A Mixed-Domain Self-Attention Network for Multilabel Cardiac Irregularity Classification Using Reduced-Lead Electrocardiogram

Hao-Chun Yang, Wan-Ting Hsieh, Trista Pei-Chun Chen

Abstract: Electrocardiogram(ECG) is commonly used to detect cardiac irregularities such as atrial fibrillation, bradycardia, and other irregular complexes. While previous studies had great success classifying these irregularities with standard 12-lead ECGs, there existed limited evidence demonstrating the utility of reduced-lead ECGs in capturing a wide-range of diagnostic information. In addition, classification model's generalizability across multiple recording sources also remained uncovered. As part of the Phys-ioNet/Computing in Cardiology Challenge 2021, our team HaoWan_AIeC, proposed Mixed-Domain Self-Attention Resnet (MDARsn) to identify cardiac abnormalities from reduced-lead ECG. Our classifiers received scores of 0.4, 0.33, 0.37, 0.34, and 0.34 (ranked 18th, 23rd, 20th, 23rd, and 22nd) for the 12-lead, 6-lead, 4-lead, 3-lead, and 2-lead versions of the hidden test set with the evaluation metric defined by the challenge.