PulseAI wins FDA clearance for an AI that reads smartwatch ECGs and cuts the "inconclusive" pile

PulseAI said on Thursday, Nov. 5, 2026, that the U.S. Food and Drug Administration granted 510(k) clearance for its AI-powered arrhythmia analysis model for wearables, which analyzes ECGs recorded on smartwatches and other devices.

Smartwatches have turned ECGs into something people record in line at the grocery store, and every "inconclusive" reading lands on a clinician who has to review it, repeat it or call the patient back. A model that clears most of those up is less a gadget feature and more a staffing fix. The catch: this is one 296-person study, and PulseAI still has to get onto the watches and platforms people actually use.

The U.S. Food and Drug Administration granted 510(k) clearance for PulseAI's AI-powered arrhythmia analysis model for wearables. A 510(k) is the path the agency uses when it decides a device is close enough to one already allowed on the market. Arrhythmia means a heart rhythm that is not steady. The FDA database lists the device as PulseAPI SmartWatch, number K254226, with a decision date of Oct. 27, 2026. PulseAI announced the clearance on Nov. 5. The device name, the number, and the decision date are in the FDA database. The announcement is PulseAI's.

In a prospective study of 296 participants published in Heart Rhythm, PulseAI's algorithm improved the sensitivity of atrial fibrillation detection and cut inconclusive interpretations by 92% compared with the Apple Watch's native algorithm. Atrial fibrillation, or AFib, is an irregular rhythm in the heart's upper chambers. Sensitivity, here, is how often the model catches a rhythm that is really there. Native means the algorithm built into the watch. Inconclusive means the recording came back unresolved, so a person still has to sort it out. The unresolved share fell from 17.6% of recordings to 1.4%. Those figures are from the study, as PulseAI reported them.

PulseAI's release headline says the model delivers up to a 10-times performance improvement when it is used with leading wearables. That line is the company's claim.

The technology is built to work with compatible ECG devices and software platforms. An ECG, an electrocardiogram, is a tracing of the heart's electrical signal. Recordings are sent for automated analysis in the cloud, and structured results go back to the connected app for a clinician to review. Those lines are PulseAI's. A shopper does not buy this from PulseAI. It is meant to sit inside the devices and software a clinic already uses.

PulseAI cites IDC, a market-research firm: 164 million smartwatches were sold worldwide in 2025. That count is IDC's, as PulseAI reported it.

PulseAI was founded in 2020 and is based in Belfast, Northern Ireland. Innovation Ulster Limited, Ulster University's investment arm, led its recent investment round alongside the Mortara Group. Those lines are PulseAI's.

Dr. Alan Kennedy, chief executive and founder of PulseAI, said: "Wearables have made ECG recording widely accessible. Our ambition is to match expanded access with improved accuracy to help deliver a truly scalable solution to providers." He also said: "With FDA clearance, our focus now is bringing this capability into the devices and platforms in use by millions every day." Those quotations are his.

The picture shows PulseAI's arrhythmia analysis across devices: a smartwatch displaying an ECG trace, a phone running the PulseAI Smartwatch app with heart-rate and ECG readings, and a desktop monitor showing an ECG review screen. The credit is PulseAI.

In plain terms, the FDA cleared a Belfast company's model for reading heart-rhythm recordings from smartwatches and other wearables. In one study of 296 people, unresolved readings fell from 17.6% to 1.4% against the algorithm already on the Apple Watch. The result still has to be reviewed by a clinician, and the model still has to get onto the devices people actually wear.