A Lightweight CNN-Attention-BiLSTM Architecture for Multi-Class Arrhythmia Classification on Standard and Wearable ECGs

The introduction of a lightweight deep learning model for classifying cardiac arrhythmias marks a significant advancement in health technology. By integrating 1D Convolutional Neural Networks (CNN), attention mechanisms, and Bidirectional Long Short-Term Memory (BiLSTM), this model effectively addresses the critical need for early and accurate detection of arrhythmias, which is essential for timely medical intervention. Evaluated on the CPSC 2018 dataset, the model not only demonstrates superior accuracy and F1-scores over existing baseline models but also maintains a compact size with just 0.945 million parameters. This efficiency makes it particularly suitable for real-time deployment in wearable health monitoring systems, thereby enhancing the accessibility of arrhythmia detection for patients. As wearable technology continues to evolve, the integration of such advanced models could lead to improved health outcomes through proactive monitoring and timely responses to cardiac events.