Heterogeneous developmental trajectories and predictors of self-harm in early adolescence: A machine learning study

Area of Science:

• Developmental Psychology
• Adolescent Mental Health
• Public Health

Background:

• Limited understanding of dynamic, state-specific self-harm development in early adolescence.
• Few prospective studies examining predictors of adolescent self-harm pathways.
• Need for research on heterogeneous self-harm trajectories beyond frequency and severity.

Purpose of the Study:

• To delineate distinct self-harm trajectories in early adolescence using a person-centered approach.
• To identify baseline predictors associated with these identified self-harm pathways.
• To understand the dynamic interplay between self-harm ideation and behavior over time.

Main Methods:

• Utilized a large-scale Chinese adolescent cohort (N=11,366) with four annual waves of data.
• Employed a person-centered approach to identify distinct self-harm developmental trajectories.
• Applied interpretable machine learning to pinpoint baseline predictors for each trajectory.

Main Results:

• Identified five distinct self-harm trajectories: persistently low-risk, persistent ideation, ideation remission, behavior-to-ideation, and ideation-to-behavior.
• Depressive symptoms, self-blame, family stress, and gender were significant overall predictors.
• Predictors varied considerably across trajectories, highlighting unique developmental drivers for each pathway.

Conclusions:

• Adolescent self-harm exhibits heterogeneous developmental trajectories.
• Trajectory-specific risk predictors underscore the need for tailored prevention strategies.
• Interventions should address both common and unique risk factors across different self-harm pathways.