Machine learning based identification of structural brain alterations underlying suicide risk in adolescents

Area of Science:

• Neuroimaging
• Psychiatry
• Machine Learning

Background:

• Suicide is a leading cause of death in adolescents, necessitating improved identification of at-risk individuals.
• Current neuroimaging biomarkers for suicide risk lack sufficient accuracy.
• Identifying reliable neural signatures is crucial for early intervention and prevention strategies.

Purpose of the Study:

• To apply machine learning algorithms to structural MRI data for discriminating individuals with suicide risk from typically developing adolescents.
• To identify region-specific brain structural alterations associated with adolescent suicide risk.
• To develop an accurate, individual-level assessment of suicide risk using neuroimaging data.

Main Methods:

• Structural MRI data from 79 adolescents with suicide risk and 79 typically developing controls were analyzed.
• Region-specific cortical and subcortical volumes (CV/SCV) were extracted after whole-brain parcellation.
• Support vector machine (SVM), K-nearest neighbors, and ensemble algorithms were employed for classification.

Main Results:

• The SVM classifier achieved the highest accuracy (74.79%) in distinguishing suicide risk from typically developing adolescents.
• Key discriminative regions included reduced cortical volume in frontal/temporal lobes and increased volume in the cuneus/precuneus.
• High sensitivity (75.90%) and specificity (74.07%) were observed, with an AUC of 87.18%.

Conclusions:

• Machine learning effectively identifies structural brain biomarkers for adolescent suicide risk.
• The study presents a robust, unbiased framework for assessing suicide risk using neuroimaging.
• Further validation with larger cohorts and clinical controls is recommended to confirm findings.