Abstract
Electrocardiograms (ECGs) are essential tools for the diagnosis and monitoring of heart diseases. Accurate and automatic detection of cardiac abnormalities from ECG signals is crucial for timely intervention and effective patient care. In this paper, we propose a novel 1D Convolutional Long-Short-Term Memory (1D-ConvLSTM) network for the detection of heart diseases in ECG data. The 1D-ConvLSTM network combines the advantages of convolutional neural networks (CNNs) for local feature extraction and long short-term memory (LSTM) networks for capturing temporal dependencies in ECG signals. Our proposed method demonstrates high coherence in feature extraction and low perplexity in the decision-making process, resulting in improved performance for heart disease detection. We evaluate the 1D-ConvLSTM network on a large-scale, publicly available dataset containing various types of heart diseases, including arrhythmias, myocardial infarctions, and heart failure. Results show high performance in terms of different evaluation parameters including accuracy, sensitivity, and specificity, and demonstrating the effectiveness of the 1D-ConvLSTM network for automatic heart disease detection on ECG data. This novel approach has the potential to facilitate early diagnosis and enhance healthcare outcomes for patients with cardiovascular conditions.