Automated Detection of Cardiac Arrhythmia Using Recurrent Neural Network

Abstract

Cardiac arrhythmia is a medical condition characterized by an irregular heartbeat, which may be too fast, too slow, or inconsistent. It
can lead to severe complications if not diagnosed and treated in a timely manner. Electrocardiogram (ECG) signals play a crucial role in detecting
and analyzing cardiac arrhythmia. The goal of this paper is to apply deep learning techniques to the diagnosis of cardiac arrhythmia using ECG
signals while minimizing the amount of data preprocessing required.. However, deep learning techniques provide a more automated and data-
driven approach by directly learning from raw ECG signals. Our results demonstrate that the CNN-LSTM hybrid model achieves a five-fold
cross-validation accuracy of 0.834 in distinguishing normal and abnormal ECG signals associated with cardiac arrhythmia. This indicates that
combining CNNs with LSTMs enhances the model's ability to capture both spatial and temporal dependencies in ECG signals. Furthermore, the
accuracy obtained by other hybrid architectures, such as CNN-GRU and CNN-RNN, is comparable to that of the CNN-LSTM model, suggesting
that deep learning-based approaches are highly effective in identifying cardiac arrhythmia. In conclusion, this study highlights the potential of
deep learning techniques in automating the diagnosis of cardiac arrhythmia with minimal data preprocessing. The results suggest that hybrid
architectures, particularly CNN-LSTM, offer promising accuracy and can serve as reliable diagnostic tools. CNN with an accuracy of 95.22%. An
accuracy of 84.54% was achieved in the detection of inferior MI in ECG using CNN.