New software predicts seizures 17 minutes before occurrence

Researchers from the University of Windsor have created a software program that can predict an oncoming seizure 17 minutes before it happens. Prof. Robin Gras and PhD student Abba Golestanti tested their software, GenericPred, on data from epilepsy patients.

It was able to predict an upcoming seizure up to 17 minutes in advance with 100 per cent accuracy. The results of their tests were published in the journal Nature last month. The two also tested their software on stock market trends and global temperature data.

The researchers used their software to analyze the brain activity of 21 patients. About 30 minutes of a patient's electroencephalography (EEG) readings was enough data to predict an upcoming seizure. Gras said that the software is a major improvement over available predictive tools, which can predict oncoming seizures about five minutes before they occur.

He told CTV's Canada AM that giving patients advanced warning about an upcoming seizure can give them more time to remove themselves from dangerous situations, like when a patient is driving.

"That's the goal of this tool, to be able to alert people before they're in a dangerous situation," Gras said, noting that with 17 minutes warning drivers can pull over, park their car and call the hospital.

The researchers have patented GenericPred, but are now hoping to team up with a firm to develop a device that will utilize the software. Gras said eventually they'd like to apply the software to other health issues or illnesses, including Parkinson's disease and strokes. "That's something we'd like to do, it seems that our tool is very generic," he said.