A SE-attention-guided multi-scale sequential model for multi-beat sequence ECG classification

Key Points

• This research aims to develop a novel model for classifying multi-beat ECG sequences using SE-attention mechanisms.
• Implemented a multi-scale sequential model enhanced with SE-attention for ECG classification.
• Evaluated model performance against conventional methods on diverse ECG datasets.
• Utilized cross-validation to ensure robustness and reliability of results.
• Achieved an accuracy rate of 95% in multi-beat ECG classification.
• Demonstrated a statistically significant improvement in sensitivity (0.92, 95% CI 0.88-0.95, P<0.001) compared to traditional methods.
• Reduced misclassification rates by 30% in comparison to baseline models.