Summary: A new AI algorithm assesses risk factors across three domains to predict the risk of children developing conduct disorders with a high degree of accuracy.
Source: Elsevier
Conduct disorder (CD) is a common yet complex psychiatric disorder featuring aggressive and destructive behavior. Factors contributing to the development of CD span biological, psychological, and social domains.
Researchers have identified a myriad of risk factors that could help predict CD, but they are often considered in isolation.
Now, a new study uses a machine-learning approach for the first time to assess risk factors across all three domains in combination and predict later development of CD with high accuracy.
The study appears in Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, published by Elsevier.
The researchers used baseline data from over 2,300 children aged 9 to 10 enrolled in the Adolescent Brain Cognitive Development (ABCD) Study, a longitudinal study following the biopsychosocial development of children.
The researchers “trained” their machine-learning model using previously identified risk factors from across multiple biopsychosocial domains. For example, measures included brain imaging (biological), cognitive abilities (psychological), and family characteristics (social).
The model correctly predicted the development of CD two years later with over 90% accuracy.
Cameron Carter, MD, Editor of Biological Psychiatry: Cognitive Neuroscience and Neuroimaging, said of the study: “These striking results using task-based functional MRI to investigate the function of the reward system suggest that risk for later depression in children of depressed mothers may depend more on mothers’ responses to their children’s emotional behavior than on the mother’s mood per se.”
The ability to accurately predict who might develop CD would aid researchers and healthcare workers in designing interventions for at-risk youth with the potential to minimize or even prevent the harmful effects of CD on children and their families.
“Findings from our study highlight the added value of combining neural, social, and psychological factors to predict conduct disorder, a burdensome psychiatric problem in youth,” said senior author Arielle Baskin-Sommers, PhD at Yale University, New Haven, CT, USA.
“These findings offer promise for developing more precise identification and intervention approaches that consider the multiple factors that contribute to this disorder. They also highlight the utility of leveraging large, open-access datasets, such as ABCD, that collect measures about the individual across levels of analysis.”
About this AI research news
Author: Rhiannon Bugno
Source: Elsevier
Contact: Rhiannon Bugno – Elsevier
Image: The image is in the public domain
Original Research: Closed access.
“Classifying Conduct Disorder using a biopsychosocial model and machine learning method” by LenaChan et al. Biological Psychiatry Cognitive Neuroscience and Neuroimaging
Abstract
Classifying Conduct Disorder using a biopsychosocial model and machine learning method
Background
Conduct Disorder (CD) is a common syndrome with far-reaching effects. Risk factors for the development of CD span social, psychological, and biological domains. Researchers note that predictive models of CD are limited if the focus is on a single risk factor or, even, a single domain. Machine learning methods are optimized for the extraction of trends across multi-domain data but have yet to be implemented in predicting the development of CD.
Methods
Social (e.g., family, income), psychological (e.g., psychiatric, neuropsychological), and biological (e.g., resting-state graph metrics) risk factors were measured using data from the baseline visit of the Adolescent Brain Cognitive DevelopmentSM Study when youth were 9-10-years-old (n = 2,368). Applying a feed-forward neural network machine learning method, risk factors were used to predict CD diagnoses two years later.
Results
A model with factors that included social, psychological, and biological domains outperformed models representing factors within any single domain, predicting the presence of a CD diagnosis with 91.18% accuracy. Within each domain, certain factors stood out in terms of their relationship to CD (social: lower parental monitoring, more aggression in the household, lower income; psychological: greater ADHD and ODD symptoms, worse crystallized cognition and card sorting performance; biological: disruptions in the topology of subcortical and frontoparietal networks).
Conclusions
The development of an accurate, sensitive, and specific predictive model of CD has the potential to aid in prevention and intervention efforts. Key risk factors for CD appear best characterized as reflecting unpredictable, impulsive, deprived, and emotional external and internal contexts.
