Poster No:
571
Submission Type:
Abstract Submission
Authors:
Jason Smucny1, Avery Wood2, Ian Davidson2, Cameron Carter3
Institutions:
1University of California, Davis, Sacramento, CA, 2University of California Davis, Sacramento, CA, 3University of California Irvine, Irvine, CA
First Author:
Co-Author(s):
Avery Wood
University of California Davis
Sacramento, CA
Introduction:
Previous work has identified a set of risk factors that significantly predict conversion to psychosis in adolescents at clinical high risk (CHR) for the disorder (Cannon et al. 2016). It is unknown, however, if these same factors also predict transition to a high risk state in preteens aged 12-13.
Methods:
A logistic regression (LR) model was used to fit a binary CHR outcome, defined as having a highest distress score ≥2 on any Prodromal Questionnaire-Brief Child (PQ-BC) version (Karcher et al. 2018) question at Adolescent Brain and Cognitive Development (ABCD) study (https://abcd.org) year 4 (with concurrent distress scores <2 at year 3). Features included age, having a first-degree relative with psychosis, Rey Auditory Verbal Learning Test (RAVLT) n correct trials, NIH Toolbox Pattern Recognition test raw score, n negative life events, n trauma types, and showing a significant drop in grades from the previous year. Site-adjusted overall fractional anisotropy (FA), diffusivity, and cortical thickness were included as MRI predictors in a separate model.
Results:
5237 children were included in analyses. The LR model was significant (p<.001, R2=.042) with 65% accuracy (67% of non-"transitioners" and 52% of transitioners). Including MRI features improved LR model fit (R2 = .051) but not accuracy (65%) with FA and cortical thickness being significant predictors. Machine learning (xgboost with random forest) using all features improved accuracy to 82% (85% of non-transitioners and 52% of transitioners).
Conclusions:
These findings suggest that factors previously shown to predict conversion to psychosis in CHRs may also predict transition to a pseudo-CHR state in preteens. Model prediction may be enhanced by incorporating MRI-based features and using machine learning.
Disorders of the Nervous System:
Psychiatric (eg. Depression, Anxiety, Schizophrenia) 1
Lifespan Development:
Early life, Adolescence, Aging
Modeling and Analysis Methods:
Classification and Predictive Modeling 2
Novel Imaging Acquisition Methods:
Anatomical MRI
Diffusion MRI
Keywords:
Machine Learning
MRI
Psychiatric
Psychiatric Disorders
Schizophrenia
STRUCTURAL MRI
White Matter
WHITE MATTER IMAGING - DTI, HARDI, DSI, ETC
Other - Clinical High Risk for Psychosis
1|2Indicates the priority used for review
Provide references using author date format
Cannon T.D. (2016). An Individualized Risk Calculator for Research in Prodromal Psychosis. American Journal of Psychiatry vol. 173, no 10, pp. 980-988.
Karcher N.R. (2018). Assessment of the Prodromal Questionnaire–Brief Child Version for Measurement of Self-reported Psychoticlike Experiences in Childhood. JAMA Psychiatry vol. 75, no. 8, pp. 853-861.