Guillaume Dumas is an associate professor in the psychiatry department of the University of Montreal, a researcher at the Azrieli research center at Sainte-Justine University Hospital (Montreal) and at Mila, the Quebec artificial intelligence (AI) institute directed by Yoshua Bengio. He directs the laboratory of “precision psychiatry and social physiology”in which AI plays an important role.
How does your research constitute an innovation in the field of psychiatry?
Since my thesis work, I have been interested in intercerebral synchronizations. It is now known that the brain reacts differently when engaged in social interaction; However, this discovery had consequences on the way we understand mental health. Instead of reducing mental disorders to what happens inside the skull, we now know that we must take into account the entire body as well as the social environment in which the patient evolves. Take the case of autism: autism has long been reduced to either a problem affecting a brain region or a mutated gene. We now believe that to understand autism we must have a multi-scale approach, ranging from biological data to the social environment. Our laboratory works around this complementarity between “precision psychiatry” (or “personalized psychiatry”) which relies on current technological tools, such as AI and its big data, which make it possible to refine the diagnosis and treatment. charge, and on the other side “social physiology”, which takes into account cultural variations and the social determinants of mental health.
How does AI help you in this precision medicine?
Most people think of AI as something radically new; For my part, I see a continuity with what already existed in medicine, when we worked with statistics. But classic statistics were intended to test hypotheses, when artificial intelligence allows us to produce “predictions” from data, that is to say to indicate possible solutions that we do not know. The advent of “computational” medicine, namely the use of mathematics and computer science of which AI is a part, makes it possible to detect patterns, what we call pattern recognition : we can recognize patterns, genetic for example, and thus detect a tumor, sequence it to better treat it. Oncology was a precursor in the use of AI, but this obviously remains a challenge in psychiatry, where we touch on thought, consciousness, “ineffability”, without yet knowing scientifically how it brain and cognition work exactly. Moreover, to try to better understand the mechanisms in place in psychotherapy, we are currently recording clinical interviews between psychiatrists and patients in order to collect data on the brain activity, physiological activity and language deployed.
You have 57.62% of this article left to read. The rest is reserved for subscribers.