Early warning of atrial fibrillation using deep learning

• DOI: 10.1016/j.patter.2024.100970
• Publication Date: 2024-04-18T14:35:57Z
• Authors: Marino Gavidia; Hongling Zhu; Arthur N. Montanari; Jesús Fuentes; Cheng Cheng; Sergio Dubner; Martin Chames; Pierre Maison-Bla...; Md Moklesur Rahman; Roberto Sassi; Fabio Badilini; Yinuo Jiang; Shengjun Zhang; Hai-Tao Zhang; Hao Du; Basi Teng; Ye Yuan; Guohua Wan; Zhouping Tang; Xin He; Xiaoyun Yang; Jorge Goncalves
• Keywords: Wearable Technology

Atrial fibrillation (AF), the most prevalent cardiac rhythm disorder, significantly increases hospitalization and health risks. Reverting from AF to sinus (SR) often requires intensive interventions. This study presents a deep-learning model capable of predicting transition SR on average 30.8 min before onset appears, with an accuracy 83% F1 score 85% test data. performance was obtained R-to-R interval signals, which can be accessible wearable technology. Our model, entitled Warning Fibrillation (WARN), consists deep convolutional neural network trained validated 24-h Holter electrocardiogram data 280 patients, 70 additional patients used for testing further evaluation 33 two external centers. The low computational cost WARN makes it ideal integration into technology, allowing continuous heart monitoring early detection, potentially reduce emergency interventions improve patient outcomes.