bidirectional Long Short-Term Memory (BI-LSTM) is proposed for the detection of supraventricular arrhythmia. Instead of learning every weight in the network in one training iteration, only a portion of the weights are learned through dropout. So, dynamic drop rate is chosen in improved BI-LSTM model. Three models, Long Short-Term Memory (LSTM), Gated recurrent unit (GRU), and BI-LSTM, are also implemented, and the experimental results are compared to evaluate the performance of the proposed model. This study examines the MIT-BIH Supraventricular Arrhythmia database. Accuracy, Precision, Recall, F-score, and Matthews Correlation Coefficient are used to calculate performance. The BI-LSTM algorithm with dynamic dropout layer outperformed with accuracy of 99.65%.