AbstractDespite numerous efforts to predict suicide risk in children, the ability to reliably identify youth that will engage in suicide thoughts or behaviors (STB) has remained remarkably unsuccessful. To further knowledge in this area, we apply a novel machine learning approach and examine whether children with STB could be differentiated from children without STB based on a combination of sociodemographic, physical health, social environmental, clinical psychiatric, cognitive, biological and genetic characteristics. The study sample included 5,885 unrelated children (50% female, 67% white) between 9 and 11 years old from the Adolescent Brain Cognitive Development (ABCD) study. Both parents and youth reported on children’s STB and based on these reports, we divided children into three subgroups: 1. children with current or past STB, 2. children with psychiatric disorder but no STB (clinical controls) and 3. healthy control children. We performed binomial penalized logistic regression analysis to distinguish between groups. The analyses were performed separately for child-reported STB and parent-reported STB. Results showed that we were able to distinguish the STB group from healthy controls and clinical controls (area under the receiver operating characteristics curve (AUROC) range: 0.79-0.81 and 0.70-0.78 respectively). However, we could not distinguish children with suicidal ideation from those who attempted suicide (AUROC range 0.49-0.59). Factors that differentiated the STB group from the clinical control group included family conflict, prodromal psychosis symptoms, impulsivity, depression severity and a history of mental health treatment. Future research is needed to determine if these variables prospectively predict subsequent suicidal behavior.
Predictors of suicidal thoughts and behavior in children: results from penalized logistic regression analyses in the ABCD study
