Developing a Mixed Deep Learning Model Approach for Arrhythmia Classification Based on ECG Signal

Nowadays, deaths from cardiovascular diseases (CVDs) are exponentially increasing due to various factors. These include underlying health conditions, perceived medical treatment expenses, and lack of timely access to medical treatment. However, Electrocardiogram (ECG) comes in as a prominent process of recording electrical heart signals and predicts specific cardiovascular diseases based on these electrical signals. Furthermore, recent researchers have shown many methods to classify heartbeats based on these signals such as deep learning using convolutional neural network, recurrent neural network, such classification further help the treatment become more convenient and automatic. Therefore, this paper proposes a mixed method based on deep learning to differentiate arrhythmia heartbeats with 5 different types of heartbeats applying PhysionNet's MIT-BIH Arrythmia Database. According to the outcome, the proposed method has a macro accuracy of 92 % and an average accuracy of 99%.