Ambulatory atrial fibrillation detection and quantification by wristworn AI device compared to standard holter monitoring
Mariska van Vliet, Jan J.J. Aalberts, Cora Hamelinck, Arnaud D. Hauer, Dieke Hoftijzer, Stefan H. J. Monnink, Jurjan C. Schipper, Jan Constandse, Nicholas S. Peters, Gregory Y.H. Lip, Steven R. Steinhubl, Eelko Ronner
Abstract
Timely detection of atrial fibrillation (AF) is crucial for the prevention of serious consequences such as stroke and heart failure, yet it remains challenging due to its often asymptomatic or paroxysmal nature. Wearable devices with artificial intelligence algorithms offer promising solutions. AF detection by the CardioWatch 287-2 (CW2), a wrist-worn photoplethysmography (PPG) and single-lead ECG device, was compared to 24-h Holter. Patient compliance, AF prevalence and AF burden were evaluated for 27 additional days. Data from 150 participants (mean age 64 ± 12 SD; 41% female) were analysed. The CW2's PPG and single-lead ECG algorithms achieved a specificity ≥98% and sensitivity ≥95% for AF detection, and 99% correlation for AF burden, compared to 24-h Holter. AF prevalence increased from 14.7% (24-h Holter) to 26.7% (28-day CW2). Thus, the wrist-worn device showed promising performance in detecting AF and determining AF burden. The trial was registered on ClinicalTrials.gov (NCT05899959) on June 2, 2023.