Precision Detection and Prediction of Atrial Arrhythmias Using Artificial Intelligence and Consumer Wearable Devices

Atrial fibrillation (AF) is the most prevalent cardiac arrhythmia affecting over one million people in the UK. It is associated with increased cardiovascular morbidity and mortality and costs the NHS between £1.4 billion and 2.5 billion annually. Current methods to detect AF include opportunistic pulse palpation, single time point 12-lead electrocardiograms (ECGs), ambulatory Holter monitoring, and implantable loop recorders (ILRs). The more widely used intermittent monitoring methods, such as ECGs and Holter monitoring, are limited in terms of duration and have lower detection yields of atrial arrhythmias. At the other end of the spectrum, the ILR can give continuous and accurate arrhythmia detection but is invasive and requires specialist expertise to implant, monitor, and analyse. In recent years, the use of wearable mobile health (mHealth) devices has emerged as a direct-to-consumer option for monitoring parameters such as heart rate and activity levels. From a clinical perspective they potentially offer a less invasive and cost-effective investigative approach, with remote monitoring solutions to possibly predict and detect AF. This technology has significant potential in terms of passive, non-invasive and continuous monitoring to aid the early diagnosis and management of AF. The original REMOTE-AF study (NCT05037136) developed novel methodology to detect AF using PPG-dervived data from a wearable. This study will further enhance this foundational work by recruiting patients to develop a AI-enabled, multi-parametric algorithm using PPG-derived data to detect AF.