Predicting antidepressant response via local-global graph neural network and neuroimaging biomarkers

Depressed mood and anhedonia, the core symptoms of major depressive disorder (MDD), are linked to dysfunction in the brain's reward and emotion regulation circuits. To develop a predictive model for treatment remission in MDD based on pre-treatment neurocircuitry and clinical features. A total of 279 untreated MDD patients were analyzed, treated with selective serotonin reuptake inhibitors for 8-12 weeks, and assigned to training, internal validation, and external validation datasets. A hierarchical local-global imaging and clinical feature fusion graph neural network model was constructed. The model achieved 76.21% accuracy (AUC = 0.78) in predicting remission. Validation on the internal and external independent datasets yielded similar performance (accuracy = 72.73%, AUC = 0.74; accuracy = 71.43%, AUC = 0.72). Key contributing brain regions included the right globus pallidus, bilateral putamen, left hippocampus, bilateral thalamus, and bilateral anterior cingulate gyrus. These findings highlight the role of specific circuits in guiding antidepressant treatment.