Smartwatch AFib Alerts: What an Irregular Rhythm Notification Means
Editorial: ONDA Life (Yakiv Bilenko) · Updated November 18, 2026

A smartwatch irregular rhythm notification is a prompt to get an ECG, not a diagnosis of atrial fibrillation. European and American guidelines say that pulse-sensor (PPG) wearables are not reliable enough to diagnose it on their own. In large studies, notifications were rare, and AF was later confirmed in only some people. Whether such screening prevents strokes is not known. HRV readings cannot detect AF.
Key points
- The European guideline requires ECG documentation before AF is diagnosed or treated, and this excludes non-ECG wearables.
- The American guideline says a pulse-sensor alert may prompt an ECG but is not reliable enough to diagnose AF.
- In large studies of Apple and Fitbit watches, few people received a notification.
- When a notification came while an ECG patch was worn, it usually matched AF on the ECG; on a patch worn later, AF was found in about a third of people.
- Watch ECG recordings are a separate tool and were accurate in pooled studies, especially when read by trained staff.
- Whether watch screening prevents strokes has not been shown.
- HRV, RMSSD and SDNN do not detect AF, and ONDA does not screen for it.
What is an irregular rhythm notification?
Many smartwatches use an optical pulse sensor (photoplethysmography, PPG) to check the timing of your pulse in the background. If the timing looks irregular several times in a row, the watch may show an irregular rhythm notification suggesting possible atrial fibrillation (AF), a common heart rhythm disorder. In short: the notification is a reason to get an ECG, not a diagnosis [S1] [S2].
This page describes what studies and guidelines say. It is not medical advice. If you have symptoms such as chest pain, fainting or severe shortness of breath, seek urgent care.
Why do guidelines require an ECG?
The European guideline requires ECG documentation before AF risk is assessed or treatment starts, to avoid wrong diagnoses [S1]. It states that this does not include non-ECG wearables that use PPG [S1]. The American guideline agrees: a PPG alert may prompt you to get an ECG, but it is not reliable enough to diagnose AF [S2].
What did large studies find?
Notifications were rare. In a large study of an Apple Watch notification, few participants received one over months of monitoring [S3]. The Fitbit Heart Study reported a similar picture [S4].
Two different numbers. When a notification came while an ECG patch was being worn, it usually matched AF on that ECG [S3] [S4]. But when people wore a patch after the notification, AF was found in only about a third of them [S3] [S4]. Both are true: the alert is often right about that moment, but AF can come and go, and not everyone notified had AF on a later recording. Both studies were supported by the device makers and each tested one device.
Screening, not diagnosis. A systematic review concludes that PPG devices are best used for screening, followed by ECG confirmation according to guidelines [S5]. It lists motion artifacts and false positives among the limits [S5].
How is a watch ECG different?
Some watches can also record a single-lead ECG when you hold a finger on the crown or frame. That is a separate tool from the background pulse notification. In a meta-analysis, smartwatch ECG recordings detected AF with high accuracy, especially when trained staff read them, though some recordings could not be read [S6]. Another meta-analysis found that accuracy varied between devices and studies, so a fair ranking of watch brands is not possible from it [S7].
What does the evidence show?
Guideline: a PPG alert is not a diagnosis. Both major guidelines require an ECG [S1] [S2].
Context-dependent: notifications in large studies. Few people are notified; a later ECG confirms AF in only some of them; alerts during simultaneous ECG usually match [S3] [S4].
Context-dependent: watch ECG. Accurate in pooled studies, best with expert reading [S6].
Unknown: strokes and outcomes. Whether screening with a watch prevents strokes or other harms has not been shown [S5].
What it does not tell you
- Not a diagnosis. Only an ECG can establish AF [S1] [S2].
- Not reassurance. No notification does not rule out AF; the feature checks only at some moments.
- Not proof of benefit. Stroke prevention through watch screening is unproven [S5].
- Not about HRV. HRV, RMSSD and SDNN describe variation between beats and depend heavily on how they are recorded [S8]. They are not a test for AF.
In ONDA
ONDA builds a personal baseline from HRV, resting heart rate and breathing rate stored in Apple Health [S9]. ONDA does not detect, screen for or diagnose AF and is not a medical device. If your watch shows an irregular rhythm notification, talk to a doctor about getting an ECG. How ONDA measures your signals is explained in How ONDA measures and interprets your signals.
Educational information, not a diagnosis or medical treatment.
Evidence at a glance
| Claim | Evidence | Limitation |
|---|---|---|
| The European guideline requires ECG documentation to start AF risk assessment and management. [S1] | Guideline / expert consensus | Guideline recommendation (expert consensus). |
| The European guideline states that this ECG documentation does not include non-ECG wearables such as pulse-sensor (PPG) devices. [S1] | Guideline / expert consensus | Guideline recommendation (expert consensus). |
| The American guideline says pulse-sensor alerts may prompt an ECG but are not reliable enough to diagnose AF. [S2] | Guideline / expert consensus | Guideline recommendation (expert consensus). |
| In a large study of an Apple Watch irregular pulse notification, few participants received a notification. [S3] | Context-dependent | Single manufacturer-supported study in self-selected users of one device and algorithm. |
| In the Apple study, AF was found on an ECG patch worn after the notification in about a third of notified participants. [S3] | Context-dependent | Only a minority of notified participants returned a patch, applied on average about two weeks later; AF that comes and goes may be missed. |
| In the Apple study, notifications that occurred while the ECG patch was worn usually matched AF on the ECG. [S3] | Context-dependent | Agreement during simultaneous recording; not the same as the share of notified people who have AF. |
| In the Fitbit Heart Study, AF was present on a later ECG patch in about a third of notified participants with analyzable patches. [S4] | Context-dependent | Single manufacturer-supported study of one algorithm; only part of notified participants had analyzable patches. |
| In the Fitbit Heart Study, a detection that occurred during ECG patch monitoring almost always matched AF on the ECG. [S4] | Context-dependent | Simultaneous agreement in a subgroup; not a population screening yield. |
| A systematic review concludes that pulse-sensor devices are best used for screening, with ECG confirmation per guidelines. [S5] | Context-dependent | Systematic review without a pooled estimate for screening outcomes. |
| The same review lists motion artifacts, false positives and limited evidence for stroke prevention among the limitations. [S5] | Unknown | Whether watch screening prevents strokes is not established by the studies reviewed. |
| In a meta-analysis, smartwatch single-lead ECG recordings detected AF with high accuracy, especially when read manually by trained staff. [S6] | Context-dependent | Diagnostic accuracy studies, often in selected patients; some recordings could not be read. |
| A diagnostic meta-analysis found that accuracy varied between devices, with very high heterogeneity, so it does not allow a fair device ranking. [S7] | Emerging | Very high heterogeneity between studies; the article has a published correction; its numbers are not used here. |
| HRV values depend on the input signal, recording length, breathing and analysis method. [S8] | Guideline / expert consensus | Methodological guideline (expert consensus) for research recordings. |
| ONDA builds a personal baseline from values in Apple Health. [S9] | Established | Describes how the app works only; not evidence for any health claim. |
Sources
- [S1] Van Gelder et al. (2024). 2024 ESC Guidelines for the management of atrial fibrillation developed in collaboration with the European Association for Cardio-Thoracic Surgery (EACTS). European Heart Journal. DOI 10.1093/eurheartj/ehae176 · PMID 39210723
- [S2] Joglar et al. (2024). 2023 ACC/AHA/ACCP/HRS Guideline for the Diagnosis and Management of Atrial Fibrillation: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. DOI 10.1161/CIR.0000000000001193 · PMID 38033089
- [S3] Perez et al. (2019). Large-Scale Assessment of a Smartwatch to Identify Atrial Fibrillation. New England Journal of Medicine. DOI 10.1056/NEJMoa1901183 · PMID 31722151
- [S4] Lubitz et al. (2022). Detection of Atrial Fibrillation in a Large Population Using Wearable Devices: The Fitbit Heart Study. Circulation. DOI 10.1161/CIRCULATIONAHA.122.060291 · PMID 36148649
- [S5] Bin Rashid et al. (2026). Wearable devices for atrial fibrillation: diagnostic and screening roles of ECG and PPG-A systematic review. Future Cardiology. DOI 10.1080/14796678.2026.2689050 · PMID 42332956
- [S6] Iqhrammullah et al. (2025). Accuracy and interpretability of smartwatch electrocardiogram for early detection of atrial fibrillation: A systematic review and meta-analysis. Journal of Arrhythmia. DOI 10.1002/joa3.70087 · PMID 40406413
- [S7] Barrera et al. (2025). Accuracy of Smartwatches in the Detection of Atrial Fibrillation: A Systematic Review and Diagnostic Meta-Analysis. JACC: Advances. DOI 10.1016/j.jacadv.2025.102133 · PMID 41182224
- [S8] Carter et al. (2026). Guidelines for rigor and reproducibility of heart rate variability within human cardiovascular research. Am J Physiol Heart Circ Physiol. DOI 10.1152/ajpheart.00041.2026 · PMID 42495990
- [S9] ONDA — product documentation: How ONDA works. How ONDA works.
Related
How ONDA Science pages are made: every number comes from one checked list of facts, every claim is mapped to its sources and graded by strength of evidence, and sources need a DOI or PMID (manufacturer documentation is used only for device facts).