Machine learning-based nomogram for non-suicidal self-injury among depressed adolescents: a multicentre study

Area of Science:

• Clinical Psychology
• Psychiatry
• Adolescent Health

Background:

• Non-suicidal self-injury (NSSI) is a significant concern in adolescents with depression.
• Limited tools exist for precise, individualized NSSI risk assessment in this demographic.
• Accurate risk stratification is crucial for timely intervention and management.

Purpose of the Study:

• To develop and validate a clinically applicable nomogram for estimating NSSI probability in depressed adolescents.
• To identify key clinical and psychosocial predictors of NSSI in this population.
• To provide a tool for enhanced clinical decision-making regarding NSSI risk.

Main Methods:

• A nationwide multicenter cohort of 2,343 depressed adolescents was utilized.
• Random Forest and SHAP values, combined with logistic regression, identified predictive variables.
• Model performance was rigorously assessed using AUC, Hosmer-Lemeshow tests, calibration curves, and DCA.

Main Results:

• Eight variables were identified as significant predictors of NSSI: depression score, sleep medication use, difficulty identifying feelings, age, perceived family support, sex (female), hallucination, and externally oriented thinking.
• The developed nomogram demonstrated strong predictive performance in both training (AUC=0.754) and validation (AUC=0.748) cohorts.
• Calibration curves and decision curve analysis confirmed the nomogram's clinical utility and reliability.

Conclusions:

• A validated nomogram integrating eight key variables can effectively estimate NSSI risk in depressed adolescents.
• This tool offers a practical method for clinicians to assess current NSSI probability.
• The nomogram supports targeted psychosocial assessments and personalized clinical management strategies for at-risk youth.