The co-occurrence of depressive symptoms and problematic smartphone use (PSU) in adolescents poses a growing public health concern, but its component-level conditional associations, cross-wave patterns, and variation across depression-severity groups remain insufficiently understood. Using a two-wave observational design with a 12-month interval (N = 1200 students from one high school in a county of Gansu Province, China), this study integrated Gaussian graphical modeling (GGM), network comparison testing (NCT), and latent profile analysis (LPA). Cognitive Flexibility showed the largest absolute bridge expected influence at both waves, and the ranking was unchanged when the comparison was restricted to nodes measured at the same level of aggregation. NCT detected no significant global difference between the T1 and T2 contemporaneous networks, although several local cross-community edges differed between the two estimates. LPA retained four nearly parallel, severity-ordered classes at both waves, interpreted as depression-severity groups rather than qualitatively distinct profiles. Subgroup bootstrap analyses showed acceptable edge-weight accuracy; expected-influence CS coefficients ranged from 0.286 to 0.750, whereas all bridge-expected-influence CS coefficients were below 0.25, so the subgroup findings are confined to cross-community connectivity. Between-group differences in cross-community connectivity persisted after equal-n down-sampling. These findings should be interpreted in light of the observational design, single-school convenience sample, and reliance on self-report measures. Cognitive flexibility nevertheless emerged as a consistent cross-community correlate in the overall networks and warrants further longitudinal and intervention evaluation rather than being regarded as an established intervention target.

Problematic gambling, problematic gaming, problematic social media use, and problematic pornography use differ in nosological status but may share some psychological correlates. The present cross-sectional study examined associations between early maladaptive schemas (EMSs) and these four problematic excessive behaviors, and compared theory-informed serial and parallel indirect-effect specifications involving desire thinking and craving. A community sample of 811 Iranian adults (43.4 % males; mean age = 27.70 years, SD = 9.00) completed self-report measures. Desire thinking and craving were assessed with reference to the same personally selected activity, although the selected activity was not recorded. Item-level confirmatory factor analyses supported a relative distinction between desire thinking, craving frequency, and craving strength, although substantial overlap remained. The parallel specification showed poor fit, whereas the serial specification showed acceptable fit (χ[2](38) = 227.97, CFI = 0.967, TLI = 0.943, RMSEA = 0.079, and SRMR = 0.028). In the serial specification, EMSs were positively associated with desire thinking, desire thinking was strongly associated with craving, and serial indirect associations were observed for all four problematic excessive behaviors. EMSs also remained directly associated with the outcomes. Craving was positively associated with all four behavioral severity outcomes. However, because the selected activity was not recorded, these associations cannot be interpreted as behavior-matched craving-severity associations. The findings suggest possible cross-domain relevance of EMSs, desire thinking, and craving, but do not establish temporal or causal ordering, diagnostic equivalence, or a single underlying addiction process. Because the referent of the desire-thinking and craving measures was not recorded, behavior-specific interpretation of these associations remains limited.