Heterogeneity in non-suicidal self-injury ideation and treatment response among inpatient female adolescents with mood disorders: Insights from ecological momentary assessment

Non-suicidal self-injury (NSSI) is highly prevalent and exhibits substantial fluctuations among adolescents with mood disorders, yet the heterogeneity and differential treatment responses accompanied by such fluctuations remain unclear. The current study employed ecological momentary assessment (EMA) and latent profile analysis to identify latent subtypes of NSSI ideation via temporal features, and constructed machine-learning models to validate the predictive value of dynamic features for post-treatment self-injury risk. Thirty-five Chinese female adolescents with mood disorders and co-occurring NSSI ( M _age = 14.4 years) receiving inpatient psychiatric treatment completed baseline assessments and 5 daily surveys of NSSI ideation throughout the course of treatment. Two subtypes were identified: the high-mean paroxysmal subtype (54.3%) and the low-mean sporadic subtype (45.7%). The high-mean paroxysmal subtype exhibited significantly more severe clinical symptoms, more emotional abuse experiences, lower self-esteem, higher impulsivity and greater difficulties in emotion regulation. The temporal-feature-based model outperformed the baseline-feature-based model in predicting self-injury risk. These findings underscore the dynamic nature and heterogeneity of NSSI ideation among adolescents with mood disorders, and offering implications for tailoring interventions to distinct youth subgroups.