Classification of 12-lead ECGs: the PhysioNet/Computing in Cardiology Challenge 2020

Matthew Reyna1, Erick Andres Perez Alday2, Annie Gu1, Chengyu Liu3, Salman Seyedi1, Ali Bahrami Rad4, Andoni Elola5, Qiao Li6, Ashish Sharma1, Gari Clifford7
1Emory University, 2Oregon Health and Science University, 3Southeast University, 4University of Tampere, 5UPV/EHU, 6Emory, 7Emory University and Georgia Institute of Technology


The PhysioNet/Computing in Cardiology Challenge 2020 focuses on the identification of cardiac abnormalities in 12-lead electrocardiogram (ECG) recordings. For the unofficial phase, a total of 9847 recordings were sourced from hospital systems from three distinct countries; with all records being annotated with clinical diagnoses. 6877 were posted for training.

For this Challenge, we asked participants to design working, open-source algorithms for identifying cardiac abnormalities in 12-lead ECG recordings. This Challenge provides several innovations. First, we sourced data from multiple institutions from around the world with different demographics, allowing us to assess the generalizability of the models. Second, we required participants to submit both their trained models and code for reproducing their trained models from the training data, which improves the generalizability and reproducibility of the Challengers' approaches. Third, we proposed a novel evaluation metric that reflects the impact of different misclassification errors for cardiac abnormalities, particularly for missed cardiac abnormalities from ECG recordings.

So far, 188 teams have submitted 678 algorithms (365 successfully ran) representing a diversity of approaches from both academia and industry for identifying cardiac abnormalities.