Real-time smartphone alerts during atrial fibrillation episodes with implantable cardiac monitors and wearable devices: SMART-ALERT study.
Briosa E Gala A., Sharp AJ., Pope MTB., Leo M., Varini R., Paisey J., Curzen N., Banerjee A., Betts TR.
BACKGROUND: Atrial fibrillation (AF) is a major cause of stroke, with dynamic thromboembolic risk during and shortly after episodes. Implantable cardiac monitors and wearable devices have emerged as tools for real-time AF detection, yet their comparative performance in supporting anticoagulation strategies remains underexplored. OBJECTIVE: This single-center feasibility study investigated the performance of real-time AF detection and notification for episodes lasting longer than 30 minutes. METHODS: Phase 1 evaluated the integration of an implantable cardiac monitor and bespoke cloud-based software (SMART-ALERT) in sending automated short message service notifications (n = 50) for episodes uploaded within 24 hours. Phase 2 evaluated the Apple Watch (n = 23) and the CART Ring (n = 23) detecting and notifying participants of AF episodes. The primary outcome was the successful AF notification rates via (1) SMART-ALERT software and (2) wearable devices. Secondary outcomes included acknowledgment rates, notification times, and adherence. RESULT: Among 4943 AF episodes detected in 31 participants, the SMART-ALERT software successfully notified 511 of 691 eligible AF episodes (74%), with a 99.6% acknowledgment. In contrast, wearable devices showed poor notification performance: the Apple Watch identified 76 of 389 episodes (19.5%); and the CART Ring, 72 of 474 (15.1%). This performance difference was partly explained by suboptimal device adherence (Apple Watch: 66.3%; CART Ring: 23.9%), with 24.6% and 55.7% of AF episodes missed because of devices not being worn. CONCLUSION: The SMART-ALERT system demonstrated the feasibility of real-time AF detection and automated notifications, achieving a 74% notification success but facing important connectivity challenges. Wearable devices showed poor notification rates (<20%) and adherence, highlighting significant technical barriers to their current use in clinical AF monitoring.