Enhancing EEG Data Analysis and Brain Health monitoring through Self-Supervised Learning and Corticothalamic Modeling

Sleep is a central function of many species, and is studied widely in neurophysiology and computational neuroscience, yet its underlying mechanisms remain elusive. The common method for recording brain activity during sleep is polysomnography (PSG) which is expensive to operate and can prove to be heavy and bothersome for the recording subjects. In recent years, low-cost and lightweight mobile EEG (mEEG) headsets have revolutionized sleep assessments by democratizing the access to sleep EEG. Another considerable development in the field has been the development of mathematical models of brain physiology that can, crucially, simulate EEG activity. In this project, we will combine the use of a semi-supervised learning method and a neurophysiological model of the corticothalamic circuitry to extract explainable representations of the sleep data and identify significant biomarkers of healthy and disordered states of the brain.

Faculty Supervisor:

John Griffiths

Student:

Partner:

InteraXon Inc

Discipline:

Engineering

Sector:

Professional, scientific and technical services; Retail trade; Wholesale trade

University:

University of Toronto

Program: