ECG-TransCovNet: A hybrid transformer model for accurate arrhythmia detection using Electrocardiogram signals

Abnormalities in the heart's rhythm, known as arrhythmias, pose a significant threat to global health, often leading to severe cardiac conditions and sudden cardiac deaths. Therefore, early and accurate detection of arrhythmias is crucial for timely intervention and potentially life-saving treatment. Artificial Intelligence, particularly deep learning, has revolutionised the detection and diagnosis of various health conditions, including arrhythmias. A unique hybrid architecture, ECG-TransCovNet, that combines Convolutional Neural Networks and Transformer models for enhanced arrhythmia detection in Electrocardiogram signals is introduced. The authors' approach leverages the superior temporal pattern recognition capabilities of Transformers and the spatial feature extraction strengths of convolutional neural networks, providing a robust and accurate solution for arrhythmia detection. The performance and generalisability of the authors' proposed model are validated through tests on the MIT-BIH arrhythmia and PhysioNet databases. The authors conducted experimental trials using these two benchmark datasets. The authors' results demonstrate that the proposed ECG-TransCovNet model achieves state-of-the-art (SOTA) performance in terms of detection accuracy, reaching 98.6%. Additionally, the authors conducted several experiments and compared the results to the most recent techniques utilising their assessment measures. The experimental results demonstrate that the authors' model can generally produce better results.