Smartwatch app detects seizures with low rate of false alarms

Notice: News releases are not subject to review by MedLink Neurology’s Editorial Board.

• In people with epilepsy, a smartwatch app accurately detected tonic-clonic seizures with a low rate of false alarms.
• Tonic-clonic seizures involve major convulsions. Following such a seizure, people briefly lose muscle tone and can have airway obstructions that increase the risk for sudden unexpected death in epilepsy (SUDEP).
• For the study, participants were monitored for just over two days on average using both video-electroencephalograph and the app.
• Researchers compared the results and found the app detected 46 of 47 tonic-clonic seizures with a false-alarm rate 90% lower than the rate reported for other seizure detection devices.
• The authors say having a device that is easy to wear with fewer false alarms can promote consistent use and deliver dependable alerts to caregivers.
• They note this is important because seizure detection and alerts can help caregivers render first aid, potentially reducing the risk of SUDEP.

In people with epilepsy, a new study has found that a smartwatch application accurately detected tonic-clonic seizures, seizures with major convulsions, with a low rate of false alarms. The study was published May 27, 2026, in Neurology^® Open Access, an official journal of the American Academy of Neurology.

“For people who have uncontrolled tonic-clonic seizures, which can include a loss of muscle tone and airway obstruction, the risk of sudden unexpected death in epilepsy, called SUDEP, is high, particularly for those who sleep alone,” said study author James W. Wheless MD, of Le Bonheur Children's Hospital in Memphis, Tennessee, and a Fellow of the American Academy of Neurology. “Wearable seizure detection devices can alert caregivers so they can provide first aid, but some devices have high false alarm rates that can discourage use and timely intervention. Our study found the EpiWatch smartwatch app detected nearly all tonic-clonic seizures in participants and had a lower rate of false alarms.”

The study included 242 adults and children with epilepsy, with an average age of 23, who had tonic-clonic seizures or the potential to have such seizures. Participants were each monitored for an average of two-and-a-half days in a specialized hospital unit with video-electroencephalograph (EEG) monitoring to track their seizures. Neurologists who specialize in epilepsy classified tonic-clonic and other types of seizures based on the video-EEG monitoring.

A total of 83 people had seizures that involved muscle activity, and 37 experienced at least one tonic-clonic seizure.

Participants also wore smartwatches with the EpiWatch app to track tonic-clonic seizures. Researchers compared video-EEG results with results from the app.

Researchers found the app detected 46 of 47 tonic-clonic seizures, noting that one seizure was missed due to a caregiver restraining a participant's arm.

After over 16,000 hours of study monitoring, there were 56 false alarms, a rate of 0.08 per day, equivalent to one false alarm every 12.4 days. Researchers determined this rate was 90% lower than the rate of other devices, which had a range of 0.67 to 2.52 per day.

Of participants, 87% had no false alarms, 9% had one, and 4% had two or more. Of the false alarms, 35 were associated with activities such as playing video games or other repetitive movements. There were no adverse effects.

The app had an overall sensitivity of 98%, compared to other devices that had a range of 76% to 94%.

“Wearing some seizure monitoring devices may carry a stigma, but using a common smartwatch with an app does not, which is important for encouraging regular use,” said Wheless. “Being prescribed an app with fewer false alarms can encourage long-term use and provide reliable caregiver alerts to help reduce sudden unexpected deaths and other risks associated with tonic-clonic seizures.”

A limitation of the study was that all seizures occurred in the controlled environment of an epilepsy monitoring unit and may not represent the variety of conditions in which people experience seizures.

Source: News Release
American Academy of Neurology
May 28, 2026