Wearable Devices for Early Detection of Atrial Fibrillation: A Systematic Review and Meta-Analysis of Diagnostic Accuracy and Screening Performance

Abstract

Background: Atrial fibrillation (AF) is the most prevalent ongoing cardiac arrhythmia globally, with an associated high incidence of stroke, heart failure, and death. Early identification of atrial fibrillation will facilitate the initiation of anticoagulant therapy and significantly reduce the risk of thrombogenesis. Wearable technology, such as smartwatches and wristband monitor systems, has evolved into scalable modalities to identify atrial fibrillation both in the community and medically. However, because diagnostic performance varies widely across devices, a pooled quantitative evaluation is required to assess their diagnostic effectiveness.

Objective: This study aimed to evaluate the diagnostic accuracy of wearable devices for detecting atrial fibrillation and to synthesize evidence on screening performance using meta-analytic methods. Methods: The review and meta-analysis followed guidelines established by PRISMA (public reporting in a systematic way) and included studies that examined how well wearable devices identified AF using ECG or PPG only. Two random-effects models were used to calculate pooled sensitivity and specificity, and heterogeneity was assessed via I² statistics. Publication bias was assessed by means of funnel plots.

Results: There were 438,780 total patients in the analysis from 10 different studies. The combined sensitivity of wearable devices for detecting atrial fibrillation was 0.92 (95% CI: 0.87–0.95), with a combined specificity of 0.96 (95% CI: 0.91–0.98) using a random effects model. Significant heterogeneity was found for both the sensitivity (I² = 97.1%) and the specificity (I² = 98.0%). The funnel plot showed moderate asymmetry, meaning that there may have been bias in publishing results that did not show significance.

Conclusion: Wearable devices demonstrate high diagnostic accuracy for atrial fibrillation detection and represent a feasible approach for largescale screening. Standardization of algorithms and evaluation of longterm clinical outcomes remain critical for widespread clinical adoption.