This presentation will provide an overview of our group’s work using electrophysiological measures (i.e. electroencephalography (EEG) and event-related potentials (ERP)) of brain activity to study individuals who have sustained a brain injury. In order to examine the consequences of brain injury, we have taken advantage of the component structure of ERPs in which particular time-based features of brain responses to stimulus environments reflect functions such as attention, memory, and language comprehension. Work will be described demonstrating the value of this approach in assessing functional capacity in a variety of patients including those who have been diagnosed as being in a vegetative state, in a coma, or having sustained a concussion. The more traditional approaches to signal processing will be discussed followed by a presentation of the significant “value-added” in using machine learning (ML) methods to provide more fine-grained assessment of the neurophysiological signals obtained in these different patient populations. The talk will discuss a number of different approaches that were taken to apply ML to these types of biological signals, challenges that particularly affect these applications (e.g., limited sample sizes, skewed classes, and varying SNRs), and strategies our group has taken to bridge that gap in the domain of clinical EEG/ERP.

Professor John F. Connolly, Ph.D. uses brain imaging tools to study cognitive functioning in acquired brain injuries (concussion, coma, traumatic brain injury, stroke) and autism; the common theme being the creation of tools enabling objective, quantitative and replicable assessment of brain-injured patients including those who are impossible to assess with traditional measures due to their clinical state. He holds a patent for EEG technology capable of assessing cognitive consequences of brain injury and disorders of consciousness and is the CSO for VoxNeuro (VoxNeuro.com).

He is the Senator William McMaster Chair, Cognitive Neuroscience and the founding Director of the ARiEAL Research Centre where he co-directs the Language Memory & Brain Laboratory (LMBLab). He holds appointments in the Faculties of Engineering, Science, and Humanities and is appointed in the Neuroscience Graduate Program in the Faculties of Health Sciences and Science at McMaster University. He has published extensively in basic systems neuroscience and in application areas including neurology and neuropsychiatry.

His research has received support from agencies including: CIHR, NSERC, SSHRC, NIH, IRAP, Ontario Neurotrauma Foundation, Ontario Brain Institute, Ontario Centres of Excellence, March of Dimes (USA), Scottish Rite (USA & Canada). He has also consulted for public and private organizations in Australia, Belgium, Canada, Finland, Hong Kong, New Zealand, the United Kingdom and the USA.

Rober Boshra, Ph.D. is a researcher of electrophysiology and clinical applications of machine learning to improve assessment capability and outcome. His research focuses on using explainable machine learning to answer neuroscience questions. His recent work targets objective brain injury assessment using EEG/ERP, as well as investigates automated detection of consciousness markers in coma and other disorders of consciousness. He has multidisciplinary training as a computer scientist, neuroscientist, and biomedical engineer. Rober is an alumnus of McMaster's AI-focused MacData institute and a postgraduate affiliate of the Vector Institute. Rober is currently the Director of AI & technology at VoxNeuro. He has conducted work that received financial support from a number of sources including: the Ontario Ministry of Education, The Hamilton Spectator, CIHR, NSERC, MacData, and NRC IRAP.