Abstract
BACKGROUND: Atrial fibrillation (AF) is the most prevalent arrhythmia and a significant risk factor for stroke and heart disease, making early diagnosis crucial for prevention. Although several smartwatches can detect AF, their accuracy varies, and there is limited information regarding their effectiveness. OBJECTIVES: The purpose of this study was to analyze the overall accuracy of current available smartwatches in identifying AF. METHODS: A comprehensive systematic literature search was performed across major databases for articles published until January 2025. The specificity, sensitivity, and area under the curve (AUC). Statistical analysis was conducted using R software. The review protocol was registered in PROSPERO (CRD420251007932). RESULTS: Twenty-six studies were included in the study, including 17,349 patients. Overall sensitivity was 95% (95% CI: 92%-97%; I(2) = 95%) and specificity 97% (95% CI: 94%-98%; I(2) = 99%), with a pooled AUC of 0.97 (95% CI: 0.96%-0.99%). Device performance varied, with the Apple Watch achieving 94% sensitivity (95% CI: 89%-96%; I(2) = 71%) and 97% specificity (95% CI: 93%-99%; I(2) = 87%), Samsung devices yielding 97% sensitivity (95% CI: 92%-99%; I(2) = 92%) and 96% specificity (95% CI: 91%-98%; I(2) = 92%), and the Withings Scan Watch showing 89% sensitivity (95% CI: 75%-96%; I(2) = 94%) and 95% specificity (95% CI: 88%-99%; I(2) = 93%). Accuracy was comparable between photoplethysmography and electrocardiogram-based models. CONCLUSIONS: Smartwatches possess excellent diagnostic accuracy for AF detection and may, therefore, represent a suitable and fitting option for patients.