Simon Fraser University Brain Behavior Lab

Epilepsy in children poses significant diagnostic challenges, particularly in distinguishing between benign and non-benign forms. The subtleties of Electroencephalograms (EEGs), used in diagnosis, often confound even experienced EEG readers, making correct diagnosis a difficult but critical task.

The Project

Praxis partnered with the Brain Behavior Lab at Simon Fraser University to employ an early machine learning Artificial Intelligence (AI) for epilepsy diagnosis. By creating a multi-level neural network trained on verified epilepsy data, Praxis’s system was able to categorize unexamined EEGs into benign and non-benign groups.

Mature patient waiting for electrodes to be fixed
Computer recording brain waves of mature gentleman
SFU Graph

The Result

The AI-driven solution delivered an impressive accuracy rate of over 80% in distinguishing between benign and non-benign epilepsy. With larger training data, accuracy could potentially reach the mid to high 90% range. This research was partly funded by the National Research Council of Canada.