ARTIFICIAL INTELLIGENCE (AI)–ENABLED WEARABLES FOR CARDIAC RHYTHM MONITORING: DIAGNOSTIC PERFORMANCE, REGULATORY TRENDS, AND CLINICAL INTEGRATION.

Wearable technologies extend cardiac-rhythm monitoring beyond conventional clinical settings. Smartwatches, handheld single-lead electrocardiogram (ECG) devices, adhesive ECG patches, chest straps, rings, and remote-monitoring platforms increasingly use artificial intelligence (AI) and machine-learning methods to detect atrial fibrillation (AF), classify rhythm abnormalities, assess signal quality, prioritize ambulatory recordings, and estimate arrhythmia burden. This review examined the technological, regulatory, diagnostic, and clinical-integration considerations relevant to AI-enabled wearable cardiac monitoring. This review evaluates evidence concerning wearable ECG and photoplethysmography (PPG)-based rhythm-monitoring technologies, with attention to diagnostic performance, regulatory authorization, algorithmic limitations, and requirements for safe clinical implementation.
Wearable ECG devices can identify AF with high sensitivity and specificity in selected and supervised populations. However, diagnostic performance varies according to device modality, recording quality, patient characteristics, AF prevalence, reference standard, software version, and the handling of inconclusive or unclassifiable recordings. PPG-based irregular-rhythm notifications can provide useful screening signals but do not independently establish an AF diagnosis because they detect pulse irregularity rather than cardiac electrical activity. In contrast, interpretable wearable ECG recordings may provide rhythm documentation when incorporated into an appropriate clinical evaluation pathway.
Regulatory authorization defines a device’s permitted intended use but does not establish universal diagnostic accuracy, generalizability, clinical-outcome benefit, or suitability for autonomous clinical decision-making. Effective implementation requires external validation; transparent reporting of unclassifiable recordings; clinician-overread and confirmatory ECG pathways; governance of algorithm updates; privacy and cybersecurity protections; and ongoing post-market performance monitoring. AI-enabled wearable rhythm monitoring is most likely to improve care when used to support, rather than replace clinical judgment.