A Dataset and Benchmarks for Atrial Fibrillation Detection from Electrocardiograms of Intensive Care Unit Patients.

Across both datasets, electrocardiograms foundation models generally performed best, followed by deep learning, then feature-based classifiers.

However, the difference between deep learning and feature-based classifiers is minimal and highly dependent on the model selected, with feature-based classifiers obtaining a very slightly higher average performance on our intensive care unit test set but deep learning achieving a higher overall maximum performance.