Grant: Project Grant

Epilepsy, one of the most prevalent and stigmatized neurological conditions, affects 1% of Canadians of all ages and is characterized by recurrent seizures as a result of excessive neuronal brain activity. Unfortunately, more than a third of people living with epilepsy are not adequately controlled with currently available medical or surgical treatment options. The most disabling aspect of living with epilepsy is the unpredictable nature of seizures, which creates a constant source of worriment and danger for patients and their caretakers. This has motivated researchers to investigate the possibility of forecasting seizures. Similar to meteorologists who use equations of fluid dynamics and thermodynamics with measures provided by weather stations to forecast the weather, we propose to predict seizures using a) equations measuring particular nonlinear correlations across the different frequencies of the brain's electrical activity provided by implanted EEG electrodes; and b) new approaches based on artificial neural networks, a form of artificial intelligence capable of learning to perform a task, in our case automatically detecting the pre-seizure state, by seeing many training examples. Our algorithm would then be implemented into a personalized framework which would continuously forecasting the probability of being in a pre-seizure state. Real-time seizure forecasting would offer a life-changing solution to patients with drug-resistant epilepsy. Simple devices such as automated triggered alarms (to warn patients to lie down, take a rapid-acting antiseizure drug, or to alert caretakers) can realistically be implemented on a short-term basis with high impact (improved quality of life and reduction of seizure-related injuries and deaths).