Multi-level predictors of depression symptoms in the Adolescent Brain Cognitive Development (ABCD) study

Area of Science:

• Neuroscience
• Developmental Psychology
• Psychiatry

Background:

• Identifying adolescent depression risk factors is crucial for early intervention.
• Previous studies often examined isolated factors, limiting comprehensive understanding.
• This study investigates multi-level factors for predicting depression symptoms in children.

Purpose of the Study:

• To examine multi-level factors that maximize the prediction of depression symptoms in US children.
• To compare linear and non-linear predictive models for depression symptom prediction.
• To identify key predictors of concurrent and future depression symptoms.

Main Methods:

• Utilized data from 7,995 participants in the Adolescent Brain and Cognitive Development (ABCD) study.
• Employed Child Behavior Checklist for depression symptom measurement.
• Used elastic net regression (EN) and gradient-boosted trees (GBT) with demographic, environmental, and fMRI data.

Main Results:

• Both EN and GBT models showed comparable prediction accuracy for baseline and 1-year follow-up depression symptoms.
• Top predictors included parental depression history, family conflict, and shorter sleep duration.
• Caudate functional connectivity was the strongest neural predictor, though overall weaker than other factors.

Conclusions:

• Parental mental health, family environment, and sleep quality are significant risk factors for youth depression.
• Caudate functional connectivity may serve as a potential biomarker for depression risk.
• A multi-faceted approach is necessary for understanding and predicting adolescent depression.