Po-661-03 use of a deep learning algorithm to predict paroxysmal atrial fibrillation based on printed electrocardiographic records acquired during sinus rhythm

Atrial fibrillation (AF) is a common type of sustained arrhythmia worldwide. Asymptomatic AF, which occurs frequently, is associated with an increased incidence of ischemic stroke, heart failure, and mortality. A large number of patients with paroxysmal atrial fibrillation (PAF) remain undiagnosed due to the absence of electrocardiographic evidence of AF (AF-ECGs). If PAF could be predicted, targeted screening could improve early detection and treatment of this condition.