Clinical validation of five direct-to-consumer smartwatches to detect atrial fibrillation in a real-world cohort of patients

Abstract Funding Acknowledgements Type of funding sources: None. Introduction Atrial fibrillation (AF) is the most common cardiac arrhythmia with an estimated lifetime risk of one in four. Multiple smartwatches capable to "screen" AF are presently available. The sensitivity and specificity for the detection of AF may differ between the available smartwatches, but this has not yet been adequately investigated. Methods We enrolled patients presenting to a cardiology service at a tertiary referral center in a prospective, observational study. The aim of this study was to assess and compare the accuracy of five smartwatches (Apple Watch 6, AliveCor Kardia Mobile, Fitbit Sense, Samsung Galaxy Watch 3 and Withings Scanwatch) in identifying AF compared to a nearly simultaneously acquired physician-interpreted 12-lead ECG in a real world cohort of patients. Each patient was assessed with at least 4/5 smartwatches. Secondary objective was to assess patient preference using smartwatches via predefined questionnaire. Results We prospectively enrolled 163 patients (32.9% female, mean age 64.7 years). AF was present in 47 patients (30.3%) at time of recording. We included 155 patients with 4 or 5 individual recordings for further analysis. Sensitivity and specificity for the detection of AF was similar between smartwatches: 90% and 95% for the Apple Watch 6, 97% and 96% for the AliveCor Kardia Mobile, 86% and 100% for the Fitbit Sense, 97% and 91% for the Samsung Galaxy Watch 3 and 91% and 95% for the Withings Scanwatch, respectively (Figure1). The rate of inconclusive tracings, meaning the algorithm was not able to determine the heart rhythm, was 17%, 26%, 21%, 20% and 24% for the Apple Watch 6, AliveCor Kardia Mobile, Fitbit Sense, Samsung Galaxy Watch 3 and Withings Scanwatch, respectively. Among inconclusive individual tracings from all devices, 63 tracings (40%) were due to high or low heart rate and 50 tracings (31%) due to motion artifacts. The inconclusive recordings were interpreted by blinded cardiologists to determine if these tracings are still clinically useful. By manual review the rhythm could be determined in 98.8% of 741 total individual recorded single-lead ECGs. Regarding patient acceptance, the Apple Watch was ranked first (37%) and the AliveCor Kardia Mobile last (7.6%) when patients were asked about their preferred device for daily rhythm monitoring. Conclusion In this clinical validation of five direct-to-consumer smartwatches we found a high dia