Prediction and impact of maladaptive perfectionism on non-suicidal self-injury in adolescents: a study based on machine learning and structural equation modeling.

Background: Non-suicidal self-injury (NSSI) is highly prevalent among adolescents, yet the psychological mechanisms, particularly the synergistic effects of maladaptive perfectionism, impulsivity, and emotion regulation difficulties, remain inadequately understood. While recent longitudinal studies have focused on predicting NSSI occurrence using multiple machine learning algorithms, the current study addresses a complementary scientific question by examining the underlying psychological mechanisms through a cross-sectional multi-method approach. Methods: In a sample of 3,865 Chinese adolescents, we employed a multi-method approach combining machine learning (Support Vector Machine for classification) with structural equation modeling to analyze self-report data. It should be noted that due to the cross-sectional nature of our data, the relationships identified represent associations rather than causal effects. Results: The machine learning model demonstrated good discriminatory power for identifying NSSI (AUC = 0.79, 95% CI [0.76, 0.82]), with emotion regulation difficulties emerging as the strongest predictor. The chain mediation model revealed that maladaptive perfectionism is associated with NSSI through a sequential pathway: it is linked to heightened impulsivity, which in turn is associated with exacerbated emotion regulation difficulties, ultimately relating to self-injury. Conclusion: Maladaptive perfectionism is associated with risk for NSSI through a cascade of psychological processes involving impulsivity and emotion regulation deficits. These findings underscore the necessity for multi-target interventions that simultaneously address perfectionistic cognitions, impulsive tendencies, and regulatory skills in at-risk adolescents.