Construction and verification of nomogram prediction model for non-suicidal self-injury in adolescents with depression

TL;DR

This study created a predictive model to identify adolescents with depression at high risk of non-suicidal self-injury, using factors like sleep and social support.

Contribution

A novel nomogram-based prediction model for non-suicidal self-injury in depressed adolescents was developed and validated.

Findings

The model achieved an AUC of 0.880 with high sensitivity and acceptable specificity.

Internal and external validations confirmed strong discrimination and calibration.

The model showed good clinical applicability through decision curve analysis.

Abstract

Accurate identification of high-risk individuals for NSSI and timely intervention are critical for mitigating self-harm risk. This study aimed to develop a predictive model for NSSI behaviors in adolescents with depression.

A convenience sample of 596 adolescents with depression was assessed, with 455 assigned to the training and internal validation set and 144 to the external validation set. Nine key predictors were identified through univariate analysis, LASSO regression, and binary logistic regression, including birth mode, history of peer self-harm, parental psychiatric disorders, sleep duration, social life events, self-esteem, psychological resilience, social support, and depression severity. A nomogram-based prediction model was constructed from these factors, with model performance evaluated via ROC curves, AUC values, Hosmer-Lemeshow test, and calibration curves. Clinical applicability was determined using decision curve analysis (DCA).

The model exhibited an AUC of 0.880 (P < 0.001), with sensitivity of 0.933 and specificity of 0.765. The Hosmer-Lemeshow test confirmed good model fit (χ2 = 7.19, P = 0.516). Both internal and external validations demonstrated strong discrimination, calibration, and clinical relevance.

The nomogram-based risk model developed in this study effectively predicts NSSI behaviors in adolescents with depression, offering significant scientific and clinical value and warranting further implementation.

The online version contains supplementary material available at 10.1186/s40359-025-02789-8.

Figures