Network analysis of psychological heterogeneity and stratified suicide risk management in depressed adults under the RDoC framework: a cross-sectional study

Based on the Research Domain Criteria (RDoC), this study systematically explored the heterogeneous structure of psychological function in adult clinical samples, clarified the interaction mechanism between suicide risk and four core dimensions (Cognition, Emotion, Behavior, Physiology/Sleep; CEBP for short), and aimed to provide empirical support for accurate clinical psychological risk assessment and stratified intervention. A cross-sectional study enrolled 2167 Adult patients with depression. The CEBP dimensions were constructed using SCL-90, HAMD, HAMA, and PSQI.

Suicide risk was assessed by combining the HAMD suicide item and clinical interviews. Latent Profile Analysis (LPA) identified potential psychological function subtypes; psychological network analysis (EBICglasso method) constructed networks for the full sample and each subtype, and network comparison tests revealed inter-subtype structural differences. All statistical analyses were performed using R software (version 4.5.2).

LPA showed the 2167 samples could be divided into 4 distinct potential subtypes (4-class model optimal, Entropy = 0.91): Class 1 (n = 414, 19.10%), Class 2 (n = 180, 8.31%), Class 3 (n = 577, 26.63%), Class 4 (n = 996, 45.96%), each with a low-to-high symptom gradient and distinct configurations. Emotion was the full-sample network core, with strong positive connections to other dimensions, and most significantly correlated with suicide risk. Subtypes had specific structures: Mild multi-domain impairment (Class 1), Physiology/sleep core dominant subtype (Class 2), Emotion-cognition coupled moderate impairment (Class 3), Low overall symptom burden (Class 4).

Network comparison tests confirmed significant overall structure and global strength differences between Class 1 & 4 and Class 2 & 3 (adjusted p