Enhanced ECG arrhythmia detection through deep learning techniques

Cardiac arrhythmias, characterized by irregular heart rhythms, can lead to severe health issues. Early and precise detection is essential to reduce the mortality rates linked to heart conditions. Consequently, an automated system for detecting cardiac arrhythmias is necessary. This study introduces a deep learning-based method for the automatic detection of cardiac arrhythmias using ECG signals from the MIT-BIH database. We utilized five deep learning models—1D-CNN, DNN, RNN, LSTM, and GRU—to compare their performance and identify the most effective model. Our proposed approach can accurately detect various arrhythmias, such as atrial fibrillation, ventricular fibrillation, and bradycardia. Results showed that the 1-D CNN achieved an accuracy of 99.6%, DNN reached 97.2%, Simple RNN recorded 35%, LSTM attained 97%, and GRU obtained 98.3%.