Grant: Project Grant

Epilepsy is a neurological condition that affects 1% of the world's population. It is characterized by recurrent seizures. One of the main tests used for the diagnosis of epilepsy is the electroencephalogram (EEG). The EEG is a recording of the brain's electrical activity, typically for 30-minutes, using non-invasive electrodes placed on the scalp. After recording, the EEG is interpreted by a neurologist in search of interictal spikes. These "spiky", asymptomatic discharges occur between seizures and are a biomarker of epilepsy. Unfortunately, they occur infrequently, and a routine EEG will manage to capture spikes in less than half of patients with epilepsy. In the absence of spikes, neurologists cannot distinguish the EEG of patients with from those without epilepsy. Researchers have turned to computers to identify subtle patterns, hidden to the naked eye, in the EEG of patients with epilepsy. Preliminary studies, tested on a small number of patients, showed that quantitative characteristics extracted from the EEG can help identify patients with epilepsy. In addition, deep learning (DL), a type of artificial intelligence, has shown good performances in identifying subtle patterns in complex data. This project aims to develop an algorithm capable of predicting long-term seizure risk based on brain signals, even in the absence of spikes. Our specific aims are to: 1) Develop a machine learning (ML) algorithm using quantitative characteristics extracted from the EEG; 2) Develop a DL algorithm capable of predicting long-term seizure risk based on raw EEG signals; 3) ML and DL algorithms will be compared and the best performing model will be tested prospectively at the Centre hospitalier de l'Université de Montréal (CHUM). This project could have a significant impact on clinical practice. The identification of new biomarkers of epilepsy will increase the capacity of the EEG to detect this condition and allow to identify and treat patients earlier.