Interpretable Hybrid CNN-BiLSTM-Attention Model for ECG Arrhythmia Classification

Abstract

Accurate detection of cardiac arrhythmias from electrocardiogram (ECG) signals is critical for early diagnosis and patient care. Although deep learning has achieved high classification performance, many models act as black boxes, limiting clinical adoption. This paper proposes an interpretable hybrid deep learning framework that combines convolutional neural networks (CNNs), bidirectional long short-term memory (BiLSTM) networks, and a self-attention mechanism for the classification of singlelead ECG arrhythmias. CNN layers extract local morphological features, BiLSTM captures temporal dependencies, and self-attention highlights diagnostically relevant signal regions. Using the MIT-BIH Arrhythmia Database and AAMI-recommended class grouping, the model achieves an overall accuracy of 99.52%, demonstrating competitive or superior performance compared to several existing approaches reported in the literature. Attention maps and Grad-CAM visualizations confirm that the network focuses on physiologically meaningful ECG segments, demonstrating both high predictive performance and interpretability. These results suggest the potential of the framework for clinically reliable real-time ECG monitoring and decision support.

More by Ualikhan Sadyk

• Development of Novel Neural Network Architecture for Automated Pharmaceutical and Vitamin Identification Using Image ClassificationInternational Conference on Electronics, Computer and Computation · 2026
• Development of Multimodal Neural Network for Suicidal Ideation Detection Using Kazakh-Russian BERT and Attention MechanismInternational Conference on Electronics, Computer and Computation · 2026
• Benchmarking Deep Transfer Learning for Banknote Recognition: Vision Transformers Vs. CNNs on the KZ-BD DatasetInternational Conference on Electronics, Computer and Computation · 2026
• Kazakh banknote image datasetData in Brief · 2026