Mapping the dimensional architecture of adolescent problematic smartphone use: a network and bayesian analysis of...

Adolescent problematic smartphone use (PSU) commonly coexists with internalizing/externalizing difficulties, psychotic-like experiences and sleep disturbances, yet their dimension-level interconnections and dependency structure remain underexplored. This study constructed a multidomain, dimension-level network to detect central and bridge dimensions, adopted Bayesian models to infer probable variable ordering, and further compared network disparities by gender and educational stage.

A cross-sectional survey included 12,320 Chinese senior/junior high school students and university freshmen for analysis. Validated questionnaires yielded 20 subscale scores for analysis. Regularized undirected network analysis estimated and screened core and bridge nodes; Bayesian directed acyclic graphs revealed network probabilistic dependency sequences; network invariance tests compared network architectures across genders and educational groups.

Withdrawal symptoms possessed the highest strength and expected influence as the most interconnected dimension. Emotional problems exhibited the maximum bridge metrics and served as the primary cross-domain connector. Though thought content abnormality had relatively high overall connectivity, its links were mostly concentrated inside the psychopathology community. The Bayesian directed acyclic graphs positioned emotional problems as the leading source-like node, while withdrawal symptoms appeared early in PSU-related paths; conduct problems, negative consequences and hypnotic medication use were located mainly at receiving ends. Network structure and global strength showed no gender gaps. While global connectivity stayed stable across educational stages, network architectures varied remarkably, with sleep-related dimensions presenting significant stage-specific differences.

Adolescent PSU is integrated into an extensive multidomain system consisting of psychopathological and sleep variables. Withdrawal symptoms stand out via widespread overall connectivity, and emotional problems dominate cross-domain linkage, representing two distinct structural roles. The probabilistic variable sequences derived from Bayesian directed acyclic graphs only provide testable hypotheses for longitudinal research rather than definitive causal evidence. This study highlights the necessity of developmentally tailored assessments and pinpoints core dimensions worthy of prospective tracking as promising targets for adolescent PSU prevention and intervention.