Machine Learning and Arrhythmia: Advances in Atrial Fibrillation Detection and Management

In this paper we review recent advancements in the diagnosis and management of atrial fibrillation through machine learning (ML).

Machine learning models developed from clinical records, electrocardiograms (ECGs) as well as data from implantable and wearable devices can now detect and even predict new-onset atrial fibrillation. Other models have improved prediction of stroke risk, increased the success of electrical cardioversions and facilitated catheter ablation of AF. Machine learning presents exciting new opportunities to enhance detection and management of atrial fibrillation. However, these developments need to be weighed against considerations of generalizability, equity, and transparency of these models for real-world utilization in clinical practice. We suggest targeted approaches for evaluation and utilization of ML models to allow for informed clinical implementation.