Deep Learning for ECG Classification: Uncovering Learned Features Aligned with Diagnostic Criteria

This paper investigates the learned features of a deep learning model trained for 12-lead ECG classification, demonstrating that these features align with established diagnostic criteria. The study provides insights into the model's decision-making process, highlighting its interpretability and potential clinical utility. By analyzing the model's learned features, the authors reveal a strong correlation between the model's predictions and recognized diagnostic indicators, suggesting a potential for using deep learning models in clinical settings for ECG interpretation.