Using machine learning to identify early predictors of adolescent emotion regulation development.

Area of Science:

• Developmental Psychology
• Adolescent Mental Health
• Machine Learning Applications

Background:

• Approximately 20% of adolescents experience difficulties with emotion regulation.
• Identifying early predictors is essential for timely intervention.
• Understanding developmental trajectories of emotion regulation is key.

Purpose of the Study:

• To identify early predictors of emotion regulation development in adolescents.
• To rank the importance of 87 candidate variables assessed at age 13.
• To utilize machine learning for predicting developmental trajectories.

Main Methods:

• Employed the SEM-forests machine learning algorithm.
• Assessed 87 candidate variables in 497 Dutch families at age 13.
• Predicted quadratic latent trajectory models of emotion regulation from ages 14 to 18.

Main Results:

• Individual differences (e.g., personality), relationship quality, and parent-peer conflict were significant predictors.
• Internalizing and externalizing problems strongly predicted emotion regulation development.
• Demographics, bullying, delinquency, and substance use were less predictive, with negative parenting practices ranking higher than positive ones.

Conclusions:

• Early identification of at-risk adolescents is feasible through specific predictors.
• Findings inform theoretical models of emotion regulation development.
• An open-source risk assessment tool (ERRATA) is presented for practical application.