AAmii's Advances in Schizophrenia Prediction

AI tool helps predict schizophrenia and severity of symptoms

A group out of the University of Alberta is making media headlines after publishing research that highlights the development of an artificial intelligence tool to predict schizophrenia by analyzing brain scans.

The study, published in Nature, was conducted by Sunil Vasu Kalmady (Senior Machine Learning Specialist, Faculty of Medicine, University of Alberta) under the co-supervision of Russ Greiner (Fellow and Canada CIFAR AI Chair at Amii, Professor of Computing Science at the University of Alberta).

It details a tool which can successfully predict early symptoms of schizophrenia in relatives of already-diagnosed individuals. Previously, the tool also achieved 87% accuracy in predicting a diagnosis of the disorder through examining brain scans.

About the tool

The machine learning tool “EMPaSchiz” (Ensemble algorithm with Multiple Parcellations for Schizophrenia prediction) was co-developed with Andrew Greenshaw and Serdar Dursun (University of Alberta Professors from the Department of Psychiatry), and a team of researchers from the National Institute of Mental Health and Neurosciences in India.

By examining patient brain scans, EMPaSchiz was previously used in a 2017 study to help identify more reliable multivariate patterns for accurate prediction of schizophrenia and its symptom severity. The tool could provide a data-driven means for clinical psychologists and psychiatrists to improve their diagnosis and symptom assessment of patient conditions.

The tool has since been developed with a machine learning algorithm to predict if an individual has schizophrenia, based on features extracted from MRI scans. According to the research team, this research “demonstrates the potential of machine-learned diagnostic models to predict state-independent vulnerability, even when symptoms do not meet the full criteria for clinical diagnosis.”

As with much of the AI research being conducted in precision medicine, this tool is designed to provide information and insights to support clinical decision-making and empower human psychiatrists with more data-driven diagnosis and assessments methods.