SMART-ALERT study: continuous cardiac monitoring and real-time automated smartphone alerts for atrial fibrillation episodes

Abstract Introduction 'Pill-in-the-pocket' oral anticoagulation (OAC) strategy is based on the concept that thromboembolic risk fluctuates—rising during and shortly after an atrial fibrillation (AF) episode and normalizing in sinus rhythm. Limiting OAC use in low-risk patients with infrequent episodes may offer comparable stroke protection, reduce bleeding risk, and enhance adherence. Safely implementing a 'pill-in-the-pocket' approach requires a closed-loop system that provides patients with automated real-time alerts during AF episodes, minimizing delays in starting OAC. Purpose This feasibility study assesses the performance of implantable cardiac monitors (ICM) and wearable devices (Apple Watch and CART-I ring) in accurately detecting AF episodes lasting over 30 minutes and delivering real-time alerts to participants' smartphones. Methods This single-centre feasibility study was conducted in two phases, each lasting 3 months (Figure 1) •Phase 1: Bespoke cloud-based software reviewed Carelink transmissions from the LINQ II ICM every 5 minutes, triggering SMS alerts only for the first AF episode lasting over 30 minutes within a 24-hour period. Participants acknowledged the alert by replying by replying via SMS to a designated study number. •Phase 2: Participants were assigned a wearable device that alerted their smartphone with a pop-up notification upon detecting AF episodes over 30 minutes. Participants acknowledged the alert by replying by replying via SMS to a designated study number. During this phase, the LINQ II ICM continued monitoring but did not send alerts. Results Fifty participants were prospectively enrolled, with a median age of 62 years; 30% were female, 76% had paroxysmal AF, and the median CHA2DS2-VASc and HAS-BLED scores were both 1. Forty-six patients progressed to phase 2 and were assigned a wearable device. Over a mean follow-up of 8.5±2.5 months, 1,065 AF episodes meeting alert criteria occurred in 32 participants. The LINQ II ICM achieved the highest rate of automated alerts sent within 24 hours at 74.0%, compared to 22.1% (p<0.001) for the watch and 23.4% (p<0.001) for the ring (Figure 2A). Similarly, acknowledgment rates for SMS alerts from the LINQ II ICM were markedly higher than those from the watch and ring pop-up notifications (99.4%, 43.5%, and 0, respectively) (Figure 2B). Failure to send SMS alerts from the LINQ II ICM was due to smartphones not uploading data to servers within 24 hours and connectivity issues between our software and Carelink. The lower detection rate from wearables may be partly attributed to intermittent monitoring and variability in signal quality. Conclusions This feasibility study demonstrates promising efficacy and compliance with LINQ II ICM when integrated with bespoke software. Wearables exhibited lower detection and acknowledgement rates, suggesting that their role in 'pill-in-the-pocket' OAC strategies may be limited without improvements in monitoring consistency. Figure 2