Citation
Saxe, Glenn N. (2018). Computational Child Psychiatry: New Methods to Address the Complexity of Child and Adolescent Psychiatric Disorders.Abstract
ObjectivesThis Symposium introduces the emerging field of computational psychiatry as applied to child and adolescent mental health. Four presentations will describe research studies using a variety of computational methods to showcase their use for predictive classification, causal discovery, and the generation of useful clinical tools. Computational psychiatry encompasses a new approach to mental health research that has gained increasing attention in recent years. Its central rationale concerns the complexity of mental illnesses and the limitations of conventional research methods and data analytic techniques to address the complexity of these illnesses. These methods and/or data techniques include machine-learning (ML) predictive classification, causal network modeling, and network science based on graph theory.
Methods
A brief overview of computational psychiatry and its methods will be presented followed by 4 research presentations. Each of the research presentations will feature one or more computational methods applied to specific mental health problems in childhood, including ASD, PTSD, and suicide. Each of the presenters will also offer their perspective on the use of computational psychiatry for the field.
Results
The first presentation describes 2 studies using ML and causal network modeling to predict the development of PTSD in acutely traumatized children and suicide attempts in adolescents and to create causal network models of PTSD and suicide attempts, respectively. The second presentation describes a study of causal network modeling to understand the etiology of PTSD in a longitudinal cohort of young children at risk for maltreatment. The third presentation describes a study designed to create clinically relevant and accurate decision trees to support the prediction of suicide in adolescents. The final presentation describes research applying ML to accurately classify children with ASD based on information contained in a brief videotape of the child.
Conclusions
Computational psychiatry provides a set of powerful research methods that can address the complexity of mental disorders in children. These methods can enable accurate prediction, etiologic discovery, and the creation of useful clinical tools.