Resting-state spontaneous brain activity as a neural marker for suicidal ideation in adolescents with non-suicidal self-injury: a voxel-wise and machine learning study.

Background: Non-Suicidal Self-Injury (NSSI) is a primary risk factor for suicide, but objective biomarkers to assess this risk are urgently needed. The "prefrontal-limbic dysregulation" model provides a neurobiological framework for self-injurious behaviors. This study aimed to identify resting-state neural markers of suicidal ideation severity in adolescents with NSSI and to build a predictive model for individualized risk assessment. Methods: We recruited 64 adolescent psychiatric inpatients with NSSI. Suicidal ideation was measured using the Beck Scale for Suicide Ideation (BSI). Resting-state functional MRI (rs-fMRI) was used to measure spontaneous brain activity via the amplitude of low-frequency fluctuation (ALFF). We performed a whole-brain correlation analysis between ALFF and BSI scores. A support vector regression (SVR) model was then developed using the identified neural feature to predict individual BSI scores. Results: A significant negative correlation was found between BSI scores and ALFF values in the left Middle Frontal Gyrus (MFG). Lower spontaneous activity in this region was associated with more severe suicidal ideation. The SVR model, based on the left MFG ALFF values, successfully predicted individual BSI scores with significant accuracy (r = 0.492, p < 0.001), a finding confirmed by permutation testing. Conclusion: Diminished resting-state activity in the left MFG is a key neural correlate of suicidal ideation severity in adolescents with NSSI. The functional activity of the left MFG is a promising biomarker for suicide risk assessment and may serve as a potential target for novel neuromodulatory therapies in this high-risk population.