ECGGAN: A Framework for Effective and Interpretable Electrocardiogram Anomaly Detection

Heart is the most important organ of the human body, and Electrocardiogram (ECG) is an essential tool for clinical monitoring of heart health and detecting cardiovascular diseases. Automatic detection of ECG anomalies is of great significance and clinical value in healthcare. However, performing automatic anomaly detection for the ECG data is challenging because we not only need to accurately detect the anomalies but also need to provide clinically meaningful interpretation of the results. Existing works on automatic ECG anomaly detection either rely on hand-crafted designs of feature extraction algorithms which are typically too simple to deliver good performance, or deep learning for automatically extracting features, which is not interpretable. In this paper, we propose ECGGAN, a novel reconstruction-based ECG anomaly detection framework. The key idea of ECGGAN is to make full use of the characteristics of ECG with the periodic metadata, namely beat, to learn the universal pattern in ECG from representative normal data. We establish a reconstruction model, taking leads as constraints to capture the unique characteristics of different leads in ECG data, and achieve accurate anomaly detection at ECG-level by combining multiple leads. Experimental results on two real-world datasets and their mixed-set confirm that our method achieves superior performance than baselines in terms of precision, recall, F1-score, and AUC. In addition, ECGGAN can provide clinically meaningful interpretation of results by revealing the extent to which abnormal sites deviate from the normal pattern.