ExplainabilityTransparencyTrustworthinesscs.AI
CoFE: A Framework Generating Counterfactual ECG for Explainable Cardiac AI-Diagnostics
Jong-Hwan Jang, Junho Song, Yong-Yeon Jo
August 22, 2025
IEEE International Conference on Acoustics, Speech, and Signal Processing
4 citations
Key Takeaways
-
Counterfactual explanations make AI-ECG decisions transparent: The CoFE framework generates modified ECG signals showing doctors exactly which features (like heart rhythm patterns and electrical intervals) the AI model relies on when making diagnostic predictions, moving beyond "black box" decision-making.
-
AI explanations align with established medical knowledge: The framework's outputs matched clinically recognized indicators for conditions like atrial fibrillation, suggesting the AI is learning medically valid patterns rather than spurious correlations—a critical validation for clinical adoption.
-
Locating important features matters as much as identifying them: By clarifying both where relevant ECG features appear and how they influence predictions, the framework helps clinicians understand not just what the AI noticed, but whether it's looking at the right parts of the signal in clinically appropriate ways.
-
Explainability is essential for clinical integration: The paper emphasizes that AI-ECG models cannot be safely integrated into real medical practice without interpretability tools, positioning explainable AI as a prerequisite rather than an optional enhancement for responsible healthcare AI deployment.
Abstract
Recognizing the need for explainable AI (XAI) approaches to enable the successful integration of AI-based ECG prediction models (AI-ECG) into clinical practice, we introduce a framework generating \textbf{Co}unter\textbf{F}actual \textbf{E}CGs (i,e., named CoFE) to illustrate how specific features, such as amplitudes and intervals, influence the model's predictive decisions. To demonstrate the applicability of the CoFE, we present two case studies: atrial fibrillation classification and potassium level regression models. The CoFE reveals feature changes in ECG signals that align with the established clinical knowledge. By clarifying both \textbf{where valid features appear} in the ECG and \textbf{how they influence the model's predictions}, we anticipate that our framework will enhance the interpretability of AI-ECG models and support more effective clinical decision-making. Our demonstration video is available at: https://www.youtube.com/watch?v=YoW0bNBPglQ.