Smartwatch ECG AI Wins FDA Clearance to Spot AFib

PulseAI said on Thursday that the United States Food and Drug Administration has cleared its smartwatch ECG AI for analysing heart rhythms recorded on wearable devices. The Belfast company announced the clearance on November 5, 2026, and the FDA database lists the device as PulseAPI SmartWatch under number K254226, with a decision date of October 27. The model reads electrocardiograms from smartwatches and other wearables and flags possible arrhythmia for clinical review.

The clearance covers an artificial intelligence model built to sit behind the ECG apps people already use. PulseAI designed the technology for integration with compatible ECG devices and software platforms: recordings go to the cloud for automated analysis, and structured results return to the connected application for a clinician to review. The company has not named the devices or platforms that will carry the model, and its chief executive said the focus now is bringing the capability into the devices and platforms in use by millions every day, calling the clearance "an important step toward that goal."

What the study found

The evidence behind the clearance comes from a prospective study of 296 participants published in the journal Heart Rhythm. In that study, PulseAI's algorithm improved the sensitivity of atrial fibrillation detection and cut inconclusive interpretations by 92 percent compared with the Apple Watch's native algorithm. The share of recordings left unresolved fell from 17.6 percent to 1.4 percent. The published study reports that the model delivers up to tenfold performance improvement when used with leading wearables.

Atrial fibrillation, often shortened to AFib, is the most common sustained heart rhythm disorder, affecting tens of millions of people worldwide. It can occur without symptoms and, when untreated, is linked to around a fivefold higher risk of ischaemic stroke, depending on a person's individual risk factors. Finding AFib matters because it lets clinicians assess stroke risk and consider treatment before a clot causes harm. A test that leaves fewer readings unresolved means fewer patients called back for repeat recordings and fewer borderline results waiting for a specialist to sort through.

Why inconclusive readings clog the system

Smartwatches have put ECG recording in pockets and on wrists around the world. According to IDC, 164 million smartwatches were sold worldwide in 2025, and most models now offer some form of heart rhythm monitoring. That spread has pushed heart rhythm recording well beyond hospitals and clinics, but the clinical labour needed to interpret the recordings has not kept pace. Every inconclusive result can mean manual review, a repeat recording, or a follow-up call with the patient.

The benchmark in the study was the algorithm built into the Apple Watch, whose ECG app received FDA marketing authorisation in 2018. When that algorithm cannot classify a tracing, it returns an inconclusive result, leaving the wearer without an answer and the reading in a queue for human review. PulseAI's model was tested against that same setup, and the trial measured how many recordings each approach left unresolved.

PulseAI's pitch is that cutting the unresolved pile lets clinical services handle more patient-generated ECGs with the staff they already have. The model does not replace a clinician: results are structured for review inside a connected application, and a clinician still makes the call on what a reading means. That distinction matters for a tool aimed at providers rather than sold directly to watch owners, since a cleared algorithm still has to earn trust in everyday practice before it changes how many readings reach a doctor's desk.

What happens next

Clearance is a starting point, not a launch. For now, the model exists as a cleared component for device makers, not as an app watch owners can download. The 510(k) route the FDA used means the agency found the model substantially equivalent to a device already allowed on the market, rather than approving it from scratch. PulseAI still has to strike deals with device makers and platform owners, and no launch date or named partner was included in the announcement. The evidence also rests on a single study, so independent confirmation and real-world results will shape how far the model spreads.

The company behind the model was founded in 2020 in Belfast, Northern Ireland, and has built deep learning algorithms trained on millions of ECG recordings, validated across smartwatches, patches, hospital monitors, and defibrillators. Its recent investment round was led by Innovation Ulster Limited, the investment arm of Ulster University, alongside the Mortara Group. Professor Paul Bartholomew, vice chancellor at Ulster University, called the clearance "a significant achievement for the company, the University and Northern Ireland" in a statement issued with the announcement.

The FDA has cleared several first-of-their-kind treatments in recent weeks, including a hepatitis B drug that offers some patients a shot at stopping daily medicine, as reported in earlier coverage. Readers interested in more health coverage can find it on the Health topic page. For the full announcement, see the PulseAI press release and the Morningstar republication of the announcement.