Loading ...

A seizure-monitoring app developed for the Apple Watch detected tonic-clonic seizures (TCSs) with high sensitivity and a low false-alarm rate (FAR). 

Results of a multicenter phase 3 study of children and adults with epilepsy showed the EpiWatch app identified 98% of adjudicated TCSs and generated roughly one false alarm every 12 days during inpatient monitoring in findings that may help address a key limitation of existing wearable seizure-detection technologies. 

“The app demonstrates sensitivity and latency results comparable to other devices but with a substantially lower FAR,” wrote investigators led by Gregory L. Krauss, MD, professor of neurology at Johns Hopkins School of Medicine, Baltimore, Maryland.

“This improvement in FAR is anticipated to promote the long-term user adherence necessary for reliable caregiver alerts” and ultimately help mitigate sudden unexpected death in epilepsy (SUDEP) and other risks associated with TCSs, the investigators noted. 

The findings were published online May 27 in Neurology.

Wearable Detection Devices Recommended

Uncontrolled TCSs significantly increase the risk of SUDEP. However, research shows people with TCSs who share a bedroom have a threefold reduction in SUDEP risk, suggesting that timely caregiver intervention may mitigate this risk.

Based on this observation, the International League Against Epilepsy recommends wearable devices for TCS detection and alerting. Although other devices have adequate sensitivities, they have relatively high FARs that discourage consistent use and caregiver responses, and they may unintentionally disclose a medical condition that contributes to social stigma.

However, although existing devices generally offer adequate sensitivity, many are associated with relatively high FARs, which can undermine consistent use and reduce caregiver responsiveness. In addition, some devices may inadvertently disclose a patient's epilepsy diagnosis, potentially contributing to social stigma.

The EpiWatch app uses heart rate and motion data collected by the Apple Watch to identify patterns associated with TCS. When a seizure is detected, the app can send an alert to a designated caregiver's smartphone while allowing users or caregivers to cancel alerts triggered by false alarms.

Because FARs can discourage patients and caregivers from using seizure-alert devices consistently, investigators conducted a multicenter phase 3 trial to evaluate the sensitivity and false alarm rate of EpiWatch.

Phase 3 Results

The multicenter study included 242 children and adults with a history of TCSs or a clinical potential for TCSs who were undergoing inpatient video EEG monitoring at six epilepsy centers. Participants represented a broad range of ages, races, and ethnicities and had a mean age of 22.7 years.

Participants wore an Apple Watch equipped with the EpiWatch app throughout their inpatient monitoring, which lasted a median of about 60 hours, or roughly 2.5 days. An independent panel of three epileptologists, blinded to the app's detections, reviewed and classified seizure events. Of the 242 participants, 83 experienced seizures with motor features that were evaluated by the panel.

One of the study's two co-primary endpoints was positive percent agreement (PPA), which measured the sensitivity of the EpiWatch app for detecting TCSs compared with seizures classified as TCSs by the blinded expert panel. The other co-primary endpoint was the FAR.

Of 47 events classified as a TCS, 46 were detected by EpiWatch, for an overall sensitivity of 98% (CI, 95%-100%) and a conservatively adjusted overall PPA of 94%. The authors noted this is comparable to sensitivities reported for other wearable devices for TCS monitoring, which have ranged from 76% to 94%.

The app failed to detect one TCS, which investigators said may have occurred because a caregiver firmly restrained the participant's watch-wearing arm throughout the seizure. “All wrist-worn devices for TCS monitoring are susceptible to this source of false-negative detections, which does not interfere with seizure alerting,” the authors noted.

The second primary outcome was the FAR, which measured the daily frequency of false alarms. The EpiWatch app demonstrated a FAR of 0.08 per day (CI, 0.02-0.12), which translates to a false alarm every 12.4 days.

“This is an order of magnitude lower than the FARs reported for prospective studies of other wearable TCS monitoring devices” that ranged from 0.67 per day to 2.52 per day, the investigators noted. 

Fewer False Alarms

The authors noted that the low FAR may help minimize "alarm fatigue,” a phenomenon in which frequent alerts lead patients and caregivers to become less responsive to warnings.

Although previous studies have reported higher FARs in children than in adults, the FAR for EpiWatch was consistent across all age groups evaluated (5-12 years, 13-21 years, and 22 years and older). Sensitivity analyses also found no significant effect of race on FAR.

EpiWatch detected all TCSs that occurred during sleep, a finding the authors said is particularly important given the association between nocturnal TCSs and an increased risk for SUDEP. 

The median latency of TCS detection with the EpiWatch app (31.5 seconds) was comparable to detection latencies reported for other wearable devices.

No adverse events, both anticipated and unanticipated, were reported by any of the trial sites.

The authors noted that using a widely adopted consumer device such as the Apple Watch may reduce the stigma associated with wearable seizure monitoring. 

They added that the widespread use of Apple Watch means patients don’t require specialized training to operate the EpiWatch app effectively, encouraging long-term and regular use.

The authors acknowledged several limitations, including that all seizures were recorded in the controlled environment of an epilepsy monitoring unit, which may not reflect real-world performance. 

In addition, the findings only apply to TCSs in patients aged 5 years and older. Patients with Lennox-Gastaut syndrome and certain other populations were excluded, and the device was not specifically evaluated in patients with motor dysfunction, autonomic dysfunction, or cardiac pacemakers. 

Based on these trial results, EpiWatch received FDA 510(k) clearance and is available by prescription.

The study was funded by EpiWatch Inc. The investigators’ disclosures can be found in the original paper.