Prediction modeling in transdiagnostic risk: results from the PROCAN study.
Authors
Affiliations (17)
Affiliations (17)
- Department of Psychiatry, Hotchkiss Brain Institute, University of Calgary, Mathison Centre, 3280 Hospital Dr NW, Calgary, AB, T2N 4Z6, Canada. [email protected].
- Department of Psychology, St.Mary's University, Calgary, AB, Canada. [email protected].
- Department of Biomedical Engineering, University of Calgary, Calgary, AB, Canada.
- Department of Psychiatry, Hotchkiss Brain Institute, University of Calgary, Mathison Centre, 3280 Hospital Dr NW, Calgary, AB, T2N 4Z6, Canada.
- Electrical and Software Engineering, University of Calgary, Calgary, AB, Canada.
- Department of Radiology, University of Calgary, Calgary, AB, Canada.
- Department of Radiology, Alberta Children's Hospital Research Institute, Calgary, AB, Canada.
- Department of Radiology, Child and Adolescent Imaging Research Program, Calgary, AB, Canada.
- Centre for Youth Bipolar Disorder, Center for Addiction and Mental Health, Toronto, ON, Canada.
- Department of Psychiatry, Faculty of Medicine, University of Toronto, Toronto, ON, Canada.
- Department of Pharmacology, Faculty of Medicine, University of Toronto, Toronto, ON, Canada.
- Department of Community Health and Epidemiology, Faculty of Medicine, Dalhousie University, Halifax, NS, Canada.
- Department of Psychiatry, University Health Network, Toronto, ON, Canada.
- Department of Psychiatry, St. Michael's Hospital, Toronto, ON, Canada.
- Arthur Sommer Rotenberg Chair in Suicide and Depression Studies, St. Michael's Hospital, Toronto, ON, Canada.
- Li Ka Shing Knowledge Institute, St. Michael's Hospital, Toronto, ON, Canada.
- Krembil Research Institute, University Health Network, Toronto, ON, Canada.
Abstract
Identifying biomarkers for serious mental illnesses (SMI) has significant implications for early intervention and prevention. The current study uses machine learning to build a model of risk prediction and transition based on multi-modal neuroimaging, clinical, and behavioral data from youth at transdiagnostic risk. Participants aged 12-25 were recruited at two sites in Canada, and followed for 4 years. Symptom severity was measured using the Scale of Psychosis-Risk Symptoms (SOPS) and K10 Distress Scale, and a range of cognitive and behavioral measures were collected, as well as magnetic resonance imaging (MRI) data. Participants were assigned to one of 5 groups: healthy controls (HC; n = 42), familial risk (stage 0; n = 40), mild symptoms (stage 1a; n = 48), attenuated syndromes (stage 1b; n = 82), or discrete disorder (transition; n = 31). Constrained spherical deconvolution was used to generate whole brain tractography maps from diffusion MRI, which were then used to calculate connectivity matrices for graph theory analysis. Graph theory was also used to analyze correlations of functional MRI signal between pairs of brain regions. All measures were evaluated in a model to predict transition between groups. Random Forest analysis identified diffusion MRI-derived nodal metrics of betweenness centrality in the angular gyrus, inferior temporal gyrus, amygdala and calcarine fissure as potential features which can discriminate between the groups. Additionally, SOPS and K10 Distress Scales were useful behavioral predictors of transdiagnostic risk. Our findings show that combining neuroimaging with clinical characteristics may result in a promising predictive model for transdiagnostic risk and transition to SMI.