Dynamic Evolution of Anxiety and Depression Symptoms in Adolescents: A Cross-Lagged Panel Network Analysis

Objective Adolescent anxiety and depression are often co-infected and affect each other. The traditional research paradigm based on total scores make it difficult to reveal the dynamic interaction structure among symptoms. This study adopts the network analysis method to explore the cross-sectional characteristics and longitudinal prediction relationship of the network of adolescent anxiety and depression symptoms and identify central symptoms and key driving nodes, providing an empirical basis for accurate psychological intervention. Methods 831 adolescents aged 10–18 who were diagnosed with anxiety and depression in the outpatient clinic were included and completed the Symptom Checklist-90 (SCL-90) during the baseline period (T1) and the follow-up period (T2, with an interval of 3–12 months).EBICglasso regularized partial correlation networks were estimated to construct cross-sectional networks at T1 and T2, with core symptoms identified using node strength indices. Longitudinal directed associations were modeled using a Cross-Lagged Panel Network (CLPN) analysis with LASSO regularization, from which In-Expected Influence and Out-Expected Influence centrality metrics were derived to assess temporal influence. Network stability was evaluated using 3,000 non-parametric bootstrap resamples. Results Cross-sectional network analysis revealed that at T1, depression (DEP) emerged as the core symptom (strength = 1.20), with the strongest edge identified between interpersonal sensitivity (INT) and paranoid ideation (PAR) (r = 0.38). At T2, the core symptom shifted to anxiety (ANX) (strength = 1.30), and the strongest connection was observed between Somatization (SOM) and Anxiety (ANX) (r = 0.32). Global network strength exhibited a slight decline from T1 (4.97) to T2 (4.77). In the longitudinal CLPN analysis, SOM demonstrated the highest positive Out-Expected Influence (0.269), acting as the primary driver of subsequent symptom exacerbation, whereas phobic anxiety (PHOB) uniquely exhibited a negative Out-Expected Influence (-0.787). In-Expected Influence values across symptoms ranged from − 0.145 to 0.025. Conclusion The symptom network structure of anxiety and depression in adolescents underwent structural shifts as the course of illness progressed, with the core symptom transitioning from depression to anxiety. Somatization (SOM) emerged as a key driver of future psychopathological deterioration, whereas phobic anxiety (PHOB) exhibited a unique negative predictive effect in the temporal sequence. Targeted intervention against core nodes may effectively disrupt the vicious cycle of symptom reinforcement.