Yilong Zhang

Anxiety and depressive symptoms frequently co-occur during adolescence. Physical activity is associated with adolescent mental health at the scale-score level, but its links with individual symptoms and their temporal ordering remain less clear. This study examined contemporaneous and one-year prospective associations between physical activity and individual anxiety/depression symptoms, evaluated the robustness of pooling three annual wave-pair cohorts, and compared network patterns between male and female adolescents. Three adjacent annual wave-pair cohorts (2020–2021, 2021–2022, and 2022–2023) from a Chinese national adolescent health survey were analyzed ( N = 13,785 wave-pair records). Physical activity was measured with the PAQ-A and anxiety/depression symptoms with the SCL-90. Gaussian graphical models and cross-lagged panel networks (CLPNs) were estimated. Annual-cohort invariance, missingness, centrality, bridge centrality, accuracy, stability, and male/female differences were examined. Annual-cohort comparisons showed no detectable network-structure differences (M = 0.07–0.08, p = 0.120–0.270) or global-strength differences (ΔS = 0.12–0.32, p = 0.150–0.320). Little’s test was nonsignificant (χ² = 79.39, df = 89, p = 0.531), providing no evidence against a completely-at-random missingness mechanism. In the pooled contemporaneous network, physical activity had negative partial correlations with low motivation (weight = -0.0732), low energy (-0.0651), and nervousness (-0.0532). Worthlessness (expected influence [EI] = 1.39), tension (EI = 1.38), and nervousness (EI = 1.34) were influential nodes. Panic had the highest reported bridge EI (0.659). In the CLPN, physical activity was prospectively associated with lower low motivation ( β = -0.0602) and low mood ( β = -0.0561), while nervousness was associated with lower subsequent physical activity ( β = -0.0539). Male and female networks differed in structure (M = 0.112, p < 0.001) but not global strength (ΔS = 0.19, p = 0.709). Physical activity occupied a peripheral position in the contemporaneous network but showed small prospective associations with selected anxiety and depressive symptoms. These temporal associations do not establish causality. The stable annual-cohort pattern and male/female-specific edge configuration support further confirmatory work using designs that can address unmeasured confounding and within-person change.