Performance comparison of CNN and LSTM algorithms for arrhythmia classification

One of the critical CVDs is cardiac arrhythmia and has caused significant fatalities. Recently, deep learning models are utilized for the classification of arrhythmia disease through electrocardiogram (ECG) signal analysis. Among the existing deep learning model, convolutional neural network (CNN) and long short-term memory (LSTM) algorithms are extensively used for arrhythmia classification. However, there is a lack of study that analyzes the performance comparison of CNN and LSTM algorithms for arrhythmia classification. In this paper, the performance of CNN and LSTM algorithms for arrhythmia classification is compared for a publicly available dataset. Specifically, the MIT-BIH arrhythmia dataset is used and the performance is measured in terms of area under the curve (AUC) and receiver operating characteristic (ROC) curve. Analyzing the performance of these algorithms will further assist in the development of an enhanced deep learning model that offers improved performance.