Consumer ECG and Atrial Fibrillation Features

Start here: when a result is not a "wait and see" moment

A watch or phone cannot rule out a heart emergency. In the US, call 911 if you have a fast or irregular heartbeat together with chest pain, shortness of breath, fainting, weakness or numbness on one side of the body, trouble speaking or confusion. The NHS lists these among the symptoms for which UK readers should call emergency services. [1] Any other concern about a reading, a notification or your symptoms is a reason to talk with a doctor, not a reason to rely on the device.

What these features are

Atrial fibrillation (AF) is a heart rhythm problem. The NHS describes it as a problem with the electrical system that makes the heart beat, with symptoms that can include an irregular pulse and pounding, racing or fluttering; some people have no symptoms and it is found during a routine check-up. [1] Two different kinds of consumer features look for it:

An ECG records the heart's electrical activity. [2] A watch tracing is only one view of that signal. In one clinic study, a single-lead smartwatch ECG and a 6-lead handheld ECG were each compared with a 12-lead ECG; single-lead agreement was weaker for some measurements, such as the PR interval. [3]

What the FDA has authorized

In the US, "electrocardiograph software for over-the-counter use" is a Class II device type. The regulation says it creates, analyzes and displays ECG data, can provide information for identifying cardiac arrhythmias, and "is not intended to provide a diagnosis." [4] The FDA record for Apple's ECG App (DEN180044) lists this regulation (870.2345), product code QDA and a De Novo decision granted 09/11/2018. [5] Makers must report detection performance as sensitivity plus specificity or positive predictive value, and labeling must describe situations where the device may not perform as expected. [4]

This clears one function (flagging possible arrhythmia from an ECG recording) in the US. It does not cover other features on the same device, other countries, or the general wellness claims in marketing.

What studies show

Notifications from the pulse sensor (Apple Heart Study, 2019, funded by Apple). Of 419,297 participants, 0.52% received an irregular-pulse notification over a median of 117 days. Of 450 who returned an analyzable ECG patch, 34% had AF on the patch; the positive predictive value for a notification matching AF on simultaneous ECG was 0.84. [6] Most people were not notified, and most notified people who wore the patch later did not show AF during that wear period. Patches were worn about 13 days after notification on average. [6]

Head-to-head in a clinic (BASEL, 201 patients, 31% with AF, 12-lead ECG as reference). Sensitivity and specificity were 85% and 75% for the Apple Watch 6 and 85% and 75% for the Samsung Galaxy Watch 3, 58% and 75% for the Withings Scanwatch, 66% and 79% for the Fitbit Sense, and 79% and 69% for the AliveCor KardiaMobile. Between 17% and 26% of tracings were "inconclusive" depending on the device. [7] Results do not transfer between models or to newer versions.

Pooled results. A 2025 meta-analysis of 26 studies (17,349 patients) found overall sensitivity of 95% and specificity of 97%, with very high variation between studies (I2 of 95% and 99%). [8] A meta-analysis of the Apple Watch ECG alone (11 studies, 4,241 participants, mean age 62.56, 28% women) found sensitivity of 94.8% and specificity of 95%. [9] Pooled numbers come from varied settings, so they say less about any one watch in everyday life than they may seem to.

Accuracy is not the same as benefit

Detecting AF in a test is not proof that screening healthy people improves health. For adults 50 or older without symptoms, the US Preventive Services Task Force concluded in 2022 that evidence is insufficient to assess the balance of benefits and harms of screening for AF. It says clinicians should use clinical judgment. [10] That is insufficient evidence about screening, not a finding that screening is harmful.

Who should be careful

• People with a known heart rhythm condition or symptoms should follow their own clinician's plan, not a watch result.
• A normal watch result does not rule out AF or other problems, and an "AF" label is a prompt to see a clinician, who can confirm with a medical ECG. The NHS notes AF treatment aims to lower the risk of blood clots, stroke and heart failure. [1]
• Do not change or stop any medicine because of a watch reading.

How to judge a claim

Ask which exact model and version was tested, in whom, against what reference (a 12-lead ECG or a patch), and how many readings were inconclusive. See how validation against a clinical reference works and consumer data versus clinical measurements.