NASHVILLE, Tennessee ― Scheduling video-electroencephalography (vEEG) monitoring on the basis of personalized seizure risk forecasts may improve diagnostic accuracy, emerging research suggests.
In a validation study, the use of personalized seizure forecasts that were based on data from an electronic diary increased the yield of vEEG for participants.
If all works out in further studies, this approach could help determine the best time for patients to present to the epilepsy monitoring unit (EMU), researchers note. And adding heart-rate data from a smartwatch could optimize this timing even more, they add.
Co-investigator Philippa Karoly, PhD, senior lecturer, Department of Biomedical Engineering, University of Melbourne, Australia, told Medscape Medical News the new results should address skepticism about whether seizure forecasting that is based on high-risk periods is accurate or useful.
“Even if we can increase the yield of EMU stays by 20%, that’s a pretty massive saving of people’s time and of hospital costs,” said Karoly
“We’re not changing someone’s standard care; we’re not changing their treatment; we are just trying to shift the timing of their treatment so it’s better for everyone involved,” she added.
The findings were presented here at the American Epilepsy Society (AES) 76th Annual Meeting 2022.
Seizure Activity Can Be Rare
Monitoring brain activity using vEEG is a “gold standard” method of diagnosing epilepsy. Such monitoring is also used to plan epilepsy-related surgery.
However, a diagnosis of epilepsy requires a record of a seizure, which “can be quite rare” in the monitoring unit, said Karoly. “In up to a third of cases, people are going through that long process [of] coming into the hospital, taking up a bed, getting set up. And then they don’t get the recordings they need; they don’t get a seizure,” she said.
Low vEEG yields necessitate repeated monitoring, which leads to delayed diagnoses. Increasing these yields could mean not only more rapid diagnoses but also improved quality of life and cost savings.
Previous research has suggested that an individual’s multiday seizure cycle activity can help forecast seizure risk ― and that adding a smartwatch that measures heart rate may improve the computation as to where a patient is in the cycles.
The aim is to determine when there will be periods of high seizure risk “and to see what happens if we schedule monitoring for that high-risk period,” said Karoly
Her poster presentation had three parts ― a baseline study, a validation study, and a prospective study.
Baseline, Validation Data
In the baseline study, researchers extracted data for 2521 ambulatory vEEG monitoring studies with linked mobile seizure diaries. They stratified these data by patients who had been diagnosed with epilepsy at the time monitoring was initiated and those who had not yet been diagnosed with epilepsy.
In this study, 66% of those with epilepsy and 30% of those who had not been definitively diagnosed with epilepsy had an abnormal report.
This revealed how frequently epilepsy events are captured ― “or what is the yield without our intervention,” Karoly said.
The validation study was an attempt to determine retrospectively whether forecasting would increase the yield. The investigators used data from 149 ambulatory vEEG monitoring studies and linked mobile seizure diaries. Participants had to have had more than 10 reported seizures in the 6 months preceding the vEEG monitoring.
For participants who were monitored during a time frame deemed high risk, the percentage of abnormal reports was 76% among those with epilepsy and 57% among those without the diagnosis.
Using the forecasting technique increased the yield for patients with epilepsy by 1.15 (or a 15% increased chance in an abnormal report). For people without a clear diagnosis of epilepsy, there was a relative change of 1.9 (or a 90% increased chance of an abnormal report).
“For everyone, we saw an increase in the captured seizures when they came in at high risk, so that’s what we expected, and that’s good,” said Karoly.
“The increase was steeper if they were unsure of the diagnosis; and that’s probably just because people who aren’t diagnosed yet might be having less frequent events,” she added.
Real-Time Forecasting
The investigators are currently carrying out a prospective study involving adults who were referred for vEEG monitoring and were set up with a mobile diary app to see how the forecasting performs “in real time,” Karoly reported.
“We tell people who need to get monitored when to come in. We actually book the monitoring using our risk forecasting,” she noted.
To date, they have recruited 38 participants, 11 of whom have come in for their scheduled EEG. Initial results are promising, Karoly said.
“It’s a little early to report the full outcome, but we can say we have seen an increase in epileptic activity across the board when these people came in during high risk,” she added.
The current risk forecasting is based on seizure diary data alone, but study participants also wear a smartwatch. That data will be used to see whether smartwatches can augment the forecast.
“We will see if we can forecast more quickly using an extra signal,” said Karoly.
The researchers are preparing the validation study results for publication and aim to complete the prospective clinical study in the first quarter of next year. They hope to launch a larger trial in which some participants will be randomly assigned either to scheduled monitoring or not.
“Once results of that clinical trial are reported, this technique could be deployed fairly easily through existing referral networks,” Karoly noted.
Useful Tool?
Commenting for Medscape Medical News, Daniel M. Goldenholz, MD, PhD, assistant professor, Harvard Beth Israel Deaconess Medical Center, said he was “very excited” to see whether the researchers “further validate” their technique with a larger sample in a prospective cohort.
“I think this could be very useful to many epilepsy centers, if it works out,” said Goldenholz, who was not involved with the research.
The validation study showed that the technique “helps a little bit for people with epilepsy” and helps “a lot for people without a clear diagnosis,” he said.
“The prospective study is too small to be sure if it addresses the problem, but the little data shown is very encouraging. With a larger sample of patients, this might become a very useful tool,” Goldenholz added.
However, he noted that the fact that the forecasting tool requires participants to have had at least 10 seizures is “an important caveat.”
The study was funded by the National Health and Medical Research Council in Australia. Karoly has a financial interest in Seer Medical (CHECK), the company that runs the diagnostic monitoring. Goldenholz is on the advisory boards for Epilepsy AI, Eysz, and Magic Leap and has received a grant from the National Institutes of Health.
American Epilepsy Society (AES) 76th Annual Meeting 2022: Abstract 1.472. Presented December 3, 2022.
For more Medscape Neurology news, join us on Facebook and Twitter.
Source: Read Full Article