Skip to Main Content

Study Strives to Forecast Seizures

graphic of human brain overlaid on top of world mapFor people living with epilepsy, not knowing when they will have their next seizure significantly limits their ability to work and participate in activities. Yet research now underway could change all that by providing a warning when a seizure is imminent.

Dr. Mona Nasseri, assistant professor of electrical engineering, is collaborating with Mayo Clinic on a study that found patterns when researchers compared physiological data – collected by a monitoring device worn on the wrist -- with the actual time of a seizure. Through analysis of data, such as heart rate, body temperature and movement, researchers discovered that they would have been able to forecast most of the seizures about 30 minutes before they occurred. As a result, they recently published their findings showing that it is possible to provide reliable seizure forecasts without directly measuring brain activity.

That gift of time would offer hope for a better life, allowing patients to take fast-acting medications or alter their activities. “We just hope to help these people,” Nasseri said. “I have seen these patients, and I know that they need something like this … when they have a lot of seizures that are resistant to medications, they have to avoid so many activities. We hope to be able to help them with this project.”

The study is part of the Epilepsy Foundation of America’s Epilepsy Innovation Institute, and the My Seizure Gauge project, which includes international collaboration. The project is based at Mayo Clinic in Rochester, Minnesota. Nasseri worked for three years at that location with Dr. Benjamin Brinkmann, an epilepsy scientist and lead researcher for the study, before joining the faculty at UNF in the fall of 2020. Involved from the project’s inception, Nasseri will continue to analyze data and collaborate with Brinkmann.

This is the first study that followed people during their daily activities for six to 12 months, rather than previous work that was based on in-hospital data recording of patients, according to Nasseri. They tracked six people with drug-resistant epilepsy and an implanted neurostimulation device that monitors electrical brain activity. Because of the device in the brain, the researchers were able to receive data that indicated exactly when the seizure occurred, rather than having to rely on participants noting the time in personal diaries, which is less reliable.

Nasseri is contributing to the study by implementing machine learning and signal processing techniques to develop these detection algorithms and seizure predictions. “We collected the data from the wrist-worn devices and designed a machine-learning algorithm,” she said. “Based on the actual times of the seizures, we selected data prior to the seizure to train the machine learning classifier and were able to develop the algorithm to recognize the pre-seizure data.”

Though the data collection is nearly complete, the analysis will continue and may take close to a year to accomplish. The next steps will include perfecting the algorithm and developing the hardware that can apply the algorithm in a real-time application, something a few years in the future.

Nasseri is pleased to be able to continue working with Mayo on this research. “Before I worked at Mayo, I was working on telecommunication systems, but then when I started working at Mayo, I found this project very interesting. It gives meaning to your work. You see that you probably are going to help some people. That’s different and what I truly enjoy.”