Metric
Wearable ECG
Does your watch's irregular-heartbeat alert actually catch atrial fibrillation, and does catching it earlier actually help? Here's what the accuracy and screening research shows.
Consumer wearables now offer two related but distinct features: a single-lead ECG you can trigger yourself, and a background PPG-based algorithm that watches for an irregular pulse and alerts you when it finds one. Both are marketed around atrial fibrillation (AFib), the most common heart-rhythm disorder and a real stroke risk factor. The two questions worth separating are whether these features actually detect AFib accurately, and whether catching it this way actually changes outcomes. The evidence answers them very differently.
What your device actually calls this
Scientific measurementElectrocardiogram (electrical heart signal) or photoplethysmography (PPG, optical pulse-based irregular-rhythm detection). These are two different sensing methods, not two names for the same thing.
| Device | What they call it | What it actually measures |
|---|---|---|
| Apple Watch | "Irregular Rhythm Notification" and the "ECG app" (two separate features) | The notification is PPG-based passive background monitoring; the ECG app is a genuine single-channel ECG, FDA-cleared, requires actively holding the crown. Not the same feature and don't provide the same evidence quality. |
| Fitbit | "Fitbit ECG App" (branded the "Google ECG App" in newer materials), plus a separate passive "Irregular Heart Rhythm Notifications" feature | Sense 2 only. The ECG app is a genuine single-channel ECG, FDA-cleared and CE-marked, requiring an on-demand spot check. Irregular Heart Rhythm Notifications uses the same PPG sensor as everyday heart rate tracking to flag possible AFib passively, without an active check. |
| Samsung Galaxy Watch | "Irregular Heart Rhythm Notification" and "ECG" (two separate features) | FDA-cleared May 2023. Same passive/on-demand two-tier design as Apple and Fitbit; accuracy drops with motion or poor sensor contact, the same limitation every consumer ECG shares. |
| Amazfit | On-demand single-lead ECG | In three studies of Amazfit-made bands and watches, the ECG algorithm caught AFib 87 to 94 percent of the time, with specificity from 67 to 100 percent depending on the setting. Corrected 2026-10-09: this card previously credited 93.28 percent sensitivity from a technician-read study to automatic detection. |
| Withings ScanWatch | "ECG" | On-demand bezel-touch reading, FDA-cleared October 2021 for AFib detection specifically. |
The Apple Watch's "Irregular Rhythm Notification" and "ECG app" get conflated constantly in casual conversation, but they're different tools with different evidence bases. The notification is a passive background screen; the ECG app is an actively-triggered, FDA-cleared diagnostic-adjacent recording. Fitbit's Sense 2 runs the same two-tier design. Garmin, Oura, and Whoop don't offer ECG at all: Garmin hasn't pursued single-lead certification, Oura's ring form factor has no room for the electrode setup it requires, and Whoop's band doesn't include one either.
Device measurement accuracy: Established · on the Research DeskHow each device actually shows this
Apple Watch
"Irregular Rhythm Notification" and the "ECG app" (two separate features)
Fitbit
"Fitbit ECG App" (branded the "Google ECG App" in newer materials), plus a separate passive "Irregular Heart Rhythm Notifications" feature
Samsung Galaxy Watch
"Irregular Heart Rhythm Notification" and "ECG" (two separate features)
Amazfit
On-demand single-lead ECG
Withings ScanWatch
"ECG"
See every device →
Sources: Apple's own support page (support.apple.com) for the two-feature distinction, primary. Fitbit/Google's own product documentation and FDA clearance materials, primary.
More on Wearable ECG
Does Wearable ECG Accurately Detect Atrial Fibrillation?
Against a physician-read 12-lead ECG, wearable single-lead accuracy lands in the low-to-mid 90s for both sensitivity and specificity, across several devices.
Why Does Apple Watch's ECG Come Back Inconclusive So Often?
Apple Watch is one of several consumer ECG devices tested against physician-read readings, and every one produced a share of inconclusive tracings.
Seven Trials: AFib Screening Led to Only 7% Fewer Strokes
Seven trials, 76,458 people: screening led to 7% fewer strokes (RR 0.932, CI 0.873 to 0.996). Without the implant trial, not significant.
Does a Cuffless Blood Pressure Wearable Actually Work?
No blanket answer. A wearable ultrasound sensor passed clinical validation, while several wrist devices already on sale failed standard accuracy protocols.
Finding More AFib Has Not Clearly Meant Fewer Strokes
Screening raised AFib diagnoses (6.8% vs 5.4% in AMALFI) but stroke was 2.7% vs 2.5%. The big smartwatch study measured accuracy, not strokes.
Is AFib Overdiagnosed by Wearable Screening? A Real Estimate
A hypothesis-generating analysis found AFib screening nearly tripled diagnoses (OR 2.84), with overdiagnosis estimated at 8% to 29% of the extra cases.
Can AI Detect Hidden AFib Risk From a Normal ECG?
An algorithm reading a normal-rhythm ECG can flag elevated AFib risk, and a trial found previously unrecognised cases. It is a risk signal, not a diagnosis.
GUARD-AF Found More AFib and Not Fewer Strokes
GUARD-AF screened 11,905 adults aged 70 and over with a two-week ECG patch. It found more atrial fibrillation and no reduction in stroke hospitalization.
Fewer Strokes After AFib Screening Only With Risky Genes
In one analysis of the LOOP trial, AFib screening went with fewer strokes only at high genetic risk (HR 0.65). The authors call it hypothesis-generating.
Cardiologists Miss AFib on Single-Lead Strips About 1 in 4
Cardiologists reading single-lead ECGs found AFib 77.4% of the time (range 50% to 90.6%). Two reviewers agreed only moderately (kappa 0.48 to 0.58).
Can Your Blood Pressure Cuff Catch AFib You Didn't Know You Had?
AFib detection isn't only an ECG or PPG wearable's job. Several studies have tested whether an ordinary home blood pressure monitor can screen for it too.
A Borderline ECG Reading Predicted Later AFib in One Trial
In the VITAL-AF trial, a 'possible AFib' single-lead ECG reading carried nearly 2.5 times the risk of a later AFib diagnosis versus a normal reading.
Only Two Wrist Blood Pressure Watches Have Real Accuracy Data
A 2026 hypertension society review found only 2 wrist blood pressure watches sold, Omron HeartGuide and Huawei Watch D/D2, carry standard-grade validation.
Does Exercise Actually Raise or Lower Your Risk of AFib?
Endurance circles worry exercise can trigger AFib. A 93,669-person wearable study on guideline-level activity points the other way.
Can You Screen Yourself for AFib by Feeling Your Own Pulse?
Cardiology guidelines endorse self-pulse checks for AFib. A study built to test that recommendation against real ECG readings ran in 1,010 patients.
Apple Watch vs. Fitbit: Which Tracks Your Heart Rate More Accurately During a Dangerous Arrhythmia?
A hospital comparison fitted patients with both watches during monitored heart rhythm problems to see whose reading held up as the rhythm worsened.
What an Irregular Heart Rhythm Notification Actually Confirms, and What It Doesn't Yet
In the watch's own confirmatory research, only 34 percent of people notified of an irregular pulse were later found to have AFib on a medical-grade ECG patch.
Other metrics