From Clinic to Cloud: A Comprehensive Systematic Review on AI-Driven Wearable Devices for Atrial Fibrillation - Bridging Digital Health and Clinical Practice
Divya Ravikumar
ABSTRACT
Wearable devices (smartwatches, rings, ECG patches) are increasingly used to screen for atrial fibrillation (AF). Recent large-scale studies report high diagnostic accuracy for wearable AF detection: pooled analyses find smartphone and smartwatch-based AF algorithms often >90% sensitive and specific. For example, the Apple Heart Study (419,297 users) found an 84% positive predictive value (PPV) for smartwatch-detected AF, and the Fitbit Heart Study (455,269 users) found a 98.2% PPV. Wearable ECG patches (e.g. Zio XT) detect substantially more AF than 24h Holter monitors. Many wearables (Apple Watch, Fitbit, Samsung, Withing’s, AliveCor) now have FDA (or CE) clearance for AF detection. However, most screen-detected AF is low-burden (median ~0.5% of time), and the net clinical benefit of mass screening is unproven. Challenges include false positives from motion or ectopy, data privacy/security, interoperability, and equitable access. While preliminary trials suggest possible stroke-prevention benefits, large randomized outcome trials are still needed. This review (PRISMA-compliant) summarizes current evidence on wearable AF screening – covering device technologies, diagnostic accuracy, trial results, and implementation issues – to inform researchers and clinicians about this emerging paradigm [1-10].
