Why the study?
While deep learning models have improved ECG-based cardiac arrhythmia detection, proper explanations are required and it has been unclear which feature attribution methods are appropriate for ECG.
Grad-CAM is identified as a highly effective feature attribution method for providing explanations in deep learning-based ECG arrhythmia detection.
Three customized metrics enable rigorous ECG explanation assessment; leaves open optimal attribution method selection for clinical DL models.
The performance of cardiac arrhythmia detection with electrocardiograms(ECGs) has been considerably improved since the introduction of deep learning models. In practice, the high performance alone is not sufficient and a proper explanation is also required. Recently, researchers have started adopting feature attribution methods to address this requirement, but it has been unclear which of the methods are appropriate for ECG. In this work, we identify and customize three evaluation metrics for feature attribution methods based on the characteristics of ECG: localization score, pointing game, and degradation score. Using the three evaluation metrics, we evaluate and analyze eleven widely-used feature attribution methods. We find that some of the feature attribution methods are much more adequate for explaining ECG, where Grad-CAM outperforms the second-best method by a large margin.
No takes yet. Share an insight, caveat, or question.
Suh et al. (2022) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: