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

Aishwarya Arunkumar Kamble¹, Sreejith Vijayakumar², Selvy Ketanbhai Patel³, Maria Maqsood⁴, Zubair Ahmed⁵, Abdul Eizad Asif⁶, Imdad Ullah^7*

¹St. Mary Medical Center, Pennsylvania, USA

²Mercy St. Vincent Medical Centre, Ohio, USA

³M.K. Shah Medical College and Research Centre, India

⁴Quaid-e-Azam Medical College, Bahawalpur, Punjab, Pakistan

⁵Richmond University Medical Center, New York, USA

⁶Shalamar Medical and Dental College, Lahore, Pakistan

⁷Khyber Medical College, Pakistan

*Corresponding author: Imdad Ullah, Khyber Medical College, Pakistan.

Received: 19 April 2026; Accepted: 24 April 2026; Published: 28 May 2026

Introduction

Atrial fibrillation is the most common sustained arrhythmia and represents a significant healthcare burden worldwide. This condition affects more than 50 million people worldwide and is associated with five times the risk of ischemic stroke and two times the mortality rate [1,2]. Early detection of atrial fibrillation is important because prompt initiation of anticoagulation therapy can reduce the risk of stroke by more than 60% [3]. Traditional diagnostic methods rely on intermittent electrocardiograms performed during a doctor's visit. However, atrial fibrillation is often paroxysmal and asymptomatic, making it difficult to detect with short-term monitoring [4]. Therefore, continuous rhythm monitoring techniques have attracted increasing attention as a tool for early diagnosis and prevention of complications. Wearable devices with photoplethysmography and electrocardiogram sensors enable real-time rhythm monitoring outside the medical setting. These techniques provide continuous physiological data and enable early detection of irregular heart rhythms in large populations [5,6]. Large-scale population-based studies have demonstrated the feasibility of screening programs based on wearable devices. For example, the Apple Heart Study included over 400,000 participants and demonstrated that smartwatch-based monitoring can detect previously undiagnosed atrial fibrillation in a real-world setting [7]. Despite promising results, diagnostic performance varies by device and population due to differences in algorithm design, monitoring duration, and clinical parameters [8,9]. Some studies have reported high specificity but decreased sensitivity in postoperative populations, while others have shown excellent diagnostic yield in community screening programs [10]. These differences highlight the need to synthesize systematic evidence to determine the overall reliability of wearable atrial fibrillation detection technologies. Therefore, the purpose of this study was to conduct a systematic review and meta-analysis to evaluate the diagnostic accuracy of wearable devices for detecting atrial fibrillation and provide pooled estimates of sensitivity and specificity in different clinical populations.

Material and Method

This systematic review and meta-analysis were conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines [11]. The systematic literature search and study selection process, conducted in accordance with PRISMA 2020 guidelines, is illustrated in the flowchart. Following the identification of potential records and the removal of duplicates, studies were screened based on predefined inclusion criteria regarding wearable technologies, population characteristics, and diagnostic outcomes. After a rigorous assessment of full-text articles for eligibility and data availability, the screening process culminated in the inclusion of 10 studies involving a total of 438,780 participants for the final qualitative synthesis and meta-analysis.

Figure 1: Prisma Flowchart

Eligible studies included prospective or retrospective observational studies evaluating wearable devices capable of detecting atrial fibrillation in adult populations. Studies were required to use a recognized diagnostic reference standard, such as a 12-lead electrocardiogram or continuous telemetry monitoring. Only studies reporting diagnostic accuracy outcomes were included. Data extraction included study design, population characteristics, device type, monitoring technology, and diagnostic performance metrics. Sensitivity and specificity were selected as the primary outcomes because they represent the most clinically relevant measures of screening performance. Random-effects meta-analysis was performed using generalized linear mixed models to account for variability between studies. Heterogeneity was assessed using the I² statistic. Publication bias was evaluated using funnel plots.

Results

Characteristics of Included Studies

A total of ten studies involving 438,780 participants were included in the analysis. The studies represented a range of wearable technologies, including smartwatch electrocardiography systems, photoplethysmography-based monitoring devices, and hybrid wearable sensors. Study populations varied substantially across settings. Some studies evaluated community-based screening programs involving asymptomatic adults, while others focused on high-risk clinical populations such as patients undergoing cardiac surgery or cardioversion procedures. The largest study included in the analysis enrolled more than 419,000 participants using smartwatch-based irregular pulse notification algorithms, demonstrating the feasibility of large-scale digital health trials conducted outside traditional healthcare environments [7]. Several smaller validation studies evaluated wearable electrocardiography devices in controlled clinical settings. These studies reported high diagnostic accuracy and demonstrated the ability of wearable devices to detect atrial fibrillation in both symptomatic and asymptomatic individuals [12-14].

Table 1: PICO-structured characteristics of included studies

Diagnostic Accuracy

The pooled sensitivity of wearable devices for detecting atrial fibrillation was 0.92 (95% CI: 0.87–0.95) under a random-effects model. Substantial heterogeneity was observed across studies, with an I² value of 97.1%, indicating considerable variability in diagnostic performance.

Figure 2: Forest Plot of Sensitivity

The pooled specificity of wearable devices was 0.96 (95% CI: 0.91–0.98) under a random-effects model. High specificity indicates that wearable devices demonstrate a low rate of false-positive diagnoses across multiple clinical settings.

Figure 3: Forest Plot of Specificity.

Figure 4: Funnel Plot for Sensitivity

Figure 5: Funnel Plot for Specificity

Overall Effect Estimate

Meta-analysis of pooled risk ratios demonstrated consistent diagnostic performance across wearable devices, with statistically significant detection capability across studies.

Figure 6: Overall Effect / Risk Ratio Forest Plot

Discussion

This meta-analysis found that wearable devices demonstrated high overall diagnostic performance for detecting atrial fibrillation, with a pooled sensitivity of 0.92 and a pooled specificity of 0.96 under random-effects modeling. These findings support the growing view that wearable-based monitoring is no longer a speculative adjunct but a credible screening and case-finding tool for atrial fibrillation in both community and selected clinical settings [5-15]. The pooled results are broadly consistent with prior literature showing that wearable electrocardiographic and photoplethysmography systems can identify AF with clinically meaningful accuracy, especially when algorithms are combined with confirmatory ECG pathways [5-15]. The clinical importance of these findings lies in the fact that atrial fibrillation is frequently silent, intermittent, and therefore underdiagnosed. Traditional office-based ECG captures only a brief time window and may miss paroxysmal episodes entirely [3,4]. Wearable devices address this limitation by extending monitoring into daily life, allowing repeated or continuous rhythm assessment without requiring resource-intensive in-person encounters [7-20]. That matters because undetected AF can lead to preventable stroke, heart failure progression, and repeated hospital utilization [1-21]. The appeal of wearables is therefore not just convenience; it is the possibility of shifting AF detection upstream, before catastrophic events occur. At the same time, the pooled estimates should not be interpreted naively. The most obvious problem in this evidence base is heterogeneity. Both sensitivity and specificity showed very high I² values, meaning the included studies were not estimating one clean, uniform effect. That is not statistical noise; it reflects real differences in populations, device algorithms, data quality, monitoring conditions, and diagnostic thresholds. A smartwatch tested in a controlled cardiology clinic on patients already known to have AF is not equivalent to a passive PPG algorithm used by asymptomatic consumers in everyday life. Those are fundamentally different diagnostic environments, and collapsing them into a single pooled estimate risks overstating certainty if the heterogeneity is not openly acknowledged [12-15]. Another major issue is the difference between diagnostic accuracy and screening usefulness. Many studies show that a wearable can classify AF versus sinus rhythm with high sensitivity and specificity in a structured validation setting [12,13]. That is valuable, but it is not the same as proving benefit in a real screening program. A device can be technically accurate and still fail to improve clinical outcomes if uptake is poor, false alerts create unnecessary downstream testing, or detected low-burden AF does not translate into meaningful therapeutic benefit. This distinction has already surfaced in broader AF screening literature, where increased AF detection does not always lead to clear reductions in stroke or mortality at the trial level [16]. That gap between device performance and patient benefit is one of the central unresolved issues in this field. The role of the study setting also deserves attention. Community-based pragmatic studies such as the Apple Heart Study demonstrated the feasibility of scaling wearable monitoring to very large populations, which is a major strength from a public health perspective [7]. However, such studies often emphasize notification rates and positive predictive value rather than full diagnostic confusion matrices, making direct integration into conventional diagnostic meta-analysis more difficult. By contrast, hospital-based studies provide cleaner data and often yield more impressive performance metrics, but their populations are more selected and may exaggerate apparent accuracy relative to real-world screening conditions [10-13]. That tension between scale and diagnostic rigor is not a minor methodological inconvenience; it is the reason many AF wearable studies look impressive individually while remaining hard to compare cleanly in aggregate. Technology itself also matters. ECG-based wearables generally provide a more direct rhythm assessment and tend to perform strongly when signal quality is adequate, whereas PPG-based systems offer broader passive monitoring but may be more vulnerable to motion artifact, skin contact issues, and algorithmic misclassification [5-22]. This creates a practical trade-off. ECG-enabled wearables may produce more interpretable rhythm strips but depend on user activation or proper acquisition technique. PPG systems are more scalable and user-friendly for long-term screening but may sacrifice some precision under uncontrolled conditions. Future studies need to stop treating all wearables as interchangeable, because they are not. The device modality is part of the intervention, not a minor technical detail. The findings also have implications for implementation. High specificity is particularly important in screening programs because false-positive AF alerts can trigger cascades of anxiety, clinic visits, confirmatory testing, and unnecessary healthcare costs. The pooled specificity of 0.96 is encouraging, but again, this figure must be read in context. In low-prevalence general populations, even highly specific tools can generate substantial absolute numbers of false-positive alerts. Conversely, in enriched or postoperative populations, sensitivity becomes critically important because missed atrial fibrillation may carry immediate therapeutic consequences [10]. The acceptable trade-off between sensitivity and specificity is therefore context dependent. Screening healthy consumers is not the same as surveillance after cardiac surgery or rhythm monitoring after stroke. This review also reinforces the need for better standardization in the wearable AF literature. Too many studies use different reporting frameworks, inconsistent definitions of uninterpretable data, and selective emphasis on favorable metrics. Some report notification yield, others report positive predictive value, and others provide only per-recording classification performance. That inconsistency weakens comparability and makes pooled analysis more fragile than it should be. Future diagnostic studies should report complete 2×2 tables, predefine how unclassified traces are handled, specify whether analyses are per-patient or per-recording, and clearly separate diagnostic validation from screening-effectiveness claims [11-16]. The question of cost-effectiveness remains underdeveloped. The promise of wearables is often framed in economic terms—that scalable digital screening could reduce stroke burden and healthcare expenditures. That may be true, but the evidence is not mature enough to state it confidently. Cost-effectiveness depends not only on device accuracy but also on uptake, adherence, confirmatory testing pathways, anticoagulation decisions, baseline stroke risk, and health-system structure [16-24]. A highly accurate device is not automatically cost-effective. Until outcome-linked economic studies are stronger, claims about financial benefit should remain cautious. Several limitations should be acknowledged. First, heterogeneity across studies was substantial, which limits the precision of pooled estimates. Second, the studies included varied designs, settings, and reference standards, and some larger studies were not ideally structured for conventional diagnostic pooling. Third, funnel plot asymmetry suggests the possibility of publication bias, although visual inspection alone is not definitive. Fourth, some studies were conducted in highly selected cohorts, which may overestimate performance relative to general population screening. Finally, the current literature is much stronger for diagnostic accuracy than for downstream clinical outcome benefit, meaning that enthusiasm for wearable AF screening still outruns definitive evidence in some areas. Despite these limitations, the evidence now supports a clear conclusion: wearable devices have moved beyond novelty and can detect atrial fibrillation with high overall accuracy. The real challenge is no longer whether they can detect AF at all. The challenge is how they should be deployed, in whom, with what confirmatory pathway, and toward what clinical objective. The field needs fewer marketing-style claims and more rigorous studies linking wearable detection to stroke prevention, treatment decisions, patient-centered outcomes, and cost-effectiveness. Until then, wearable devices should be viewed as powerful screening and diagnostic support tools, not stand-alone replacements for structured clinical evaluation.

Limitations

This study has several limitations. First, substantial heterogeneity was observed across studies, reflecting differences in device algorithms and patient populations. Second, publication bias may have influenced pooled estimates. Third, long-term clinical outcomes such as stroke prevention and mortality were not consistently reported across studies.

Conclusion

Wearable devices demonstrate high sensitivity and specificity for detecting atrial fibrillation and represent a promising tool for large-scale screening and early diagnosis. Integration of wearable technologies into clinical practice should be guided by standardized validation protocols and supported by further research evaluating clinical effectiveness and economic impact.

References

1. Hindricks G, Potpara T, Dagres N, et al. 2020 ESC Guidelines for the diagnosis and management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS): The Task Force for the diagnosis and management of atrial fibrillation of the European Society of Cardiology (ESC) Developed with the special contribution of the European Heart Rhythm Association (EHRA) of the ESC. Eur Heart J 42 (2021): 373-498.
2. January CT, Wann LS, Calkins H, et al. 2019 AHA/ACC/HRS Focused Update of the 2014 AHA/ACC/HRS Guideline for the Management of Patients with Atrial Fibrillation: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines and the Heart Rhythm Society in Collaboration with the Society of Thoracic Surgeons. Circulation 140 (2019): e125-e151.
3. Healey JS, Connolly SJ, Gold MR, et al. Subclinical atrial fibrillation and the risk of stroke. N Engl J Med 366 (2012): 120-129.
4. Sanna T, Diener HC, Passman RS, et al. Cryptogenic stroke and underlying atrial fibrillation. N Engl J Med 370 (2014): 2478-2486.
5. Hannun AY, Rajpurkar P, Haghpanahi M, et al. Cardiologist-level arrhythmia detection and classification in ambulatory electrocardiograms using a deep neural network. Nat Med 25 (2019): 65-69.
6. Bumgarner JM, Lambert CT, Hussein AA, et al. Smartwatch algorithm for automated detection of atrial fibrillation. J Am Coll Cardiol 71 (2018): 2381-2388.
7. Perez MV, Mahaffey KW, Hedlin H, et al. Large-scale assessment of a smartwatch to identify atrial fibrillation. N Engl J Med 381 (2019): 1909-1917.
8. Freedman B, Camm J, Calkins H, et al. Screening for atrial fibrillation: a report of the AF-SCREEN International Collaboration. Circulation 135 (2017): 1851-1867.
9. Guo Y, Wang H, Zhang H, et al. Mobile photoplethysmographic technology to detect atrial fibrillation. J Am Coll Cardiol 74 (2019): 2365-2375.
10. Schreier K, Borger M, Sepehri Shamloo A, et al. Feasibility and reliability of a smartwatch to detect atrial fibrillation after cardiac surgery: a prospective study. Front Digit Health 7 (2025): 1718350.
11. Page MJ, McKenzie JE, Bossuyt PM, et al. The PRISMA 2020 statement: an updated guideline for reporting systematic reviews. BMJ 372 (2021): n71.
12. Abu-Alrub S, Strik M, Ramirez FD, et al. Smartwatch electrocardiograms for automated and manual diagnosis of atrial fibrillation: a comparative analysis of three models. Front Cardiovasc Med 9 (2022): 836375.
13. Campo D, Elie V, de Gallard T, et al. Atrial fibrillation detection with an analog smartwatch: prospective clinical study and algorithm validation. JMIR Form Res 6 (2022): e37280.
14. Ploux S, Strik M, Caillol T, et al. Beyond the wrist: using a smartwatch electrocardiogram to detect electrocardiographic abnormalities. Front Cardiovasc Med 115 (2022): 29-36.
15. Barrera N, Solorzano M, Jimenez Y, et al. Accuracy of smartwatches in the detection of atrial fibrillation: a systematic review and diagnostic meta-analysis. JACC Adv 4 (2025): 102133.
16. Wahab A, Nadarajah R, Larvin H, et al. Triple cardiovascular disease detection using artificial intelligence from ambulatory ECGs and risk factors: a prospective cohort study. Lancet Reg Health Eur 53 (2025): 101298.
17. Svennberg E, Engdahl J, Al-Khalili F, et al. Mass screening for untreated atrial fibrillation: the STROKESTOP study. Circulation 131 (2015): 2176-2184.
18. Steinhubl SR, Waalen J, Edwards AM, et al. Effect of a home-based wearable continuous ECG monitoring patch on detection of undiagnosed atrial fibrillation: the misstops randomized clinical trial. JAMA 320 (2018): 146-155.
19. Turakhia MP, Desai M, Hedlin H, et al. Rationale and design of a large-scale, app-based study to identify cardiac arrhythmias using a smartwatch: the Apple Heart Study. Am Heart J 207 (2019): 66-75.
20. Halcox JPJ, Wareham K, Cardew A, et al. Assessment of remote heart rhythm sampling using the AliveCor heart monitor to screen for atrial fibrillation: the REHEARSE-AF study. Circulation 136 (2017): 1784-1794.
21. Diederichsen SZ, Haugan KJ, Køber L, et al. Atrial fibrillation screening in the general population: a systematic review and meta-analysis. Eur Heart J 127 (2020): 143-154. 
22. McManus DD, Lee J, Maitas O, et al. A novel application for the detection of an irregular pulse using an iPhone 4S in patients with atrial fibrillation. Heart Rhythm 10 (2013): 315-319.
23. Koshy AN, Sajeev JK, Nerlekar N, et al. Smart watches for heart rate assessment in atrial arrhythmias. Trends Cardiovasc Med 28 (2018): 442-448.
24. Freedman B, Martinez C, Katholing A, et al. Residual risk of stroke and death in anticoagulated patients with atrial fibrillation. Heart 46 (2015): 2523-2528.
25. Chan PH, Wong CK, Poh YC, et al. Diagnostic performance of a smartphone-based photoplethysmographic application for atrial fibrillation screening in a primary care setting. J Am Heart Assoc 5 (2016): e003428.
26. Kamel H, Okin PM, Elkind MS, et al. Atrial fibrillation and mechanisms of stroke: time for a new model. Lancet Neurol 16 (2017): 183-194.
27. Lubitz SA, Yin X, Rienstra M, et al. Screening for atrial fibrillation in older adults at primary care visits: the VITAL-AF randomized controlled trial. Am Heart J 215 (2019): 147-156.
28. Guo Y, Lip GYH, Apostolakis S. The CHADS₂ and CHA₂DS₂-VASc scores for stroke risk assessment in atrial fibrillation: a systematic review and meta-analysis. Lancet 10 (2013): 258-266.
29. Reed MJ, Grubb NR, Lang CC, et al. The ROSE (Risk Stratification of Syncope in the Emergency Department) study. Heart 55 (2010): 713-721.
30. Lowres N, Neubeck L, Redfern J, et al. Screening to identify unknown atrial fibrillation. A systematic review. Thromb Haemost 110 (2013): 213-222.
31. Varma N, Stambler B, Chun S. Detection of atrial fibrillation with an implantable loop recorder in patients with cryptogenic stroke. Circulation 60 (2020): 102-106.
32. Kirchhof P, Benussi S, Kotecha D, et al. 2016 ESC Guidelines for the management of atrial fibrillation developed in collaboration with EACTS. Eur Heart J 37 (2016): 2893-2962.
33. Proesmans T, Mortelmans C, Van Haelst R, et al. Atrial fibrillation screening with photo-plethysmography through a smartphone camera. Europace 21 (2019): 1160-1168.
34. Giebel GD, Gissel C. Accuracy of mHealth devices for atrial fibrillation screening: systematic review. JMIR Mhealth Uhealth 7 (2019): e13641.
35. Lau JK, Lowres N, Neubeck L, et al. iPhone ECG application for community screening to detect silent atrial fibrillation: a novel technology to prevent stroke. Int J Cardiol 165 (2013): 193-194.
36. Guo Y, Chen Y, Lane DA, et al. Mobile photoplethysmographic technology to detect atrial fibrillation. Int J Cardiol 74 (2019): 2365-2375.
37. Svennberg E, Tjong F, Goette A, et al. How to use digital devices to detect and manage arrhythmias: an EHRA practical guide. Europace 24 (2022): 979-1005.
38. Freedman B, Camm J, Calkins H, et al. Screening for atrial fibrillation: a report of the AF-SCREEN International Collaboration. Circulation 135 (2017): 1851-1867.
39. Turakhia MP, Desai M, Hedlin H, et al. Rationale and design of the Apple Heart Study: a large-scale, app-based study to identify cardiac arrhythmias using a smartwatch. Circulation 207 (2019): 66-75.
40. Guo Y, Chen Y, Lane DA, et al. Mobile photoplethysmographic technology to detect atrial fibrillation. J Am Coll Cardiol 74 (2019): 2365-2375.