Clinical EEG and computational neuroscience

Pediatric EEG and medication

At CHU Sainte-Justine, I collaborate with Dr. Alexander G. Weil and Dr. Aristides Hadjinicolaou on EEG questions involving epilepsy, ADHD, autism, and medication. I curate recordings with medication and comorbidity metadata and compare medicated and unmedicated ADHD groups, accounting for age, sex, and comorbidities.

Signal features and learned representations

I analyze rhythms and complexity alongside representations extracted from EEG foundation models. The workflow includes dimensionality reduction, linear probing, fine-tuning, and decoding. These approaches let me ask how different descriptions of the same recordings relate to the clinical question.

This is where my work on foundation models and clinical neuroscience directly meets: the models are methods for studying real clinical data, and those data pose questions about generalization, group imbalance, and interpretation. Associations between medication groups and EEG patterns do not by themselves establish a medication effect.

A broader clinical collaboration

I also co-authored a preprint on predicting cognitive decline in prodromal synucleinopathies. That study uses clinical markers and machine learning to examine different trajectories toward dementia with Lewy bodies and Parkinson’s disease. It is a separate population and dataset from the pediatric EEG work, connected by an interest in individual differences and careful prediction.