Selective Network Reweighting in Adolescents with Mood Disorders and Problematic Smartphone Use

Selective Network Reweighting in Adolescents with Mood Disorders and Problematic Smartphone Use | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Article Selective Network Reweighting in Adolescents with Mood Disorders and Problematic Smartphone Use Xizhe Zhang, Jiayang Zhu, Jing Zhang This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-8256165/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background and purpose Problematic smartphone use (PSU) is highly prevalent in mood-disordered adolescents, but it is unclear whether it reflects global disruption of functional architecture or specific circuits, and how it relates to depressive symptom. Methods A total of 199 adolescents with mood disorders (99 major depressive disorder, 100 bipolar disorder) were stratified into PSU (n = 86) and non-PSU (n = 113) groups using DSM-5–adapted criteria. Resting-state fMRI networks were analyzed with multi-scale graph measures across proportional densities. Hamilton Depression Rating Scale items were modelled as symptom networks to derive symptom communities and brain–symptom associations. Results Global small-world indices did not differ between groups. PSU was instead associated with selective reweighting at modular and nodal levels, including migration of attention regions into a hyper-cohesive reward–limbic core and reduced broadcast roles of executive hubs. Only in the PSU group did greater executive-network cohesion predict more severe anxiety-somatic symptoms. Conclusions In mood-disordered adolescents, PSU is instantiated as selective network reweighting, not global connectome breakdown, and alters the coupling between executive control and anxiety-somatic symptom clusters. These multi-level network signatures suggest mechanistic targets for interventions that rebalance communication among reward, salience and executive systems. Health sciences/Diseases/Psychiatric disorders/Addiction Health sciences/Diseases/Psychiatric disorders/Depression Health sciences/Diseases/Psychiatric disorders/Bipolar disorder Problematic smartphone use Adolescents Mood disorders Resting-state fMRI Functional brain networks Figures Figure 1 Figure 2 Figure 3 Figure 4 Introduction Adolescence is a window of heightened neuroplasticity when large-scale functional networks consolidate into adult-like segregation and integration 1 – 3 . In parallel, smartphones have become embedded in adolescents' daily routines and deliver dense, socially punctuated cues such as alerts, vibrations and short videos that repeatedly capture attention 4 , 5 . For most young people this engagement remains within normative bounds, but among adolescents with mood disorders problematic smartphone use (PSU), defined by persistent overuse with poor control, tolerance- and withdrawal-like symptoms, and functional impairment, is especially common and clinically salient. Affected youths frequently report compulsive checking, sleep disruption and interference with school and family life 6 – 9 . Epidemiological syntheses and clinical observations indicate that, within mood-disordered cohorts, higher PSU severity tracks worse sleep and affective symptoms and greater difficulty sustaining attention, consistent with the idea that repeated cue exposure can exacerbate self-regulatory challenges already present in affective illness 10 . From a psychological perspective, adolescents with mood disorders often describe using smartphones to escape from rumination, regulate negative affect, alleviate loneliness and maintain fragile social ties 11 , 12 . The smartphone environment provides rapid, intermittent rewards and constant opportunities for social comparison and reassurance. These contingencies may transiently relieve dysphoria yet reinforce maladaptive coping strategies. Over time, such patterns can crystallize into compulsive use, nocturnal engagement and the prioritization of smartphone activities over offline responsibilities, which aggravate sleep disturbance, interpersonal conflict and functional impairment. In this context, PSU can be conceptualized as a behavioral comorbidity and modifier of adolescent mood disorders, rather than as a stand-alone diagnosis. Convergent neuroimaging and behavioral findings across technology overuse and behavioral addictions highlight three recurring features. First, salience hubs in the anterior insula and anterior cingulate cortex, together with reward and habit circuits centered on the striatum, show heightened engagement in response to disorder-relevant cues. Second, repeated pairing of digital cues with mood change and social feedback endows such cues with conditioned motivational value that increasingly recruits reward and attention networks. Third, frontoparietal executive control networks exhibit reduced recruitment or weakened coupling with salience and reward systems, suggesting impaired top-down regulation 13 – 16 . Independently, mood disorders are associated with fronto-limbic dysregulation and altered switching between default mode, salience and executive networks 17 . Taken together, these strands of evidence suggest that, in adolescents who already have mood disorders, PSU may further bias large-scale brain dynamics towards reward- and salience-driven processing at the expense of regulatory control, rather than representing an entirely separate neural phenotype 18 – 21 . Network neuroscience offers a principled framework for formalizing these ideas by modelling the brain as a graph of nodes and edges that exhibits small-world architecture, modular organization and hub-mediated communication 22 . Across a range of psychiatric conditions, including depression and behavioral addictions, global small-world metrics such as characteristic path length, clustering and efficiency are often preserved or only modestly altered, whereas more pronounced changes appear at modular and nodal scales. Graph-theoretical studies in internet-related addictions, for example, have reported altered cohesion and centrality of reward, salience and executive modules, together with shifts in the distribution of connector hubs, even when global efficiency remains within typical ranges. This pattern raises the possibility that psychopathology may not primarily reflect a diffuse breakdown of network topology but rather selective adjustments in the relative weighting of communication pathways that are embedded within an otherwise intact architecture. Beyond large-scale brain circuits, mood and anxiety symptoms themselves can be viewed as interacting elements of a network rather than as isolated manifestations of a single latent disease entity. Symptom-network approaches treat individual questionnaire items as nodes and estimate conditional associations between them, yielding symptom communities that often recapitulate clinically recognizable clusters such as core depressive mood and anhedonia, anxiety and somatic tension, sleep disturbance, and appetite or health-concern symptoms 23 . Bridge symptoms that connect these communities are thought to play a pivotal role in maintaining syndromes and mediating transitions between symptom constellations. Symptom-network analyses in depression and anxiety indicate that network structure can differ across clinical subgroups and environmental exposures, suggesting that the organization of symptom communities carries mechanistic and prognostic information. Clinical observations in adolescents with PSU highlight prominent anxiety, somatic complaints and sleep disruption, indicating that these symptom clusters may be particularly relevant for understanding how PSU modifies the course and expression of mood disorders. Despite these converging lines of evidence, several key questions remain unresolved regarding how PSU is embedded within adolescent mood disorders. First, most neuroimaging studies of technology-related overuse have been conducted in community or mixed clinical samples and have treated depressive and anxiety symptoms as covariates or secondary outcomes rather than as the primary clinical context 24 . Consequently, it remains unclear whether, within diagnosed mood disorders, PSU marks a distinct connectome phenotype or simply indexes a more severe expression of the same underlying circuit dysfunction. Second, prior work has rarely characterized the functional connectome of adolescents with and without PSU using multi-scale graph-theoretical analysis. As a result, we do not know whether PSU is associated with a genuine disruption of global integration and segregation, or instead with selective changes in the cohesion and hub roles of canonical systems such as reward, salience and executive networks. Third, brain networks and depressive symptom networks have almost always been examined in isolation; to our knowledge, no study has tested whether PSU modifies the coupling between large-scale functional organization and the structure of depressive symptom communities in mood-disordered adolescents. Addressing these gaps is essential for clarifying whether PSU acts primarily as a quantitative amplifier of mood-disorder severity or as a qualitative modifier that alters brain–symptom mappings on an already vulnerable substrate. To organize these questions, we introduce a Selective Network Reweighting framework. In this framework, PSU in mood-disordered adolescents is conceptualized not as a collapse of global functional topology, but as a redistribution of routing priorities across large-scale brain systems. The overall small-world architecture of the functional connectome is assumed to remain largely preserved, whereas the relative weight carried by communication channels linking reward, salience, executive and semantic systems is altered. This perspective generates testable predictions about the levels of the network hierarchy—global, modular and nodal—at which PSU-related differences should be expressed, and about how such differences should relate to specific constellations of depressive symptoms (Fig. 1 ). In the present study, we applied this framework to a clinically well characterized cohort of adolescents with mood disorders who were stratified according to PSU status. Using resting-state functional MRI and graph-theoretical analysis across a range of proportional densities, we quantified macro-, meso- and nodal-scale properties of the functional connectome. In parallel, we constructed depressive symptom networks from item-level Hamilton Depression Rating Scale scores and identified symptom communities reflecting anxiety and somatic tension, core depressive and psychomotor symptoms, sleep disturbance, and appetite or health-related concerns. Guided by the Selective Network Reweighting framework and by prior work in behavioral addictions and mood disorders, we focused on three questions. First, does PSU in mood-disordered adolescents involve a breakdown of global small-world organization, or are global metrics broadly preserved? Second, are PSU-related differences expressed as selective changes in the configuration and bridging roles of canonical networks such as reward, salience, executive and semantic systems? Third, does PSU alter the pattern of coupling between functional network properties and depressive symptom communities, particularly for anxiety–somatic and sleep-related clusters? By jointly modelling large-scale functional organization and symptom-network structure, this study aims to provide a mechanistic account of how PSU is instantiated in the brains and symptom profiles of adolescents with mood disorders, and to delineate circuit-level targets for early intervention. Methods Participants and clinical characterization This study enrolled a cohort of 199 adolescents aged 13 to 18 years diagnosed with mood disorders, comprising 99 patients with Major Depressive Disorder (MDD) and 100 with Bipolar Disorder (BD) presenting in the depressive episode. Diagnostic confirmation was performed by trained psychiatrists using the Schedule for Affective Disorder and Schizophrenia for School Age Children Present and Lifetime Version (K-SADS-PL) in strict accordance with DSM-5 criteria, ensuring the absence of comorbid Axis I conditions. Based on clinical adjudication, the cohort was stratified into a PSU group ( \(\:n=86\) ; mean age \(\:15.13\pm\:1.45\) years) and a non-PSU comparison group ( \(\:n=113\) ; mean age \(\:15.32\pm\:1.39\) years). PSU status was defined using modified DSM-5 criteria for Internet Gaming Disorder, with “smartphone use” substituting for “gaming” as the target behavior. Adolescents were classified as PSU if they endorsed at least five of the following nine maladaptive criteria within the preceding 12 months : (1) preoccupation, where smartphone activity becomes the dominant daily focus or involves anticipation of future use; (2) withdrawal symptoms, manifesting as irritability, anxiety, or sadness when the device is inaccessible; (3) tolerance, characterized by the need for increasing usage duration to achieve satisfaction; (4) loss of control, defined as unsuccessful attempts to curtail participation; (5) displacement of interests, involving the loss of interest in previous hobbies solely due to smartphone use; (6) continued excessive use despite awareness of resulting psychosocial problems; (7) deception of family members or therapists regarding the extent of usage; (8) dysfunctional coping, utilizing the device to escape or relieve negative affective states such as helplessness, guilt, or anxiety ; and (9)functional impairment, specifically jeopardizing or losing significant relationships, educational, or career opportunities. To provide a dimensional measure of severity, participants completed the Mandarin version of the 26-item Smartphone Addiction Inventory (SPAI) 25 . This scale evaluates four subdomains: compulsive behavior (9 items), functional impairment (8 items), withdrawal symptoms (6 items), and tolerance (3 items). Items are rated on a 4-point Likert scale ranging from “strongly disagree” to “strongly agree”, with the total score indicating addiction severity (Cronbach's \(\:\alpha\:=0.94\) in this sample). Depressive symptom severity was assessed using the 17-item Hamilton Depression Rating Scale (HAMD-17). Candidates were excluded if they presented with: (1) history of major medical illness; (2) history of moderate or severe head trauma, neurological disorders, or intellectual disability; (3) lifetime substance or alcohol dependence; (4) MRI contraindications; (5) suboptimal imaging data quality; or (6) concurrent major physical illness potentially confounding mood symptoms. MRI acquisition and preprocessing Resting-state fMRI data were acquired on a 3 T Siemens Prisma scanner using a gradient-echo EPI sequence (TR = 500 ms, TE = 30 ms) during an eyes-closed resting state for approximately 8 minutes. Participants were instructed to keep their eyes closed, remain still, and stay awake. Preprocessing was performed in SPM12 and involved discarding initial volumes to allow magnetization stabilization, slice-timing correction, rigid-body realignment to the mean functional image, co-registration of the mean EPI to each participant’s T1-weighted anatomical image, tissue segmentation, and normalization to MNI space, followed by 6-mm FWHM Gaussian smoothing 26 . Nuisance regression removed six head-motion parameters as well as mean white-matter and cerebrospinal-fluid signals, after which residual time series were band-pass filtered between 0.01 and 0.10 Hz. Scans were strictly quality-controlled for motion (translation < 3 mm and rotation < 3°); datasets exceeding these limits were excluded. Regional time series were extracted by averaging preprocessed voxel signals within the 90 regions of interest (ROIs) defined by the AAL atlas. Network construction and graph measures Subject-level functional graphs were constructed by parcellating the brain using the Automated Anatomical Labeling (AAL90) atlas 27 . For each participant, preprocessed BOLD time series were averaged within the 90 ROIs to generate a Pearson correlation matrix. Correlations were Fisher r-to-z transformed to stabilize variance. We retained only positive weights to avoid ambiguities in signed geodesics for shortest-path and triangle-based metrics 28 . To standardize sparsity and suppress weak or spurious connections, proportional thresholding was applied, ensuring that every subject retained an identical fraction of the strongest edges at each target density. Metrics requiring connectivity, such as characteristic path length, were computed on the largest connected component. For weighted geodesics, edge length was defined as \(\:\mathcal{l}=1/(w+\epsilon\:)\) with \(\:\epsilon\:>0\) to ensure monotonicity between coupling strength and distance while maintaining numerical stability. To mitigate threshold arbitrariness and emphasize effects persisting across sparsity levels, all graph measures were evaluated across proportional densities ranging from 0.05 to 0.50 in increments of 0.05 29–31 . These were summarized at the subject level by the area under the curve (AUC), a method that samples both sparse backbones—where integration relies on a few strong edges—and denser regimes characterized by local redundancy and clustering 32 . Global network metrics included the clustering coefficient ( \(\:C\) ), global efficiency ( \(\:{E}_{\text{glob}}\) ), local efficiency ( \(\:{E}_{\text{loc}}\) ), characteristic path length ( \(\:L\) ), assortativity, modularity ( \(\:Q\) ), and edge count. Nodal metrics comprised strength (sum of weights), degree (edge count at a given density), closeness (inverse of average weighted geodesic distance), betweenness (fraction of all-pairs shortest paths traversing the node), local efficiency (efficiency of the node’s neighborhood subgraph), participation coefficient ( \(\:P\) ; quantifying the diversity of inter-modular connections), and within-module degree z-score (within-Z; connectivity relative to module peers). Group contrasts followed the sign convention \(\:{\Delta\:}=\text{PSU}-\text{non-PSU}\) . For module-level summaries, nodal metrics were aggregated within aligned modules per subject to preserve the directional meaning of \(\:{\Delta\:}\) at the community scale. Community structure was estimated by applying Louvain optimization to the group-average weighted graph separately for each group 33 . To detect genuine between-group differences in modular composition, we did not impose a single consensus partition. Instead, post-hoc alignment was performed by greedily maximizing Jaccard overlap between ROI sets, yielding one-to-one module correspondences while preserving reconfiguration. Alignment quality was summarized using the adjusted Rand index (ARI) and variation of information (VI), with best-match Jaccard values reported for each module. Methodological safeguards included proportional thresholding to enforce density matching, weighted graph definitions to preserve information lost in binarization, monotonic geodesic definitions ( \(\:\mathcal{l}=1/(w+\epsilon\:)\) ), restriction of path length calculations to the largest connected component to avoid artifacts from isolated nodes, and AUC summarization to down-weight single-threshold idiosyncrasies, thereby improving robustness for adolescent resting-state fMRI analyses. Symptom measures and symptom-network analysis HAMD-17 total and item scores were obtained on the same day as MRI scanning by trained psychiatrists blinded to PSU status. Items were treated as ordered categorical variables reflecting severity within domains such as depressed mood, guilt, sleep disturbance, psychomotor change, anxiety, and somatic symptoms. To derive symptom modules, group-specific symptom networks were constructed using Gaussian graphical models (GGMs) 34 . For each group, HAMD-17 item scores were median-imputed and transformed using a rank-based inverse-normal (nonparanormal) transform followed by z-scaling. Partial correlation networks were estimated using the Graphical Lasso with five-fold cross-validation to select the regularization parameter \(\:\lambda\:\) . Precision matrices were converted to signed partial correlations with diagonals set to zero. To obtain a single shared symptom partition for cross-group comparisons, absolute partial correlations were averaged across groups, and greedy modularity maximization (Newman–Girvan) was applied to the resulting graph. This yielded four symptom communities (modules 0–3) capturing distinct clusters: (i) anxiety, somatic tension, and negative-self symptoms including guilt, psychomotor agitation, systemic and somatic anxiety, reduced sexual interest, and self-consciousness; (ii) core depressive and psychomotor symptoms including depressed mood, suicidality, work or interest loss, and psychomotor retardation; (iii) sleep disturbance encompassing initial, middle, and early-morning insomnia; and (iv) appetite and health-concern symptoms such as weight loss, hypochondriasis, and gastrointestinal complaints. On the group-specific GGMs, we computed node-wise Strength (sum \(\:\left|w\right|\) ), Expected Influence (sum \(\:w\) ), participation coefficient, within-Z, bridge strength, and bridge expected influence, alongside conventional centralities including degree, betweenness, closeness, eigenvector centrality, and weighted clustering using NetworkX. Global metrics at the symptom level included density, global strength, mean node strength, average clustering, global efficiency, mean shortest-path length, and modularity evaluated on the shared symptom partition. To test for between-group differences in symptom-network architecture, a permutation-based Network Comparison Test (NCT) was implemented. Using a fixed \(\:\lambda\:\) (mean of the two cross-validated values), GGMs were re-estimated on PSU and non-PSU data to compute: (i) differences in global strength, (ii) edge-wise differences in partial correlations, and (iii) differences in nodal and global metrics. Group labels were randomly permuted while preserving group sizes, and networks were re-estimated 1,000 times to generate null distributions. Two-sided permutation \(\:P\) -values were calculated as the proportion of null differences with absolute magnitudes equal to or greater than the observed difference. Edge-level and node-level \(\:P\) -values were controlled for false discovery rate (FDR) using the Benjamini–Hochberg procedure within families, while global-metric \(\:P\) -values were FDR-controlled across metrics. Statistical analysis All between-group effects were defined as \(\:{\Delta\:}=\text{PSU}-\text{non-PSU}\) . For each subject and graph metric, density curves were summarized by the area under the curve (AUC), providing a single value per metric per subject to mitigate multiple comparisons across thresholds. Global metrics including clustering coefficient ( \(\:C\) ), global efficiency ( \(\:{E}_{\text{glob}}\) ), local efficiency ( \(\:{E}_{\text{loc}}\) ), characteristic path length ( \(\:L\) ), modularity ( \(\:Q\) ), assortativity, and edge count were compared between groups using two-sided Welch’s $ t $ -tests with Satterthwaite degrees of freedom, ensuring robustness to unequal variances and sample sizes. Node-wise analyses treated each metric as a family of 90 ROI tests; within each metric, Welch’s \(\:t\) -tests were computed at every node, and Benjamini–Hochberg FDR correction was applied across the 90 tests, with \(\:q<0.05\) denoting statistical significance. For all contrasts, we reported the test statistic ( \(\:t\) ), the FDR-adjusted \(\:q\) value, and Hedges’ \(\:g\) (bias-corrected standardized mean difference) signed by \(\:{\Delta\:}\) to retain directionality (positive values indicate \(\:\text{PSU}>\text{non-PSU}\) ). Module-level summaries were derived by aggregating nodal metrics within post-hoc aligned modules for each subject and contrasting groups with Welch’s tests. Given the limited number of modules and dependency among constituent nodes, interpretation focused on effect directions and convergence with nodal and community findings rather than strict family-wise inference at this level. Secondary analyses examined the relationship between network-level brain metrics and symptom expression. HAMD-17 items were first summarized into subject-level symptom-cluster scores based on the symptom partition: item scores were z-standardized across the full sample and averaged within each module, yielding four continuous outcomes (module0_mean to module3_mean) per subject. A subject-level brain-metric dataset was assembled comprising (i) node-wise AUC betweenness for the left supramarginal gyrus ( \(\:SM{G}_{L}\) ) and right inferior temporal gyrus ( \(\:Tempora{l}_{In{f}_{R}}\) ), and (ii) module-level AUC metrics obtained by averaging nodal values within aligned modules, specifically focusing on executive (M1) and reward–limbic (M3) communities. For each symptom module, an ordinary least-squares (OLS) regression model was fitted: $$\:\text{moduleX\mean}\sim\:\text{group}+\text{brain\metrics}+(\text{brain_metrics}\times\:\text{group})$$ where group coded PSU status ( \(\:0=\text{non-PSU}\) , \(\:1=\text{PSU}\) ). All models included an intercept and were estimated with HC3 robust standard errors. \(\:P\) -values for all terms within each outcome were corrected using the Benjamini–Hochberg FDR method ( \(\:q<0.05\) ). Complementary within-group relationships were examined using Spearman rank correlations between each brain metric and symptom-module score separately for PSU and non-PSU groups, with BH-FDR correction applied within each group. Finally, multivariate coupling between the full set of brain metrics and the four symptom clusters was characterized using partial least squares (PLS) regression with 2 components and 5-fold cross-validation. The cross-validated \(\:{R}^{2}\) for each symptom module was compared against a null distribution generated by permuting subject labels 1,000 times; two-sided permutation \(\:p\) -values were computed as the proportion of null \(\:{R}^{2}\) values exceeding the observed magnitude. Results Global topology: small-world architecture preserved Across proportional densities (AUC 0.05–0.50), global topological metrics showed no evidence of group differences. No metrics survived Benjamini–Hochberg FDR correction ( \(\:q>0.05\) ). For example, the between-group difference in modularity was negligible ( \(\:{\Delta\:}=0.0018\) , Hedges' \(\:g=0.13\) , 95% CI \(\:[-\text{0.15,0.41}]\) ), and global efficiency was similarly stable ( \(\:{\Delta\:}=-0.0005\) , Hedges' \(\:g=-0.11\) , 95% CI \(\:[-\text{0.39,0.17}]\) ). Taken together, these findings indicate no evidence for disruption of macro-scale small-world organization in adolescents with PSU (Fig. 2 ). Module composition overview We next examined the modular organization of the functional networks. Post-hoc alignment revealed that 76 out of 90 ROIs retained their modular assignment across groups (Adjusted Rand Index \(\:=0.70\) ; Variation of Information \(\:=0.82\) ). Changes in modular assignment were observed in specific systems. Ten ROIs, including the caudate, putamen, pallidum, thalamus, and medial cingulate, shifted from the somatomotor and attention module (M0) to the reward and limbic module (M3). Additionally, four ROIs moved from the language and semantic module (M2) to the executive module (M1). These shifts corresponded to changes in module size, with M0 decreasing from 28 to 18 regions, M3 increasing from 8 to 18 regions, M2 decreasing from 22 to 18 regions, and M1 increasing from 16 to 20 regions. (Fig. 3 ) At the aggregated module level, the reward and limbic community (M3) demonstrated increased strength, degree, local efficiency, and within-module degree, combined with decreased participation coefficient and betweenness centrality in the PSU group. This pattern suggests a configuration characterized by high internal connectivity and low external integration. Conversely, the executive module (M1) and the language and semantic module (M2) showed reductions in degree, strength, and closeness centrality. The participation coefficient was also lower in M1. The somatomotor (M0) and visual (M4) modules showed minimal differences between groups. (Table 1 ) Table 1 Module-level contrasts (AUC; Δ=PSU-non-PSU). \(\:{\Delta\:}=\text{PSU}-\text{non-PSU}\) Metric M0 M1 M2 M3 M4 Strength -0.0280 -0.1420 -0.1940 0.3470 0.0410 Degree -0.0008 -0.0028 -0.0035 0.0031 0.0028 Closeness 0.0006 -0.0012 -0.0011 0.0007 0.0006 Local efficiency -0.0027 -0.0022 -0.0035 0.0072 -0.0002 Participation coefficient \(\:P\) 0.0071 -0.0040 0.0076 -0.0136 0.0074 Within-Z -0.0187 0.0308 -0.0339 0.0635 -0.0283 Betweenness 0.0008 0.0001 0.0001 -0.0007 -0.0002 Nodal scale: redistributed bridges Analysis of nodal centrality revealed two regions that survived FDR correction for betweenness centrality. Compared to the non-PSU group, the PSU group exhibited significantly higher betweenness in the left supramarginal gyrus ( \(\:SM{G}_{L}\) ; \(\:t=3.49\) , \(\:q=0.036\) , Hedges' \(\:g=0.51\) , 95% CI \(\:\left[\text{0.23,0.80}\right]\) ) and significantly lower betweenness in the right inferior temporal gyrus ( \(\:Tempora{l}_{In{f}_{R}}\) ; \(\:t=-3.41\) , \(\:q=0.036\) , Hedges' \(\:g=-0.45\) , 95% CI \(\:[-0.73,-0.16]\) ). (Fig. 4 ) Beyond these specific nodes, other centrality metrics (degree, strength, closeness) showed small, dispersed effects across the brain (mean \(\:\left|g\right|=0.15\) , \(\:0.15\) and \(\:0.14\) , respectively), with no other regions surviving multiple comparison correction. Symptom-network organization Gaussian graphical models estimated separately for PSU and non-PSU groups yielded four stable symptom communities: (i) an anxiety, somatic tension, and negative-self cluster (Module 0); (ii) a core depressive and psychomotor cluster (Module 1); (iii) a sleep disturbance cluster (Module 2); and (iv) an appetite and health-concern cluster (Module 3). Global network properties were comparable between groups. No significant differences were found in density, global strength, mean node strength, average clustering, global efficiency, mean shortest-path length, or modularity after permutation testing and FDR control (all \(\:q>0.05\) ). Similarly, permutation-based Network Comparison Tests revealed no robust between-group differences in edge-wise partial correlations after correction. At the nodal level, a single significant difference was identified: the participation coefficient of the systemic somatic symptoms item (HAMD-17 systemic symptoms, Module 0) was lower in the PSU group compared to the non-PSU group (non-PSU mean \(\:=0.73\) PSU mean \(\:=0.23\) , \(\:{\Delta\:}=-0.50\) , permutation \(\:p=0.002\) , FDR \(\:q=0.034\) ). Brain–symptom coupling We further examined the relationship between network-level brain metrics and symptom expression using the four symptom communities identified above. OLS regression models identified a significant interaction between group status and executive module cohesion (M1 within-Z) predicting the severity of the anxiety, somatic tension, and negative-self cluster (Module 0) ( \(\:{\beta\:}_{\text{interaction}}=2.58\) , \(\:p=0.0032\) , FDR \(\:q=0.044\) ; Table 2 ). Specifically, in the PSU group, greater executive network cohesion was associated with more severe Module 0 symptoms ( \(\:{\beta\:}_{\text{PSU}}=1.34\) ). In contrast, the non-PSU group showed a trend toward a negative association ( \(\:{\beta\:}_{\text{non-PSU}}=-1.24\) , \(\:p=0.013\) , FDR \(\:q=0.092\) ). Table 2 Within-group Spearman correlations between brain metrics and symptom scores Group Brain metric Symptom score Spearman \(\:\rho\:\) p FDR-q PSU M1 within-Z Module 0 Mean 0.3380 0.0015 0.0350 non-PSU M1 within-Z Module 0 Mean -0.1334 0.1595 0.6855 Within-group Spearman correlations confirmed this pattern. In the PSU group, M1 within-Z was positively correlated with Module 0 scores (Spearman \(\:\rho\:=0.338\) , \(\:p=0.0015\) , FDR \(\:q=0.035\) ). No significant correlation was observed in the non-PSU group (FDR \(\:q>0.10\) ; Table 3 ). No other brain–symptom associations survived FDR correction. Table 3 Brain–symptom OLS regression results Symptom Module Group \(\:\beta\:\) \(\:p\) FDR- \(\:q\) Module 0 Mean PSU 2.5770 0.0032 0.0442 Module 0 Mean non-PSU -1.2395 0.0131 0.0919 Discussion In this study, we examined resting-state functional network organization and depressive symptom structure in adolescents with mood disorders, stratified by the presence of problematic smartphone use (PSU), within a Selective Network Reweighting framework. Three main findings emerged. First, at the macro-scale, small-world topology did not differ meaningfully between PSU and non-PSU groups: global efficiency, path length, clustering and modularity showed no evidence of disruption. Second, at the meso- and nodal scales, there was a selective reconfiguration of network weights: somatomotor and attention-related nodes migrated into the reward–limbic module, which exhibited higher internal connectivity and lower cross-module integration, whereas executive and language–semantic modules showed reduced integrative roles and a redistribution of bridge centrality from ventral temporal to parietal attention regions. Third, at the level of symptom networks and brain–symptom coupling, the overall architecture of depressive symptom communities was highly similar between groups, but only in the PSU group was greater executive-network cohesion associated with more severe anxiety–somatic–negative self-symptoms. Taken together, these observations are consistent with a model in which PSU in mood-disordered adolescents is instantiated not as global connectome breakdown, but as selective reweighting of specific systems and a modification of the mapping between brain networks and symptom clusters. At the macro-scale, the absence of detectable group differences in small-world metrics suggests that PSU does not manifest as a generalized disintegration of functional topology in this clinical context 19 . Global efficiency, clustering and modularity were all comparable between PSU and non-PSU adolescents, with small effect sizes and confidence intervals spanning zero. This pattern contrasts with reports of widespread topological disruption in neurodegenerative disorders and severe psychotic illness, where global efficiency is often reduced and characteristic path length increased 35 – 37 . In adolescents with mood disorders, our findings indicate that the basic capacity of the functional connectome to support integrated and segregated processing appears largely intact irrespective of PSU status. Behavioral manifestations such as impulsive checking and sustained distractibility therefore cannot be straightforwardly attributed to a loss of global network capacity. Rather, they are more plausibly linked to changes in how specific subcircuits are weighted and coupled on top of a preserved macro-scale architecture. Conceptualizing PSU as a disturbance of routing priorities, rather than as a deficit in structural capability, motivates a focus on meso- and nodal-level reorganization. Against this background of global stability, we observed systematic reweighting at the module level. Post hoc alignment showed that most nodes retained their community membership across groups, but a subset of somatomotor and attention-related regions—including basal ganglia and medial cingulate structures—shifted from a somatomotor–attention module into a reward–limbic module, which increased in size. A smaller set of language–semantic regions moved into the executive module. In the PSU group, the reward–limbic community displayed higher degree, strength, local efficiency and within-module degree, together with lower participation coefficient and betweenness centrality, yielding a profile of stronger internal cohesion but weaker cross-module integration. In contrast, executive and language–semantic modules showed reductions in degree, strength and closeness, and the participation coefficient of the executive module was lower. This pattern echoes prior graph-theoretical findings in behavioral addictions and internet-related overuse, where reward and salience networks often show increased internal cohesion, while frontoparietal control systems lose some of their meditation role between communities. Our results extend these observations to adolescents with mood disorders, suggesting that in those who meet PSU criteria, smartphone-related cue processing, action routines and affective responses are more tightly embedded within a reward–limbic circuit, whereas executive and semantic systems carry less weight in cross-system integration 5 , 38 . Nodal-level results complement this modular reweighting. After correction for multiple comparisons, only two regions showed robust differences in betweenness centrality: the left supramarginal gyrus exhibited higher betweenness in the PSU group, while the right inferior temporal gyrus showed lower betweenness. The supramarginal gyrus forms part of the ventral attention network and is implicated in reorienting towards salient external stimuli; increased betweenness suggests that it lies on a greater proportion of shortest paths and mediates more inter-modular communication 39 . This topology is compatible with clinical observations that adolescents with PSU are highly responsive to notifications and other salient smartphone cues. By contrast, the inferior temporal gyrus contributes to high-level visual–semantic integration. Reduced betweenness may indicate a diminished bridging role between visual–semantic processing and other systems, implying that the route from perception to deeper semantic appraisal carries less weight in the overall shortest-path structure 40 . These interpretations remain necessarily indirect, as they are inferred from resting-state connectivity rather than task-evoked information flow, but they illustrate how a “parietal attention–dominant, temporal semantic–recessive” routing configuration could bias behavior toward rapid, cue-driven responses with less engagement of elaborative appraisal 4 . At the level of symptom networks, we found that both PSU and non-PSU groups exhibited a similar community structure when Hamilton Depression Rating Scale items were modelled as Gaussian graphical models. In each group, four communities emerged, broadly corresponding to anxiety–somatic tension–negative self, core depressive and psychomotor, sleep disturbance, and appetite/health-concern clusters. Global indices such as density, global strength, average clustering, global efficiency, mean shortest path length and modularity did not differ significantly between groups, and edge-wise network comparison tests did not reveal robust differences after correction. Thus, within this sample of mood-disordered adolescents, PSU did not appear to reshape the overall organization of depressive symptom co-occurrence. The only node-level difference that survived correction was a lower participation coefficient for the systemic somatic symptoms item in the PSU group, indicating that this symptom was more confined to its own community and played a weaker bridging role across communities. One possible interpretation is that, among youths with PSU, bodily complaints are more tightly embedded within anxiety and negative self-evaluation, and less connected to other symptom domains, aligning with clinical impressions of recurrent health worries and bodily focus that are closely tied to stress and self-criticism. Given that this effect was confined to a single item, it should be interpreted cautiously and replicated in independent samples 23 , 41 . The most distinctive evidence for a PSU-related modification emerged when we examined brain–symptom coupling. Across all tested brain metrics and symptom modules, only one interaction survived FDR correction: a group-by-executive-module cohesion effect predicting the anxiety–somatic–negative self-symptom cluster. In the PSU group, greater executive-module within-Z was associated with more severe symptoms in this community, whereas in the non-PSU group the association was negative but did not survive correction. Within-group rank correlations showed a broadly similar pattern: executive cohesion correlated positively with anxiety–somatic symptoms in the PSU group, but no significant correlation was observed in the non-PSU group after correction. These findings suggest that the clinical meaning of executive-network cohesion may depend on behavioral context 7 . In adolescents without PSU, a more cohesive executive network might reflect an effective regulatory resource that helps to contain anxiety and somatic tension. In contrast, in the context of PSU, greater executive cohesion may index a state of high effort but low efficiency, in which control resources are heavily engaged in monitoring, worry and attempts at suppression that do not translate into effective down-regulation of anxiety or bodily arousal. Because our design is cross-sectional, it is not possible to determine whether this pattern reflects a compensatory response to high symptom burden, a pre-existing vulnerability that predisposes to PSU, or a consequence of prolonged PSU-related engagement. Nevertheless, the presence of a PSU-specific interaction supports the idea that PSU can act as a modifier of brain–symptom mappings, rather than as a simple proxy for depression severity. From a theoretical perspective, a key contribution of this work is to position PSU within adolescent mood disorders as a behavioral modifier, and to provide convergent evidence for a Selective Network Reweighting account at both brain and symptom levels. We observed that, on a preserved small-world architecture, PSU was associated with a reallocation of connectivity weights among reward–limbic, attention, executive and semantic systems, altering cross-module routes and bridge nodes, while leaving global topology largely unchanged. At the same time, executive-network metrics showed different relationships to anxiety–somatic symptom communities in adolescents with and without PSU. This pattern—reweighting of modules and nodes combined with a PSU-dependent modulation of brain–symptom coupling—is consistent with the notion that PSU “sculpts” new behavioral and symptomatic phenotypes on top of an existing mood-disorder architecture 8 . More broadly, the Selective Network Reweighting framework may offer a transferable lens for understanding other digital or behavioral comorbidities, such as gaming disorder or excessive social media use, within psychiatric populations. The findings also carry potential implications for clinical and translational work. In terms of risk stratification, the combination of a more cohesive but externally less integrated reward–limbic module, reduced cross-module roles of executive and semantic communities, and elevated centrality of ventral attention nodes could define a candidate network profile for high PSU risk within mood-disordered adolescents, to be tested in longitudinal designs. In terms of intervention, the framework suggests that simply reducing total screen time may be insufficient. Instead, behavioral and psychotherapeutic strategies might aim to rebalance routing priorities by diminishing the salience and contingency of smartphone cues, strengthening flexible integration of executive and semantic systems with other modules, and reducing the closedness of reward–limbic loops 42 – 44 . Environmental modifications such as managing notification frequency and timing, structuring phone-free periods, or altering interface features to reduce cue salience may alleviate the burden on attention–reward pathways. Cognitive and emotion-regulation interventions, including training that targets shifting, inhibition and reappraisal, or approaches that enhance interoceptive and metacognitive awareness, may help restore the executive module’s role as a flexible integrator rather than an internally locked, over-engaged system. Several limitations should be considered when interpreting these findings. First, the study is cross-sectional, which precludes causal inference: we cannot determine whether the observed network reweighting precedes PSU and contributes to its development, emerges as a consequence of sustained PSU, or reflects a bidirectional process. Longitudinal and interventional studies will be essential to clarify temporal ordering. Second, the sample was drawn from a single clinical setting and included adolescents with both major depressive and bipolar disorders; although key demographic and clinical covariates were controlled statistically, medication use, illness duration and other factors may still confound the results, and generalizability to other populations remains uncertain. Third, network construction relied on a single anatomical parcellation (AAL90) and resting-state functional connectivity; spatial resolution and sensitivity to task-specific processes are limited. Future work using finer-grained parcellations, dynamic connectivity, task-based fMRI and multimodal imaging will be important to test the robustness and specificity of the selective reweighting patterns. Fourth, PSU was assessed using DSM-5–adapted criteria and self-report measures without objective smartphone usage logs; different types of smartphone activities may be associated with distinct network profiles. Finally, our symptom networks were based solely on HAMD items and did not explicitly include domains such as impulsivity, reward sensitivity or social anxiety, which are likely to be relevant for PSU. In summary, by comparing PSU and non-PSU adolescents within a mood-disordered cohort, and by integrating multi-scale graph analysis with symptom-network modelling, this study suggests that PSU is associated with selective network reweighting and altered brain–symptom coupling rather than with wholesale breakdown of functional architecture. Reward–limbic and attention systems become more tightly coupled, executive and semantic systems lose some of their cross-module integrative roles, and the association between executive-network cohesion and anxiety–somatic symptoms is reparametrized in the presence of PSU. Viewing PSU through the lens of routing priorities and selective network reweighting provides a mechanistic framework for understanding its role in adolescent mood disorders and offers a basis for developing network-informed stratification and intervention strategies. Declarations Acknowledgments This work was supported by the National Natural Science Foundation of China [62176129], National Natural Science Foundation of China-Jiangsu Joint Fund [U24A20701], the Hong Kong RGC Strategic Target Grant [grant number STG1/M-501/23-N] Ethics Ethical approval for this study was granted by the Ethics Committee of the Affiliated Nanjing Brain Hospital, Nanjing Medical University (Nanjing, China). Written informed consent was obtained from all participants and their parents or legal guardians following a full explanation of the study procedures. Conflict of interest The authors declare no conflict of interest to this work. References Han, L., Chan, M. Y., Agres, P. F., Winter-Nelson, E., Zhang, Z., & Wig, G. S. (2024). Measures of resting-state brain network segregation and integration vary in relation to data quantity: Implications for within and between subject comparisons of functional brain network organization. Cerebral Cortex , 34 (2), bhad506. Wang, R., Liu, M., Cheng, X., Wu, Y., Hildebrandt, A., & Zhou, C. (2021). Segregation, integration, and balance of large-scale resting brain networks configure different cognitive abilities. Proceedings of the National Academy of Sciences , 118 (23), e2022288118. Song, Z., Jiang, Z., Zhang, Z., Wang, Y., Chen, Y., Tang, X., & Li, H. (2024). Evolving brain network dynamics in early childhood: Insights from modular graph metrics. NeuroImage , 297 , 120740. Gavilan, D., Fernández-Lores, S., & Martinez-Navarro, G. (2020). Vividness of news push notifications and users’ response. Technological Forecasting and Social Change , 161 , 120281. Dekker, C. A., Baumgartner, S. E., Sumter, S. R., & Ohme, J. (2025). Beyond the Buzz: Investigating the Effects of a Notification-Disabling Intervention on Smartphone Behavior and Digital Well-Being. Media Psychology , 28 (1), 162–188. Mayerhofer, D., Haider, K., Amon, M., Gächter, A., O’Rourke, T., Dale, R., Humer, E., Probst, T., & Pieh, C. (2024). The Association between Problematic Smartphone Use and Mental Health in Austrian Adolescents and Young Adults. Healthcare , 12 (6), 600. Ndayambaje, E., & Okereke, P. U. (2025). The Psychopathology of Problematic Smartphone Use (PSU): A Narrative Review of Burden, Mediating Factors, and Prevention. Health Science Reports , 8 (5), e70843. Poulain, T., Meigen, C., Kiess, W., & Vogel, M. (2025). Smartphone use, wellbeing, and their association in children. Pediatric Research , 1–7. Ahmed, O., Walsh, E. I., Dawel, A., Alateeq, K., Espinoza Oyarce, D. A., & Cherbuin, N. (2024). Social media use, mental health and sleep: A systematic review with meta-analyses. Journal of Affective Disorders , 367 , 701–712. Kim, S.-G., Park, J., Kim, H.-T., Pan, Z., Lee, Y., & McIntyre, R. S. (2019). The relationship between smartphone addiction and symptoms of depression, anxiety, and attention-deficit/hyperactivity in South Korean adolescents. Annals of General Psychiatry , 18 , 1. Zhang, R., Wang, Y., Womer, F., Yang, W., Wang, X., Xu, X., Su, Z., Wang, L., Zhou, J., Qin, S., Zhou, K., Liu, F., Zhang, X., Lin, Y., Zhang, X., Yang, J., & Wang, F. (2023). School-based Evaluation Advancing Response for Child Health (SEARCH): A mixed longitudinal cohort study from multifaceted perspectives in Jiangsu, China. BMJ Mental Health , 26 (1). Zheng, J., Zong, X., Tang, L., Guo, H., Zhao, P., Womer, F. Y., Zhang, X., Tang, Y., & Wang, F. (2024). Characterizing the distinct imaging phenotypes, clinical behavior, and genetic vulnerability of brain maturational subtypes in mood disorders. Psychological Medicine , 54 (10), 2774–2784. León Méndez, M., Padrón, I., Fumero, A., & Marrero, R. J. (2024). Effects of internet and smartphone addiction on cognitive control in adolescents and young adults: A systematic review of fMRI studies. Neuroscience & Biobehavioral Reviews , 159 , 105572. Lee, S., Cheong, Y., Ro, J., Bae, J., & Jung, M. (2024). Alterations in functional connectivity in the salience network shared by depressive symptoms and smartphone overuse. Scientific Reports , 14 (1), 28679. Liu, D., Liu, X., Long, Y., Xiang, Z., Wu, Z., Liu, Z., Bian, D., & Tang, S. (2022). Problematic smartphone use is associated with differences in static and dynamic brain functional connectivity in young adults. Frontiers in Neuroscience , 16 . Áfra, E., Janszky, J., Perlaki, G., Orsi, G., Nagy, S. A., Arató, Á., Szente, A., Alhour, H. A. M., Kis-Jakab, G., & Darnai, G. (2024). Altered functional brain networks in problematic smartphone and social media use: Resting-state fMRI study. Brain Imaging and Behavior , 18 (2), 292–301. Ong, R. H. S., Sim, H. S., Bergman, M. M., How, C. H., Png, C. A. L., Lim, C. S., Peh, L. H., & Oh, H. C. (2024). Prevalence and associations of problematic smartphone use with smartphone activities, psychological well-being, and sleep quality in a household survey of Singapore adults. PLOS ONE , 19 (12), e0315364. Mo, X., Li, X., Liu, M., Hu, L., Li, Q., Wang, J., Deng, H., Lv, F., Zhou, X., Mao, Y., & Huang, Y. (2025). Resting-state fMRI graph theory analysis for predicting selective serotonin reuptake inhibitors treatment response in adolescent major depressive disorder. Frontiers in Psychiatry , 16 . Han, Y.-K., Zhang, H.-J., Chen, Y.-J., Liu, C., Zhang, Y.-H., Zhang, Z.-J., Jing, R.-T., Guo, L., Li, D., Chu, W.-Y., Wu, W.-J., Zhang, K., & Cui, L.-B. (2025). Small-world network and neuroscience. Brain-X , 3 (1), e70025. Hallquist, M. N., & Hillary, F. G. (2018). Graph theory approaches to functional network organization in brain disorders: A critique for a brave new small-world. Network Neuroscience , 3 (1), 1–26. Upshaw, J. D., Stevens, C. E., Ganis, G., & Zabelina, D. L. (2022). The hidden cost of a smartphone: The effects of smartphone notifications on cognitive control from a behavioral and electrophysiological perspective. PLOS ONE , 17 (11), e0277220. Bassett, D. S., & Sporns, O. (2017). Network neuroscience. Nature Neuroscience , 20 (3), 353–364. Cai, H., Bai, W., Liu, H., Chen, X., Qi, H., Liu, R., Cheung, T., Su, Z., Lin, J., Tang, Y., Jackson, T., Zhang, Q., & Xiang, Y.-T. (2022). Network analysis of depressive and anxiety symptoms in adolescents during the later stage of the COVID-19 pandemic. Translational Psychiatry , 12 (1), 98. Cushnie, A. K., Tang, W., & Heilbronner, S. R. (2023). Connecting Circuits with Networks in Addiction Neuroscience: A Salience Network Perspective. International Journal of Molecular Sciences , 24 (10), 9083. Lin, Y.-H., Chang, L.-R., Lee, Y.-H., Tseng, H.-W., Kuo, T. B. J., & Chen, S.-H. (2014). Development and Validation of the Smartphone Addiction Inventory (SPAI). PLOS ONE , 9 (6), e98312. Pan, C., Ma, Y., Wang, L., Zhang, Y., Wang, F., & Zhang, X. (2024). From Connectivity to Controllability: Unraveling the Brain Biomarkers of Major Depressive Disorder. Brain Sciences , 14 (5), 509. Rolls, E. T., Huang, C.-C., Lin, C.-P., Feng, J., & Joliot, M. (2020). Automated anatomical labelling atlas 3. NeuroImage , 206 , 116189. Chen, Y., Zhao, P., Pan, C., Chang, M., Zhang, X., Duan, J., Wei, Y., Tang, Y., & Wang, F. (2024). State- and trait-related dysfunctions in bipolar disorder across different mood states: A graph theory study. Journal of Psychiatry and Neuroscience , 49 (1), E11–E22. Buchanan, C. R., Bastin, M. E., Ritchie, S. J., Liewald, D. C., Madole, J., Tucker-Drob, E. M., Deary, I. J., & Cox, S. R. (2020). The effect of network thresholding and weighting on structural brain networks in the UK Biobank. NeuroImage , 211 , 116443. Theis, N., Rubin, J., Cape, J., Iyengar, S., & Prasad, K. M. (2023). Threshold Selection for Brain Connectomes. Brain Connectivity , 13 (7), 383–393. Pan, C., Zhang, X., Zheng, H., Zhang, Y., Su, Z., Zhang, C., & Zhang, W. (2025). Adaptive Control of Dynamic Networks. IEEE Transactions on Network Science and Engineering , 1–15. Hassett, J. D., Craig, B. T., Hilderley, A., Kinney-Lang, E., Yeates, K. O., MacMaster, F. P., Miller, J., Noel, M., Brooks, B. L., Barlow, K., Lebel, C., Kirton, A., & Carlson, H. L. (2024). Development of the whole-brain functional connectome explored via graph theory analysis. Aperture Neuro , 4 . Fortunato, S., & Hric, D. (2016). Community detection in networks: A user guide. Physics Reports: Review Section of Physics Letters , 659 , 1–44. Pan, C., Zhang, Q., Zhu, Y., Kong, S., Liu, J., Zhang, C., Wang, F., & Zhang, X. (2024). Module control of network analysis in psychopathology. iScience , 27 (7). Zhou, L., Xie, Y., Wang, R., Fan, Y., & Wu, Y. (2023). Dynamic segregation and integration of brain functional networks associated with emotional arousal. iScience , 26 (5), 106609. Montag, C., & Becker, B. (2023). Neuroimaging the effects of smartphone (over-)use on brain function and structure—A review on the current state of MRI-based findings and a roadmap for future research. Psychoradiology , 3 , kkad001. Ahn, J., Lee, D., Namkoong, K., & Jung, Y.-C. (2021). Altered Functional Connectivity of the Salience Network in Problematic Smartphone Users. Frontiers in Psychiatry , 12 , 636730. Wolf, N. D., Henemann, G. M., Schmitgen, M. M., Koenig, J., Bach, P., & Wolf, R. C. (2025). Screens and brains: Multimodal neuroimaging insights into mechanisms of problematic smartphone use. Progress in Neuro-Psychopharmacology and Biological Psychiatry , 142 , 111531. Einhäuser, W., Neubert, C. R., Grimm, S., & Bendixen, A. (2024). High visual salience of alert signals can lead to a counterintuitive increase of reaction times. Scientific Reports , 14 (1), 8858. Zhou, J., Duan, J., Liu, X., Wang, Y., Zheng, J., Tang, L., Zhao, P., Zhang, X., Zhu, R., & Wang, F. (2024). Functional network characteristics in adolescent psychotic mood disorder: Associations with symptom severity and treatment effects. European Child & Adolescent Psychiatry , 33 (7), 2319–2329. Wang, S.-B., Xu, W.-Q., Gao, L.-J., Tan, W.-Y., Zheng, H.-R., Hou, C.-L., & Jia, F.-J. (2023). Bridge connection between depression and anxiety symptoms and lifestyles in Chinese residents from a network perspective. Frontiers in Psychiatry , 14 . Doll, A., Hölzel, B. K., Boucard, C. C., Wohlschläger, A. M., & Sorg, C. (2015). Mindfulness is associated with intrinsic functional connectivity between default mode and salience networks. Frontiers in Human Neuroscience , 9 , 461. Bremer, B., Wu, Q., Mora Álvarez, M. G., Hölzel, B. K., Wilhelm, M., Hell, E., Tavacioglu, E. E., Torske, A., & Koch, K. (2022). Mindfulness meditation increases default mode, salience, and central executive network connectivity. Scientific Reports , 12 (1), 13219. Yue, W. L., Ng, K. K., Koh, A. J., Perini, F., Doshi, K., Zhou, J. H., & Lim, J. (2023). Mindfulness-based therapy improves brain functional network reconfiguration efficiency. Translational Psychiatry , 13 , 345. Additional Declarations The authors have declared there is NO conflict of interest to disclose Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-8256165","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Article","associatedPublications":[],"authors":[{"id":554366370,"identity":"5f816fc2-876b-4815-b9c3-9ddcd168ce7d","order_by":0,"name":"Xizhe Zhang","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAA9ElEQVRIiWNgGAWjYDACCShpwMB8AMpJIFoLWwJJWhiAWngMoEwCWuRnNz97+LXNwt6cvefbA8ucwwz87DkGDD934NbCOOeYubHMGQlmy56z2w0ktx1mkOx5Y8DYewa3FmaJBDNpiQoJNoMbudskQFoMbuQYMDO24dbCJpH+TVrCQILH4P6bZ2At9oS08EjkmEl+qJCQMLjBwwaxRYKAFgmJnDJphjMSBgZn0syAWtJ5JM48KzjYi0eL/Iz0bZI/2+rsDY4ffiYtuc1ajr89eeODn3i0gIOAB8YARhKYfQC/BmBA/4AxPhBSOgpGwSgYBSMSAABLwEg7M1rBbAAAAABJRU5ErkJggg==","orcid":"https://orcid.org/0000-0002-8684-4591","institution":"Nanjing Medical University","correspondingAuthor":true,"prefix":"","firstName":"Xizhe","middleName":"","lastName":"Zhang","suffix":""},{"id":554366371,"identity":"35e606db-c0ed-4c14-a970-4917c2ea49d3","order_by":1,"name":"Jiayang Zhu","email":"","orcid":"","institution":"Nanjing Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jiayang","middleName":"","lastName":"Zhu","suffix":""},{"id":554366372,"identity":"b810c97e-2331-43df-8ef4-00b858c0c536","order_by":2,"name":"Jing Zhang","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Zhang","suffix":""}],"badges":[],"createdAt":"2025-12-02 05:35:31","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-8256165/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-8256165/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":97320060,"identity":"a42cf312-9a85-4614-81e0-20376c901c6b","added_by":"auto","created_at":"2025-12-03 07:39:01","extension":"docx","order_by":0,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":8044737,"visible":true,"origin":"","legend":"","description":"","filename":"SelectiveNetworkReweighting1.0.docx","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/3fea7f29a8bb413e9b2c0cf3.docx"},{"id":97320064,"identity":"826b5624-1bb9-41fc-b12e-92245666ad5c","added_by":"auto","created_at":"2025-12-03 07:39:02","extension":"json","order_by":1,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":5207,"visible":true,"origin":"","legend":"","description":"","filename":"2025TP002977.json","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/e33d73d8f8c919117d58d518.json"},{"id":97320055,"identity":"921fdec9-cd02-482e-bc45-696d8052895b","added_by":"auto","created_at":"2025-12-03 07:39:01","extension":"xml","order_by":2,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":156684,"visible":true,"origin":"","legend":"","description":"","filename":"2025TP0029770enriched.xml","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/207cc4813251b9f7bd8d810e.xml"},{"id":97320061,"identity":"db18d422-554e-4b64-9a1e-6beb147fc59d","added_by":"auto","created_at":"2025-12-03 07:39:01","extension":"png","order_by":7,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":345383,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/368810387e7b6cd7eb8bc9fd.png"},{"id":97320066,"identity":"5f0e2031-106f-4969-a155-a2d0e8304abb","added_by":"auto","created_at":"2025-12-03 07:39:03","extension":"png","order_by":8,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":347703,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/8f4120a7b03c4f39afccb61c.png"},{"id":97320031,"identity":"f71c8ceb-9ea9-4216-acc1-7ddea04cd730","added_by":"auto","created_at":"2025-12-03 07:39:00","extension":"png","order_by":9,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":463962,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/85335527048c9c06d955a029.png"},{"id":97320070,"identity":"f9743c74-9dec-4571-a406-2a79e40d3a13","added_by":"auto","created_at":"2025-12-03 07:39:03","extension":"png","order_by":10,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":442719,"visible":true,"origin":"","legend":"","description":"","filename":"Onlinefloatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/2b113b9601bb7f8df340dedd.png"},{"id":97320067,"identity":"c51dad6e-7fa2-4006-9d28-9cd83553304e","added_by":"auto","created_at":"2025-12-03 07:39:03","extension":"xml","order_by":11,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":152114,"visible":true,"origin":"","legend":"","description":"","filename":"2025TP0029770structuring.xml","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/0734c2231f311511bcb037cf.xml"},{"id":97320068,"identity":"f218ca26-5481-4b7a-a56f-46fa43b586be","added_by":"auto","created_at":"2025-12-03 07:39:03","extension":"html","order_by":12,"title":"","display":"","copyAsset":false,"role":"acdc-reference","size":171441,"visible":true,"origin":"","legend":"","description":"","filename":"earlyproof.html","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/3f7aace92ac10645126559b8.html"},{"id":97320024,"identity":"5f2127e4-60c1-4ade-9783-d2170a93803a","added_by":"auto","created_at":"2025-12-03 07:39:00","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":5469874,"visible":true,"origin":"","legend":"\u003cp\u003eSelective Network Reweighting framework and analytic pipeline. \u003cstrong\u003e(A)\u003c/strong\u003e Overview of the multimodal workflow. Resting-state fMRI from 199 adolescents with mood disorders and item-level HAMD-17 scores are converted into individual functional connectivity matrices based on the AAL90 atlas and Gaussian graphical-model symptom networks. Multi-threshold graph measures are extracted at global, modular and nodal scales and then integrated in a brain–symptom coupling analysis. \u003cstrong\u003e(B)\u003c/strong\u003e Conceptual illustration of module control of network, with color-coded functional modules, control nodes and regulatory inputs that reshape inter-modular communication and routing priorities. \u003cstrong\u003e(C)\u003c/strong\u003eCompeting theoretical models of problematic smartphone use (PSU) in mood-disordered adolescents. The global disruption model predicts a diffuse breakdown of small-world topology, whereas the selective reweighting model posits a preserved global architecture with targeted redistribution of connectivity, characterized by a hyper-cohesive reward–limbic core, constrained executive broadcast and altered bridge nodes between modules.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/4d53775cf0b6c880bae298ca.png"},{"id":97320036,"identity":"82abb15e-e978-4b9e-8d76-34bd5f2685c7","added_by":"auto","created_at":"2025-12-03 07:39:00","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":666514,"visible":true,"origin":"","legend":"\u003cp\u003eGlobal AUC distributions for six graph-theoretical metrics. Violin plots display subject-wise area-under-the-curve (AUC) values across proportional densities for non-PSU (g0, blue) and PSU (g1, orange) adolescents on six global measures: clustering coefficient (C), global efficiency (E\u003csub\u003eglob\u003c/sub\u003e), local efficiency (E\u003csub\u003eloc\u003c/sub\u003e), modularity (Q), assortativity, and total edge count. The substantial overlap of the distributions and uniformly non-significant q values indicate no detectable PSU-related alterations in global small-world or modular network topology.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/e1fdc5c25bbf6dcfd532b942.png"},{"id":97320065,"identity":"88cc98d5-f1fb-4304-85dd-62134cb23570","added_by":"auto","created_at":"2025-12-03 07:39:03","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":978853,"visible":true,"origin":"","legend":"\u003cp\u003eSelective reconfiguration of modular organization in adolescents with and without problematic smartphone use. \u003cstrong\u003e(A)\u003c/strong\u003eIn the non-PSU group (g0), AAL90 regions of interest (ROIs) are displayed in canonical views and colored according to their assignment to five Louvain modules (M0–M4). \u003cstrong\u003e(B)\u003c/strong\u003e The PSU group (g1) is shown with the same layout and color scheme, illustrating changes in the spatial extent of modules. \u003cstrong\u003e(C)\u003c/strong\u003eParallel-sets plot depicting ROI migrations from g0 to g1; band width is proportional to the number of ROIs moving between modules. The most prominent streams correspond to somatomotor and attention ROIs shifting from M0 into the reward–limbic module M3, and a smaller set of language–semantic ROIs moving from M2 into the executive module M1. \u003cstrong\u003e(D)\u003c/strong\u003e Bar plots of module size (number of ROIs) for each group, showing a contraction of M0 (28 → 18 ROIs) and expansion of M3 (8 → 18 ROIs) in PSU, with modest changes in M1 and M2 and a stable visual module M4 (16 ROIs in both groups). Overall, 76 of 90 ROIs preserve their original community label, consistent with selective rather than global reorganization of community structure.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/b3ba9333281e4edb14c28d72.png"},{"id":97320062,"identity":"48cfe37c-1edd-4479-b4d5-b3fc374b3ab8","added_by":"auto","created_at":"2025-12-03 07:39:02","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":870102,"visible":true,"origin":"","legend":"\u003cp\u003eNodal-scale redistribution of betweenness centrality in adolescents with and without PSU. \u003cstrong\u003e(A)\u003c/strong\u003eTwo brain regions that showed FDR-corrected between-group differences in betweenness (AUC across densities): the left supramarginal gyrus (SMG_L, red) and the right inferior temporal gyrus (Temporal_Inf_R, blue); \u003cstrong\u003e(B)\u003c/strong\u003e Ranked bar plot of Δ betweenness (g1 − g0; AUC) for the 30 regions with the largest absolute effects; \u003cstrong\u003e(C)\u003c/strong\u003e the distribution of subject-level betweenness AUC in SMG_L and Temporal_Inf_R for each group.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/82b2d95958535627c865d084.png"},{"id":102397146,"identity":"1ba87c54-7e10-4e56-9527-bb17446ecf7e","added_by":"auto","created_at":"2026-02-11 10:04:17","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":8549087,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-8256165/v1/fe820943-ad17-4563-a51b-8a75ba87133e.pdf"}],"financialInterests":"The authors have declared there is \u003cb\u003eNO\u003c/b\u003e conflict of interest to disclose","formattedTitle":"Selective Network Reweighting in Adolescents with Mood Disorders and Problematic Smartphone Use","fulltext":[{"header":"Introduction","content":"\u003cp\u003eAdolescence is a window of heightened neuroplasticity when large-scale functional networks consolidate into adult-like segregation and integration\u003csup\u003e\u003cspan additionalcitationids=\"CR2\" citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. In parallel, smartphones have become embedded in adolescents' daily routines and deliver dense, socially punctuated cues such as alerts, vibrations and short videos that repeatedly capture attention\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e,\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. For most young people this engagement remains within normative bounds, but among adolescents with mood disorders problematic smartphone use (PSU), defined by persistent overuse with poor control, tolerance- and withdrawal-like symptoms, and functional impairment, is especially common and clinically salient. Affected youths frequently report compulsive checking, sleep disruption and interference with school and family life\u003csup\u003e\u003cspan additionalcitationids=\"CR7 CR8\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e. Epidemiological syntheses and clinical observations indicate that, within mood-disordered cohorts, higher PSU severity tracks worse sleep and affective symptoms and greater difficulty sustaining attention, consistent with the idea that repeated cue exposure can exacerbate self-regulatory challenges already present in affective illness\u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eFrom a psychological perspective, adolescents with mood disorders often describe using smartphones to escape from rumination, regulate negative affect, alleviate loneliness and maintain fragile social ties\u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e,\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e. The smartphone environment provides rapid, intermittent rewards and constant opportunities for social comparison and reassurance. These contingencies may transiently relieve dysphoria yet reinforce maladaptive coping strategies. Over time, such patterns can crystallize into compulsive use, nocturnal engagement and the prioritization of smartphone activities over offline responsibilities, which aggravate sleep disturbance, interpersonal conflict and functional impairment. In this context, PSU can be conceptualized as a behavioral comorbidity and modifier of adolescent mood disorders, rather than as a stand-alone diagnosis.\u003c/p\u003e\u003cp\u003eConvergent neuroimaging and behavioral findings across technology overuse and behavioral addictions highlight three recurring features. First, salience hubs in the anterior insula and anterior cingulate cortex, together with reward and habit circuits centered on the striatum, show heightened engagement in response to disorder-relevant cues. Second, repeated pairing of digital cues with mood change and social feedback endows such cues with conditioned motivational value that increasingly recruits reward and attention networks. Third, frontoparietal executive control networks exhibit reduced recruitment or weakened coupling with salience and reward systems, suggesting impaired top-down regulation\u003csup\u003e\u003cspan additionalcitationids=\"CR14 CR15\" citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR16\" class=\"CitationRef\"\u003e16\u003c/span\u003e\u003c/sup\u003e. Independently, mood disorders are associated with fronto-limbic dysregulation and altered switching between default mode, salience and executive networks\u003csup\u003e\u003cspan citationid=\"CR17\" class=\"CitationRef\"\u003e17\u003c/span\u003e\u003c/sup\u003e. Taken together, these strands of evidence suggest that, in adolescents who already have mood disorders, PSU may further bias large-scale brain dynamics towards reward- and salience-driven processing at the expense of regulatory control, rather than representing an entirely separate neural phenotype\u003csup\u003e\u003cspan additionalcitationids=\"CR19 CR20\" citationid=\"CR18\" class=\"CitationRef\"\u003e18\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR21\" class=\"CitationRef\"\u003e21\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eNetwork neuroscience offers a principled framework for formalizing these ideas by modelling the brain as a graph of nodes and edges that exhibits small-world architecture, modular organization and hub-mediated communication\u003csup\u003e\u003cspan citationid=\"CR22\" class=\"CitationRef\"\u003e22\u003c/span\u003e\u003c/sup\u003e. Across a range of psychiatric conditions, including depression and behavioral addictions, global small-world metrics such as characteristic path length, clustering and efficiency are often preserved or only modestly altered, whereas more pronounced changes appear at modular and nodal scales. Graph-theoretical studies in internet-related addictions, for example, have reported altered cohesion and centrality of reward, salience and executive modules, together with shifts in the distribution of connector hubs, even when global efficiency remains within typical ranges. This pattern raises the possibility that psychopathology may not primarily reflect a diffuse breakdown of network topology but rather selective adjustments in the relative weighting of communication pathways that are embedded within an otherwise intact architecture.\u003c/p\u003e\u003cp\u003eBeyond large-scale brain circuits, mood and anxiety symptoms themselves can be viewed as interacting elements of a network rather than as isolated manifestations of a single latent disease entity. Symptom-network approaches treat individual questionnaire items as nodes and estimate conditional associations between them, yielding symptom communities that often recapitulate clinically recognizable clusters such as core depressive mood and anhedonia, anxiety and somatic tension, sleep disturbance, and appetite or health-concern symptoms\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e\u003c/sup\u003e. Bridge symptoms that connect these communities are thought to play a pivotal role in maintaining syndromes and mediating transitions between symptom constellations. Symptom-network analyses in depression and anxiety indicate that network structure can differ across clinical subgroups and environmental exposures, suggesting that the organization of symptom communities carries mechanistic and prognostic information. Clinical observations in adolescents with PSU highlight prominent anxiety, somatic complaints and sleep disruption, indicating that these symptom clusters may be particularly relevant for understanding how PSU modifies the course and expression of mood disorders.\u003c/p\u003e\u003cp\u003eDespite these converging lines of evidence, several key questions remain unresolved regarding how PSU is embedded within adolescent mood disorders. First, most neuroimaging studies of technology-related overuse have been conducted in community or mixed clinical samples and have treated depressive and anxiety symptoms as covariates or secondary outcomes rather than as the primary clinical context\u003csup\u003e\u003cspan citationid=\"CR24\" class=\"CitationRef\"\u003e24\u003c/span\u003e\u003c/sup\u003e. Consequently, it remains unclear whether, within diagnosed mood disorders, PSU marks a distinct connectome phenotype or simply indexes a more severe expression of the same underlying circuit dysfunction. Second, prior work has rarely characterized the functional connectome of adolescents with and without PSU using multi-scale graph-theoretical analysis. As a result, we do not know whether PSU is associated with a genuine disruption of global integration and segregation, or instead with selective changes in the cohesion and hub roles of canonical systems such as reward, salience and executive networks. Third, brain networks and depressive symptom networks have almost always been examined in isolation; to our knowledge, no study has tested whether PSU modifies the coupling between large-scale functional organization and the structure of depressive symptom communities in mood-disordered adolescents. Addressing these gaps is essential for clarifying whether PSU acts primarily as a quantitative amplifier of mood-disorder severity or as a qualitative modifier that alters brain\u0026ndash;symptom mappings on an already vulnerable substrate.\u003c/p\u003e\u003cp\u003eTo organize these questions, we introduce a Selective Network Reweighting framework. In this framework, PSU in mood-disordered adolescents is conceptualized not as a collapse of global functional topology, but as a redistribution of routing priorities across large-scale brain systems. The overall small-world architecture of the functional connectome is assumed to remain largely preserved, whereas the relative weight carried by communication channels linking reward, salience, executive and semantic systems is altered. This perspective generates testable predictions about the levels of the network hierarchy\u0026mdash;global, modular and nodal\u0026mdash;at which PSU-related differences should be expressed, and about how such differences should relate to specific constellations of depressive symptoms (Fig.\u0026nbsp;\u003cspan refid=\"Fig1\" class=\"InternalRef\"\u003e1\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eIn the present study, we applied this framework to a clinically well characterized cohort of adolescents with mood disorders who were stratified according to PSU status. Using resting-state functional MRI and graph-theoretical analysis across a range of proportional densities, we quantified macro-, meso- and nodal-scale properties of the functional connectome. In parallel, we constructed depressive symptom networks from item-level Hamilton Depression Rating Scale scores and identified symptom communities reflecting anxiety and somatic tension, core depressive and psychomotor symptoms, sleep disturbance, and appetite or health-related concerns. Guided by the Selective Network Reweighting framework and by prior work in behavioral addictions and mood disorders, we focused on three questions. First, does PSU in mood-disordered adolescents involve a breakdown of global small-world organization, or are global metrics broadly preserved? Second, are PSU-related differences expressed as selective changes in the configuration and bridging roles of canonical networks such as reward, salience, executive and semantic systems? Third, does PSU alter the pattern of coupling between functional network properties and depressive symptom communities, particularly for anxiety\u0026ndash;somatic and sleep-related clusters? By jointly modelling large-scale functional organization and symptom-network structure, this study aims to provide a mechanistic account of how PSU is instantiated in the brains and symptom profiles of adolescents with mood disorders, and to delineate circuit-level targets for early intervention.\u003c/p\u003e"},{"header":"Methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e\u003ch2\u003eParticipants and clinical characterization\u003c/h2\u003e\u003cp\u003eThis study enrolled a cohort of 199 adolescents aged 13 to 18 years diagnosed with mood disorders, comprising 99 patients with Major Depressive Disorder (MDD) and 100 with Bipolar Disorder (BD) presenting in the depressive episode. Diagnostic confirmation was performed by trained psychiatrists using the Schedule for Affective Disorder and Schizophrenia for School Age Children Present and Lifetime Version (K-SADS-PL) in strict accordance with DSM-5 criteria, ensuring the absence of comorbid Axis I conditions.\u003c/p\u003e\u003cp\u003eBased on clinical adjudication, the cohort was stratified into a PSU group (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:n=86\\)\u003c/span\u003e\u003c/span\u003e; mean age \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:15.13\\pm\\:1.45\\)\u003c/span\u003e\u003c/span\u003e years) and a non-PSU comparison group (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:n=113\\)\u003c/span\u003e\u003c/span\u003e; mean age \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:15.32\\pm\\:1.39\\)\u003c/span\u003e\u003c/span\u003e years). PSU status was defined using modified DSM-5 criteria for Internet Gaming Disorder, with \u0026ldquo;smartphone use\u0026rdquo; substituting for \u0026ldquo;gaming\u0026rdquo; as the target behavior. Adolescents were classified as PSU if they endorsed at least five of the following nine maladaptive criteria within the preceding 12 months : (1) preoccupation, where smartphone activity becomes the dominant daily focus or involves anticipation of future use; (2) withdrawal symptoms, manifesting as irritability, anxiety, or sadness when the device is inaccessible; (3) tolerance, characterized by the need for increasing usage duration to achieve satisfaction; (4) loss of control, defined as unsuccessful attempts to curtail participation; (5) displacement of interests, involving the loss of interest in previous hobbies solely due to smartphone use; (6) continued excessive use despite awareness of resulting psychosocial problems; (7) deception of family members or therapists regarding the extent of usage; (8) dysfunctional coping, utilizing the device to escape or relieve negative affective states such as helplessness, guilt, or anxiety ; and (9)functional impairment, specifically jeopardizing or losing significant relationships, educational, or career opportunities.\u003c/p\u003e\u003cp\u003eTo provide a dimensional measure of severity, participants completed the Mandarin version of the 26-item Smartphone Addiction Inventory (SPAI)\u003csup\u003e\u003cspan citationid=\"CR25\" class=\"CitationRef\"\u003e25\u003c/span\u003e\u003c/sup\u003e. This scale evaluates four subdomains: compulsive behavior (9 items), functional impairment (8 items), withdrawal symptoms (6 items), and tolerance (3 items). Items are rated on a 4-point Likert scale ranging from \u0026ldquo;strongly disagree\u0026rdquo; to \u0026ldquo;strongly agree\u0026rdquo;, with the total score indicating addiction severity (Cronbach's \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\alpha\\:=0.94\\)\u003c/span\u003e\u003c/span\u003e in this sample). Depressive symptom severity was assessed using the 17-item Hamilton Depression Rating Scale (HAMD-17).\u003c/p\u003e\u003cp\u003eCandidates were excluded if they presented with: (1) history of major medical illness; (2) history of moderate or severe head trauma, neurological disorders, or intellectual disability; (3) lifetime substance or alcohol dependence; (4) MRI contraindications; (5) suboptimal imaging data quality; or (6) concurrent major physical illness potentially confounding mood symptoms.\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eMRI acquisition and preprocessing\u003c/h3\u003e\n\u003cp\u003eResting-state fMRI data were acquired on a 3 T Siemens Prisma scanner using a gradient-echo EPI sequence (TR\u0026thinsp;=\u0026thinsp;500 ms, TE\u0026thinsp;=\u0026thinsp;30 ms) during an eyes-closed resting state for approximately 8 minutes. Participants were instructed to keep their eyes closed, remain still, and stay awake. Preprocessing was performed in SPM12 and involved discarding initial volumes to allow magnetization stabilization, slice-timing correction, rigid-body realignment to the mean functional image, co-registration of the mean EPI to each participant\u0026rsquo;s T1-weighted anatomical image, tissue segmentation, and normalization to MNI space, followed by 6-mm FWHM Gaussian smoothing\u003csup\u003e\u003cspan citationid=\"CR26\" class=\"CitationRef\"\u003e26\u003c/span\u003e\u003c/sup\u003e. Nuisance regression removed six head-motion parameters as well as mean white-matter and cerebrospinal-fluid signals, after which residual time series were band-pass filtered between 0.01 and 0.10 Hz. Scans were strictly quality-controlled for motion (translation\u0026thinsp;\u0026lt;\u0026thinsp;3 mm and rotation\u0026thinsp;\u0026lt;\u0026thinsp;3\u0026deg;); datasets exceeding these limits were excluded. Regional time series were extracted by averaging preprocessed voxel signals within the 90 regions of interest (ROIs) defined by the AAL atlas.\u003c/p\u003e\n\u003ch3\u003eNetwork construction and graph measures\u003c/h3\u003e\n\u003cp\u003eSubject-level functional graphs were constructed by parcellating the brain using the Automated Anatomical Labeling (AAL90) atlas\u003csup\u003e\u003cspan citationid=\"CR27\" class=\"CitationRef\"\u003e27\u003c/span\u003e\u003c/sup\u003e. For each participant, preprocessed BOLD time series were averaged within the 90 ROIs to generate a Pearson correlation matrix. Correlations were Fisher r-to-z transformed to stabilize variance. We retained only positive weights to avoid ambiguities in signed geodesics for shortest-path and triangle-based metrics\u003csup\u003e\u003cspan citationid=\"CR28\" class=\"CitationRef\"\u003e28\u003c/span\u003e\u003c/sup\u003e. To standardize sparsity and suppress weak or spurious connections, proportional thresholding was applied, ensuring that every subject retained an identical fraction of the strongest edges at each target density. Metrics requiring connectivity, such as characteristic path length, were computed on the largest connected component. For weighted geodesics, edge length was defined as \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\mathcal{l}=1/(w+\\epsilon\\:)\\)\u003c/span\u003e\u003c/span\u003e with \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\epsilon\\:\u0026gt;0\\)\u003c/span\u003e\u003c/span\u003e to ensure monotonicity between coupling strength and distance while maintaining numerical stability.\u003c/p\u003e\u003cp\u003eTo mitigate threshold arbitrariness and emphasize effects persisting across sparsity levels, all graph measures were evaluated across proportional densities ranging from 0.05 to 0.50 in increments of 0.05\u003csup\u003e29\u0026ndash;31\u003c/sup\u003e. These were summarized at the subject level by the area under the curve (AUC), a method that samples both sparse backbones\u0026mdash;where integration relies on a few strong edges\u0026mdash;and denser regimes characterized by local redundancy and clustering\u003csup\u003e\u003cspan citationid=\"CR32\" class=\"CitationRef\"\u003e32\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eGlobal network metrics included the clustering coefficient (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:C\\)\u003c/span\u003e\u003c/span\u003e), global efficiency (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{\\text{glob}}\\)\u003c/span\u003e\u003c/span\u003e), local efficiency (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{\\text{loc}}\\)\u003c/span\u003e\u003c/span\u003e), characteristic path length (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:L\\)\u003c/span\u003e\u003c/span\u003e), assortativity, modularity (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Q\\)\u003c/span\u003e\u003c/span\u003e), and edge count. Nodal metrics comprised strength (sum of weights), degree (edge count at a given density), closeness (inverse of average weighted geodesic distance), betweenness (fraction of all-pairs shortest paths traversing the node), local efficiency (efficiency of the node\u0026rsquo;s neighborhood subgraph), participation coefficient (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:P\\)\u003c/span\u003e\u003c/span\u003e; quantifying the diversity of inter-modular connections), and within-module degree z-score (within-Z; connectivity relative to module peers). Group contrasts followed the sign convention \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}=\\text{PSU}-\\text{non-PSU}\\)\u003c/span\u003e\u003c/span\u003e. For module-level summaries, nodal metrics were aggregated within aligned modules per subject to preserve the directional meaning of \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}\\)\u003c/span\u003e\u003c/span\u003e at the community scale.\u003c/p\u003e\u003cp\u003eCommunity structure was estimated by applying Louvain optimization to the group-average weighted graph separately for each group\u003csup\u003e\u003cspan citationid=\"CR33\" class=\"CitationRef\"\u003e33\u003c/span\u003e\u003c/sup\u003e. To detect genuine between-group differences in modular composition, we did not impose a single consensus partition. Instead, post-hoc alignment was performed by greedily maximizing Jaccard overlap between ROI sets, yielding one-to-one module correspondences while preserving reconfiguration. Alignment quality was summarized using the adjusted Rand index (ARI) and variation of information (VI), with best-match Jaccard values reported for each module.\u003c/p\u003e\u003cp\u003eMethodological safeguards included proportional thresholding to enforce density matching, weighted graph definitions to preserve information lost in binarization, monotonic geodesic definitions (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\mathcal{l}=1/(w+\\epsilon\\:)\\)\u003c/span\u003e\u003c/span\u003e), restriction of path length calculations to the largest connected component to avoid artifacts from isolated nodes, and AUC summarization to down-weight single-threshold idiosyncrasies, thereby improving robustness for adolescent resting-state fMRI analyses.\u003c/p\u003e\n\u003ch3\u003eSymptom measures and symptom-network analysis\u003c/h3\u003e\n\u003cp\u003eHAMD-17 total and item scores were obtained on the same day as MRI scanning by trained psychiatrists blinded to PSU status. Items were treated as ordered categorical variables reflecting severity within domains such as depressed mood, guilt, sleep disturbance, psychomotor change, anxiety, and somatic symptoms.\u003c/p\u003e\u003cp\u003eTo derive symptom modules, group-specific symptom networks were constructed using Gaussian graphical models (GGMs)\u003csup\u003e\u003cspan citationid=\"CR34\" class=\"CitationRef\"\u003e34\u003c/span\u003e\u003c/sup\u003e. For each group, HAMD-17 item scores were median-imputed and transformed using a rank-based inverse-normal (nonparanormal) transform followed by z-scaling. Partial correlation networks were estimated using the Graphical Lasso with five-fold cross-validation to select the regularization parameter \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\lambda\\:\\)\u003c/span\u003e\u003c/span\u003e. Precision matrices were converted to signed partial correlations with diagonals set to zero.\u003c/p\u003e\u003cp\u003eTo obtain a single shared symptom partition for cross-group comparisons, absolute partial correlations were averaged across groups, and greedy modularity maximization (Newman\u0026ndash;Girvan) was applied to the resulting graph. This yielded four symptom communities (modules 0\u0026ndash;3) capturing distinct clusters: (i) anxiety, somatic tension, and negative-self symptoms including guilt, psychomotor agitation, systemic and somatic anxiety, reduced sexual interest, and self-consciousness; (ii) core depressive and psychomotor symptoms including depressed mood, suicidality, work or interest loss, and psychomotor retardation; (iii) sleep disturbance encompassing initial, middle, and early-morning insomnia; and (iv) appetite and health-concern symptoms such as weight loss, hypochondriasis, and gastrointestinal complaints.\u003c/p\u003e\u003cp\u003eOn the group-specific GGMs, we computed node-wise Strength (sum \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\left|w\\right|\\)\u003c/span\u003e\u003c/span\u003e), Expected Influence (sum \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:w\\)\u003c/span\u003e\u003c/span\u003e), participation coefficient, within-Z, bridge strength, and bridge expected influence, alongside conventional centralities including degree, betweenness, closeness, eigenvector centrality, and weighted clustering using NetworkX. Global metrics at the symptom level included density, global strength, mean node strength, average clustering, global efficiency, mean shortest-path length, and modularity evaluated on the shared symptom partition.\u003c/p\u003e\u003cp\u003eTo test for between-group differences in symptom-network architecture, a permutation-based Network Comparison Test (NCT) was implemented. Using a fixed \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\lambda\\:\\)\u003c/span\u003e\u003c/span\u003e (mean of the two cross-validated values), GGMs were re-estimated on PSU and non-PSU data to compute: (i) differences in global strength, (ii) edge-wise differences in partial correlations, and (iii) differences in nodal and global metrics. Group labels were randomly permuted while preserving group sizes, and networks were re-estimated 1,000 times to generate null distributions. Two-sided permutation \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:P\\)\u003c/span\u003e\u003c/span\u003e-values were calculated as the proportion of null differences with absolute magnitudes equal to or greater than the observed difference. Edge-level and node-level \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:P\\)\u003c/span\u003e\u003c/span\u003e-values were controlled for false discovery rate (FDR) using the Benjamini\u0026ndash;Hochberg procedure within families, while global-metric \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:P\\)\u003c/span\u003e\u003c/span\u003e-values were FDR-controlled across metrics.\u003c/p\u003e\u003cdiv id=\"Sec7\" class=\"Section2\"\u003e\u003ch2\u003eStatistical analysis\u003c/h2\u003e\u003cp\u003eAll between-group effects were defined as \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}=\\text{PSU}-\\text{non-PSU}\\)\u003c/span\u003e\u003c/span\u003e. For each subject and graph metric, density curves were summarized by the area under the curve (AUC), providing a single value per metric per subject to mitigate multiple comparisons across thresholds. Global metrics including clustering coefficient (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:C\\)\u003c/span\u003e\u003c/span\u003e), global efficiency (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{\\text{glob}}\\)\u003c/span\u003e\u003c/span\u003e), local efficiency (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{E}_{\\text{loc}}\\)\u003c/span\u003e\u003c/span\u003e), characteristic path length (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:L\\)\u003c/span\u003e\u003c/span\u003e), modularity (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Q\\)\u003c/span\u003e\u003c/span\u003e), assortativity, and edge count were compared between groups using two-sided Welch\u0026rsquo;s \u003cspan\u003e$\u003c/span\u003et\u003cspan\u003e$\u003c/span\u003e-tests with Satterthwaite degrees of freedom, ensuring robustness to unequal variances and sample sizes. Node-wise analyses treated each metric as a family of 90 ROI tests; within each metric, Welch\u0026rsquo;s \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:t\\)\u003c/span\u003e\u003c/span\u003e-tests were computed at every node, and Benjamini\u0026ndash;Hochberg FDR correction was applied across the 90 tests, with \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q\u0026lt;0.05\\)\u003c/span\u003e\u003c/span\u003e denoting statistical significance. For all contrasts, we reported the test statistic (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:t\\)\u003c/span\u003e\u003c/span\u003e), the FDR-adjusted \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q\\)\u003c/span\u003e\u003c/span\u003e value, and Hedges\u0026rsquo;\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:g\\)\u003c/span\u003e\u003c/span\u003e(bias-corrected standardized mean difference) signed by \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}\\)\u003c/span\u003e\u003c/span\u003e to retain directionality (positive values indicate \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\text{PSU}\u0026gt;\\text{non-PSU}\\)\u003c/span\u003e\u003c/span\u003e). Module-level summaries were derived by aggregating nodal metrics within post-hoc aligned modules for each subject and contrasting groups with Welch\u0026rsquo;s tests. Given the limited number of modules and dependency among constituent nodes, interpretation focused on effect directions and convergence with nodal and community findings rather than strict family-wise inference at this level.\u003c/p\u003e\u003cp\u003eSecondary analyses examined the relationship between network-level brain metrics and symptom expression. HAMD-17 items were first summarized into subject-level symptom-cluster scores based on the symptom partition: item scores were z-standardized across the full sample and averaged within each module, yielding four continuous outcomes (module0_mean to module3_mean) per subject. A subject-level brain-metric dataset was assembled comprising (i) node-wise AUC betweenness for the left supramarginal gyrus (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:SM{G}_{L}\\)\u003c/span\u003e\u003c/span\u003e) and right inferior temporal gyrus (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Tempora{l}_{In{f}_{R}}\\)\u003c/span\u003e\u003c/span\u003e), and (ii) module-level AUC metrics obtained by averaging nodal values within aligned modules, specifically focusing on executive (M1) and reward\u0026ndash;limbic (M3) communities.\u003c/p\u003e\u003cp\u003eFor each symptom module, an ordinary least-squares (OLS) regression model was fitted:\u003cdiv id=\"Equa\" class=\"Equation\"\u003e\u003cdiv format=\"TEX\" class=\"mathdisplay\" id=\"FileID_Equa\" name=\"EquationSource\"\u003e\n$$\\:\\text{moduleX\\mean}\\sim\\:\\text{group}+\\text{brain\\metrics}+(\\text{brain_metrics}\\times\\:\\text{group})$$\u003c/div\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003ewhere group coded PSU status (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:0=\\text{non-PSU}\\)\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:1=\\text{PSU}\\)\u003c/span\u003e\u003c/span\u003e). All models included an intercept and were estimated with HC3 robust standard errors. \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:P\\)\u003c/span\u003e\u003c/span\u003e-values for all terms within each outcome were corrected using the Benjamini\u0026ndash;Hochberg FDR method (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q\u0026lt;0.05\\)\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003cp\u003eComplementary within-group relationships were examined using Spearman rank correlations between each brain metric and symptom-module score separately for PSU and non-PSU groups, with BH-FDR correction applied within each group. Finally, multivariate coupling between the full set of brain metrics and the four symptom clusters was characterized using partial least squares (PLS) regression with 2 components and 5-fold cross-validation. The cross-validated \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{R}^{2}\\)\u003c/span\u003e\u003c/span\u003efor each symptom module was compared against a null distribution generated by permuting subject labels 1,000 times; two-sided permutation \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:p\\)\u003c/span\u003e\u003c/span\u003e-values were computed as the proportion of null \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{R}^{2}\\)\u003c/span\u003e\u003c/span\u003e values exceeding the observed magnitude.\u003c/p\u003e\u003c/div\u003e"},{"header":"Results","content":"\u003cdiv id=\"Sec9\" class=\"Section2\"\u003e\u003ch2\u003eGlobal topology: small-world architecture preserved\u003c/h2\u003e\u003cp\u003eAcross proportional densities (AUC 0.05\u0026ndash;0.50), global topological metrics showed no evidence of group differences. No metrics survived Benjamini\u0026ndash;Hochberg FDR correction (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q\u0026gt;0.05\\)\u003c/span\u003e\u003c/span\u003e). For example, the between-group difference in modularity was negligible (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}=0.0018\\)\u003c/span\u003e\u003c/span\u003e, Hedges' \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:g=0.13\\)\u003c/span\u003e\u003c/span\u003e, 95% CI\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:[-\\text{0.15,0.41}]\\)\u003c/span\u003e\u003c/span\u003e), and global efficiency was similarly stable (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}=-0.0005\\)\u003c/span\u003e\u003c/span\u003e, Hedges' \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:g=-0.11\\)\u003c/span\u003e\u003c/span\u003e, 95% CI \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:[-\\text{0.39,0.17}]\\)\u003c/span\u003e\u003c/span\u003e). Taken together, these findings indicate no evidence for disruption of macro-scale small-world organization in adolescents with PSU (Fig.\u0026nbsp;\u003cspan refid=\"Fig2\" class=\"InternalRef\"\u003e2\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003c/div\u003e\n\u003ch3\u003eModule composition overview\u003c/h3\u003e\n\u003cp\u003eWe next examined the modular organization of the functional networks. Post-hoc alignment revealed that 76 out of 90 ROIs retained their modular assignment across groups (Adjusted Rand Index \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:=0.70\\)\u003c/span\u003e\u003c/span\u003e; Variation of Information \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:=0.82\\)\u003c/span\u003e\u003c/span\u003e). Changes in modular assignment were observed in specific systems. Ten ROIs, including the caudate, putamen, pallidum, thalamus, and medial cingulate, shifted from the somatomotor and attention module (M0) to the reward and limbic module (M3). Additionally, four ROIs moved from the language and semantic module (M2) to the executive module (M1). These shifts corresponded to changes in module size, with M0 decreasing from 28 to 18 regions, M3 increasing from 8 to 18 regions, M2 decreasing from 22 to 18 regions, and M1 increasing from 16 to 20 regions. (Fig.\u0026nbsp;\u003cspan refid=\"Fig3\" class=\"InternalRef\"\u003e3\u003c/span\u003e)\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eAt the aggregated module level, the reward and limbic community (M3) demonstrated increased strength, degree, local efficiency, and within-module degree, combined with decreased participation coefficient and betweenness centrality in the PSU group. This pattern suggests a configuration characterized by high internal connectivity and low external integration. Conversely, the executive module (M1) and the language and semantic module (M2) showed reductions in degree, strength, and closeness centrality. The participation coefficient was also lower in M1. The somatomotor (M0) and visual (M4) modules showed minimal differences between groups. (Table\u0026nbsp;\u003cspan refid=\"Tab1\" class=\"InternalRef\"\u003e1\u003c/span\u003e)\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab1\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 1\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eModule-level contrasts (AUC; Δ=PSU-non-PSU).\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colspan=\"6\" nameend=\"c6\" namest=\"c1\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}=\\text{PSU}-\\text{non-PSU}\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eMetric\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eM0\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eM1\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003eM2\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003eM3\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003eM4\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eStrength\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-0.0280\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-0.1420\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-0.1940\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.3470\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.0410\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eDegree\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-0.0008\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-0.0028\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-0.0035\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.0031\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.0028\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eCloseness\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.0006\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-0.0012\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-0.0011\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.0007\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.0006\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eLocal efficiency\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-0.0027\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-0.0022\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-0.0035\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.0072\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e-0.0002\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eParticipation coefficient \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:P\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.0071\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e-0.0040\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.0076\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e-0.0136\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e0.0074\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eWithin-Z\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e-0.0187\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.0308\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e-0.0339\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e0.0635\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e-0.0283\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eBetweenness\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003e0.0008\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003e0.0001\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c4\"\u003e\u003cp\u003e0.0001\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c5\"\u003e\u003cp\u003e-0.0007\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c6\"\u003e\u003cp\u003e-0.0002\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cdiv id=\"Sec11\" class=\"Section2\"\u003e\u003ch2\u003eNodal scale: redistributed bridges\u003c/h2\u003e\u003cp\u003eAnalysis of nodal centrality revealed two regions that survived FDR correction for betweenness centrality. Compared to the non-PSU group, the PSU group exhibited significantly higher betweenness in the left supramarginal gyrus (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:SM{G}_{L}\\)\u003c/span\u003e\u003c/span\u003e; \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:t=3.49\\)\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q=0.036\\)\u003c/span\u003e\u003c/span\u003e, Hedges'\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:g=0.51\\)\u003c/span\u003e\u003c/span\u003e, 95% CI \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\left[\\text{0.23,0.80}\\right]\\)\u003c/span\u003e\u003c/span\u003e) and significantly lower betweenness in the right inferior temporal gyrus (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:Tempora{l}_{In{f}_{R}}\\)\u003c/span\u003e\u003c/span\u003e; \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:t=-3.41\\)\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q=0.036\\)\u003c/span\u003e\u003c/span\u003e, Hedges'\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:g=-0.45\\)\u003c/span\u003e\u003c/span\u003e, 95% CI \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:[-0.73,-0.16]\\)\u003c/span\u003e\u003c/span\u003e). (Fig.\u0026nbsp;\u003cspan refid=\"Fig4\" class=\"InternalRef\"\u003e4\u003c/span\u003e)\u003c/p\u003e\u003cp\u003e\u003c/p\u003e\u003cp\u003eBeyond these specific nodes, other centrality metrics (degree, strength, closeness) showed small, dispersed effects across the brain (mean \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\left|g\\right|=0.15\\)\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:0.15\\)\u003c/span\u003e\u003c/span\u003e and \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:0.14\\)\u003c/span\u003e\u003c/span\u003e, respectively), with no other regions surviving multiple comparison correction.\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec12\" class=\"Section2\"\u003e\u003ch2\u003eSymptom-network organization\u003c/h2\u003e\u003cp\u003eGaussian graphical models estimated separately for PSU and non-PSU groups yielded four stable symptom communities: (i) an anxiety, somatic tension, and negative-self cluster (Module 0); (ii) a core depressive and psychomotor cluster (Module 1); (iii) a sleep disturbance cluster (Module 2); and (iv) an appetite and health-concern cluster (Module 3).\u003c/p\u003e\u003cp\u003eGlobal network properties were comparable between groups. No significant differences were found in density, global strength, mean node strength, average clustering, global efficiency, mean shortest-path length, or modularity after permutation testing and FDR control (all \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q\u0026gt;0.05\\)\u003c/span\u003e\u003c/span\u003e). Similarly, permutation-based Network Comparison Tests revealed no robust between-group differences in edge-wise partial correlations after correction. At the nodal level, a single significant difference was identified: the participation coefficient of the systemic somatic symptoms item (HAMD-17 systemic symptoms, Module 0) was lower in the PSU group compared to the non-PSU group (non-PSU mean\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:=0.73\\)\u003c/span\u003e\u003c/span\u003e PSU mean\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:=0.23\\)\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\Delta\\:}=-0.50\\)\u003c/span\u003e\u003c/span\u003e, permutation \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:p=0.002\\)\u003c/span\u003e\u003c/span\u003e, FDR \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q=0.034\\)\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003c/div\u003e\u003cdiv id=\"Sec13\" class=\"Section2\"\u003e\u003ch2\u003eBrain\u0026ndash;symptom coupling\u003c/h2\u003e\u003cp\u003eWe further examined the relationship between network-level brain metrics and symptom expression using the four symptom communities identified above.\u003c/p\u003e\u003cp\u003eOLS regression models identified a significant interaction between group status and executive module cohesion (M1 within-Z) predicting the severity of the anxiety, somatic tension, and negative-self cluster (Module 0) (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\beta\\:}_{\\text{interaction}}=2.58\\)\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:p=0.0032\\)\u003c/span\u003e\u003c/span\u003e, FDR \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q=0.044\\)\u003c/span\u003e\u003c/span\u003e; Table\u0026nbsp;\u003cspan refid=\"Tab2\" class=\"InternalRef\"\u003e2\u003c/span\u003e). Specifically, in the PSU group, greater executive network cohesion was associated with more severe Module 0 symptoms (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\beta\\:}_{\\text{PSU}}=1.34\\)\u003c/span\u003e\u003c/span\u003e). In contrast, the non-PSU group showed a trend toward a negative association (\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:{\\beta\\:}_{\\text{non-PSU}}=-1.24\\)\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:p=0.013\\)\u003c/span\u003e\u003c/span\u003e, FDR \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q=0.092\\)\u003c/span\u003e\u003c/span\u003e).\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab2\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 2\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eWithin-group Spearman correlations between brain metrics and symptom scores\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"6\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c6\" colnum=\"6\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eGroup\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eBrain metric\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003eSymptom score\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003eSpearman \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\rho\\:\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003ep\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c6\"\u003e\u003cp\u003eFDR-q\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003ePSU\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eM1 within-Z\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eModule 0 Mean\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.3380\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.0015\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.0350\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003enon-PSU\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003eM1 within-Z\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c3\"\u003e\u003cp\u003eModule 0 Mean\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e-0.1334\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.1595\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c6\"\u003e\u003cp\u003e0.6855\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003cp\u003eWithin-group Spearman correlations confirmed this pattern. In the PSU group, M1 within-Z was positively correlated with Module 0 scores (Spearman \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\rho\\:=0.338\\)\u003c/span\u003e\u003c/span\u003e, \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:p=0.0015\\)\u003c/span\u003e\u003c/span\u003e, FDR \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q=0.035\\)\u003c/span\u003e\u003c/span\u003e). No significant correlation was observed in the non-PSU group (FDR \u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q\u0026gt;0.10\\)\u003c/span\u003e\u003c/span\u003e; Table\u0026nbsp;\u003cspan refid=\"Tab3\" class=\"InternalRef\"\u003e3\u003c/span\u003e). No other brain\u0026ndash;symptom associations survived FDR correction.\u003c/p\u003e\u003cp\u003e\u003cdiv class=\"gridtable\"\u003e\u003ctable float=\"Yes\" id=\"Tab3\" border=\"1\"\u003e\u003ccaption language=\"En\"\u003e\u003cdiv class=\"CaptionNumber\"\u003eTable 3\u003c/div\u003e\u003cdiv class=\"CaptionContent\"\u003e\u003cp\u003eBrain\u0026ndash;symptom OLS regression results\u003c/p\u003e\u003c/div\u003e\u003c/caption\u003e\u003ccolgroup cols=\"5\"\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c1\" colnum=\"1\"\u003e\u003c/div\u003e\u003cdiv align=\"left\" class=\"colspec\" colname=\"c2\" colnum=\"2\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c3\" colnum=\"3\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c4\" colnum=\"4\"\u003e\u003c/div\u003e\u003cdiv align=\"char\" char=\".\" class=\"colspec\" colname=\"c5\" colnum=\"5\"\u003e\u003c/div\u003e\u003cthead\u003e\u003ctr\u003e\u003cth align=\"left\" colname=\"c1\"\u003e\u003cp\u003eSymptom Module\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c2\"\u003e\u003cp\u003eGroup\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c3\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:\\beta\\:\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c4\"\u003e\u003cp\u003e\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:p\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/th\u003e\u003cth align=\"left\" colname=\"c5\"\u003e\u003cp\u003eFDR-\u003cspan class=\"InlineEquation\"\u003e\u003cspan class=\"mathinline\"\u003e\\(\\:q\\)\u003c/span\u003e\u003c/span\u003e\u003c/p\u003e\u003c/th\u003e\u003c/tr\u003e\u003c/thead\u003e\u003ctbody\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eModule 0 Mean\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003ePSU\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e2.5770\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.0032\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.0442\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003ctr\u003e\u003ctd align=\"left\" colname=\"c1\"\u003e\u003cp\u003eModule 0 Mean\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"left\" colname=\"c2\"\u003e\u003cp\u003enon-PSU\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c3\"\u003e\u003cp\u003e-1.2395\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c4\"\u003e\u003cp\u003e0.0131\u003c/p\u003e\u003c/td\u003e\u003ctd align=\"char\" char=\".\" colname=\"c5\"\u003e\u003cp\u003e0.0919\u003c/p\u003e\u003c/td\u003e\u003c/tr\u003e\u003c/tbody\u003e\u003c/colgroup\u003e\u003c/table\u003e\u003c/div\u003e\u003c/p\u003e\u003c/div\u003e"},{"header":"Discussion","content":"\u003cp\u003eIn this study, we examined resting-state functional network organization and depressive symptom structure in adolescents with mood disorders, stratified by the presence of problematic smartphone use (PSU), within a Selective Network Reweighting framework. Three main findings emerged. First, at the macro-scale, small-world topology did not differ meaningfully between PSU and non-PSU groups: global efficiency, path length, clustering and modularity showed no evidence of disruption. Second, at the meso- and nodal scales, there was a selective reconfiguration of network weights: somatomotor and attention-related nodes migrated into the reward\u0026ndash;limbic module, which exhibited higher internal connectivity and lower cross-module integration, whereas executive and language\u0026ndash;semantic modules showed reduced integrative roles and a redistribution of bridge centrality from ventral temporal to parietal attention regions. Third, at the level of symptom networks and brain\u0026ndash;symptom coupling, the overall architecture of depressive symptom communities was highly similar between groups, but only in the PSU group was greater executive-network cohesion associated with more severe anxiety\u0026ndash;somatic\u0026ndash;negative self-symptoms. Taken together, these observations are consistent with a model in which PSU in mood-disordered adolescents is instantiated not as global connectome breakdown, but as selective reweighting of specific systems and a modification of the mapping between brain networks and symptom clusters.\u003c/p\u003e\u003cp\u003eAt the macro-scale, the absence of detectable group differences in small-world metrics suggests that PSU does not manifest as a generalized disintegration of functional topology in this clinical context\u003csup\u003e\u003cspan citationid=\"CR19\" class=\"CitationRef\"\u003e19\u003c/span\u003e\u003c/sup\u003e. Global efficiency, clustering and modularity were all comparable between PSU and non-PSU adolescents, with small effect sizes and confidence intervals spanning zero. This pattern contrasts with reports of widespread topological disruption in neurodegenerative disorders and severe psychotic illness, where global efficiency is often reduced and characteristic path length increased\u003csup\u003e\u003cspan additionalcitationids=\"CR36\" citationid=\"CR35\" class=\"CitationRef\"\u003e35\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR37\" class=\"CitationRef\"\u003e37\u003c/span\u003e\u003c/sup\u003e. In adolescents with mood disorders, our findings indicate that the basic capacity of the functional connectome to support integrated and segregated processing appears largely intact irrespective of PSU status. Behavioral manifestations such as impulsive checking and sustained distractibility therefore cannot be straightforwardly attributed to a loss of global network capacity. Rather, they are more plausibly linked to changes in how specific subcircuits are weighted and coupled on top of a preserved macro-scale architecture. Conceptualizing PSU as a disturbance of routing priorities, rather than as a deficit in structural capability, motivates a focus on meso- and nodal-level reorganization.\u003c/p\u003e\u003cp\u003eAgainst this background of global stability, we observed systematic reweighting at the module level. Post hoc alignment showed that most nodes retained their community membership across groups, but a subset of somatomotor and attention-related regions\u0026mdash;including basal ganglia and medial cingulate structures\u0026mdash;shifted from a somatomotor\u0026ndash;attention module into a reward\u0026ndash;limbic module, which increased in size. A smaller set of language\u0026ndash;semantic regions moved into the executive module. In the PSU group, the reward\u0026ndash;limbic community displayed higher degree, strength, local efficiency and within-module degree, together with lower participation coefficient and betweenness centrality, yielding a profile of stronger internal cohesion but weaker cross-module integration. In contrast, executive and language\u0026ndash;semantic modules showed reductions in degree, strength and closeness, and the participation coefficient of the executive module was lower. This pattern echoes prior graph-theoretical findings in behavioral addictions and internet-related overuse, where reward and salience networks often show increased internal cohesion, while frontoparietal control systems lose some of their meditation role between communities. Our results extend these observations to adolescents with mood disorders, suggesting that in those who meet PSU criteria, smartphone-related cue processing, action routines and affective responses are more tightly embedded within a reward\u0026ndash;limbic circuit, whereas executive and semantic systems carry less weight in cross-system integration\u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e,\u003cspan citationid=\"CR38\" class=\"CitationRef\"\u003e38\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eNodal-level results complement this modular reweighting. After correction for multiple comparisons, only two regions showed robust differences in betweenness centrality: the left supramarginal gyrus exhibited higher betweenness in the PSU group, while the right inferior temporal gyrus showed lower betweenness. The supramarginal gyrus forms part of the ventral attention network and is implicated in reorienting towards salient external stimuli; increased betweenness suggests that it lies on a greater proportion of shortest paths and mediates more inter-modular communication\u003csup\u003e\u003cspan citationid=\"CR39\" class=\"CitationRef\"\u003e39\u003c/span\u003e\u003c/sup\u003e. This topology is compatible with clinical observations that adolescents with PSU are highly responsive to notifications and other salient smartphone cues. By contrast, the inferior temporal gyrus contributes to high-level visual\u0026ndash;semantic integration. Reduced betweenness may indicate a diminished bridging role between visual\u0026ndash;semantic processing and other systems, implying that the route from perception to deeper semantic appraisal carries less weight in the overall shortest-path structure\u003csup\u003e\u003cspan citationid=\"CR40\" class=\"CitationRef\"\u003e40\u003c/span\u003e\u003c/sup\u003e. These interpretations remain necessarily indirect, as they are inferred from resting-state connectivity rather than task-evoked information flow, but they illustrate how a \u0026ldquo;parietal attention\u0026ndash;dominant, temporal semantic\u0026ndash;recessive\u0026rdquo; routing configuration could bias behavior toward rapid, cue-driven responses with less engagement of elaborative appraisal\u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eAt the level of symptom networks, we found that both PSU and non-PSU groups exhibited a similar community structure when Hamilton Depression Rating Scale items were modelled as Gaussian graphical models. In each group, four communities emerged, broadly corresponding to anxiety\u0026ndash;somatic tension\u0026ndash;negative self, core depressive and psychomotor, sleep disturbance, and appetite/health-concern clusters. Global indices such as density, global strength, average clustering, global efficiency, mean shortest path length and modularity did not differ significantly between groups, and edge-wise network comparison tests did not reveal robust differences after correction. Thus, within this sample of mood-disordered adolescents, PSU did not appear to reshape the overall organization of depressive symptom co-occurrence. The only node-level difference that survived correction was a lower participation coefficient for the systemic somatic symptoms item in the PSU group, indicating that this symptom was more confined to its own community and played a weaker bridging role across communities. One possible interpretation is that, among youths with PSU, bodily complaints are more tightly embedded within anxiety and negative self-evaluation, and less connected to other symptom domains, aligning with clinical impressions of recurrent health worries and bodily focus that are closely tied to stress and self-criticism. Given that this effect was confined to a single item, it should be interpreted cautiously and replicated in independent samples\u003csup\u003e\u003cspan citationid=\"CR23\" class=\"CitationRef\"\u003e23\u003c/span\u003e,\u003cspan citationid=\"CR41\" class=\"CitationRef\"\u003e41\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e\u003cp\u003eThe most distinctive evidence for a PSU-related modification emerged when we examined brain\u0026ndash;symptom coupling. Across all tested brain metrics and symptom modules, only one interaction survived FDR correction: a group-by-executive-module cohesion effect predicting the anxiety\u0026ndash;somatic\u0026ndash;negative self-symptom cluster. In the PSU group, greater executive-module within-Z was associated with more severe symptoms in this community, whereas in the non-PSU group the association was negative but did not survive correction. Within-group rank correlations showed a broadly similar pattern: executive cohesion correlated positively with anxiety\u0026ndash;somatic symptoms in the PSU group, but no significant correlation was observed in the non-PSU group after correction. These findings suggest that the clinical meaning of executive-network cohesion may depend on behavioral context\u003csup\u003e\u003cspan citationid=\"CR7\" class=\"CitationRef\"\u003e7\u003c/span\u003e\u003c/sup\u003e. In adolescents without PSU, a more cohesive executive network might reflect an effective regulatory resource that helps to contain anxiety and somatic tension. In contrast, in the context of PSU, greater executive cohesion may index a state of high effort but low efficiency, in which control resources are heavily engaged in monitoring, worry and attempts at suppression that do not translate into effective down-regulation of anxiety or bodily arousal. Because our design is cross-sectional, it is not possible to determine whether this pattern reflects a compensatory response to high symptom burden, a pre-existing vulnerability that predisposes to PSU, or a consequence of prolonged PSU-related engagement. Nevertheless, the presence of a PSU-specific interaction supports the idea that PSU can act as a modifier of brain\u0026ndash;symptom mappings, rather than as a simple proxy for depression severity.\u003c/p\u003e\u003cp\u003eFrom a theoretical perspective, a key contribution of this work is to position PSU within adolescent mood disorders as a behavioral modifier, and to provide convergent evidence for a Selective Network Reweighting account at both brain and symptom levels. We observed that, on a preserved small-world architecture, PSU was associated with a reallocation of connectivity weights among reward\u0026ndash;limbic, attention, executive and semantic systems, altering cross-module routes and bridge nodes, while leaving global topology largely unchanged. At the same time, executive-network metrics showed different relationships to anxiety\u0026ndash;somatic symptom communities in adolescents with and without PSU. This pattern\u0026mdash;reweighting of modules and nodes combined with a PSU-dependent modulation of brain\u0026ndash;symptom coupling\u0026mdash;is consistent with the notion that PSU \u0026ldquo;sculpts\u0026rdquo; new behavioral and symptomatic phenotypes on top of an existing mood-disorder architecture\u003csup\u003e\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. More broadly, the Selective Network Reweighting framework may offer a transferable lens for understanding other digital or behavioral comorbidities, such as gaming disorder or excessive social media use, within psychiatric populations.\u003c/p\u003e\u003cp\u003eThe findings also carry potential implications for clinical and translational work. In terms of risk stratification, the combination of a more cohesive but externally less integrated reward\u0026ndash;limbic module, reduced cross-module roles of executive and semantic communities, and elevated centrality of ventral attention nodes could define a candidate network profile for high PSU risk within mood-disordered adolescents, to be tested in longitudinal designs. In terms of intervention, the framework suggests that simply reducing total screen time may be insufficient. Instead, behavioral and psychotherapeutic strategies might aim to rebalance routing priorities by diminishing the salience and contingency of smartphone cues, strengthening flexible integration of executive and semantic systems with other modules, and reducing the closedness of reward\u0026ndash;limbic loops\u003csup\u003e\u003cspan additionalcitationids=\"CR43\" citationid=\"CR42\" class=\"CitationRef\"\u003e42\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR44\" class=\"CitationRef\"\u003e44\u003c/span\u003e\u003c/sup\u003e. Environmental modifications such as managing notification frequency and timing, structuring phone-free periods, or altering interface features to reduce cue salience may alleviate the burden on attention\u0026ndash;reward pathways. Cognitive and emotion-regulation interventions, including training that targets shifting, inhibition and reappraisal, or approaches that enhance interoceptive and metacognitive awareness, may help restore the executive module\u0026rsquo;s role as a flexible integrator rather than an internally locked, over-engaged system.\u003c/p\u003e\u003cp\u003eSeveral limitations should be considered when interpreting these findings. First, the study is cross-sectional, which precludes causal inference: we cannot determine whether the observed network reweighting precedes PSU and contributes to its development, emerges as a consequence of sustained PSU, or reflects a bidirectional process. Longitudinal and interventional studies will be essential to clarify temporal ordering. Second, the sample was drawn from a single clinical setting and included adolescents with both major depressive and bipolar disorders; although key demographic and clinical covariates were controlled statistically, medication use, illness duration and other factors may still confound the results, and generalizability to other populations remains uncertain. Third, network construction relied on a single anatomical parcellation (AAL90) and resting-state functional connectivity; spatial resolution and sensitivity to task-specific processes are limited. Future work using finer-grained parcellations, dynamic connectivity, task-based fMRI and multimodal imaging will be important to test the robustness and specificity of the selective reweighting patterns. Fourth, PSU was assessed using DSM-5\u0026ndash;adapted criteria and self-report measures without objective smartphone usage logs; different types of smartphone activities may be associated with distinct network profiles. Finally, our symptom networks were based solely on HAMD items and did not explicitly include domains such as impulsivity, reward sensitivity or social anxiety, which are likely to be relevant for PSU.\u003c/p\u003e\u003cp\u003eIn summary, by comparing PSU and non-PSU adolescents within a mood-disordered cohort, and by integrating multi-scale graph analysis with symptom-network modelling, this study suggests that PSU is associated with selective network reweighting and altered brain\u0026ndash;symptom coupling rather than with wholesale breakdown of functional architecture. Reward\u0026ndash;limbic and attention systems become more tightly coupled, executive and semantic systems lose some of their cross-module integrative roles, and the association between executive-network cohesion and anxiety\u0026ndash;somatic symptoms is reparametrized in the presence of PSU. Viewing PSU through the lens of routing priorities and selective network reweighting provides a mechanistic framework for understanding its role in adolescent mood disorders and offers a basis for developing network-informed stratification and intervention strategies.\u003c/p\u003e"},{"header":"Declarations","content":"\u003ch2\u003e\u003cstrong\u003eAcknowledgments\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eThis work was supported by the National Natural Science Foundation of China [62176129], National Natural Science Foundation of China-Jiangsu Joint Fund [U24A20701], the Hong Kong RGC Strategic Target Grant [grant number STG1/M-501/23-N]\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003eEthics\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eEthical approval for this study was granted by the Ethics Committee of the Affiliated Nanjing Brain Hospital, Nanjing Medical University (Nanjing, China). Written informed consent was obtained from all participants and their parents or legal guardians following a full explanation of the study procedures.\u003c/p\u003e\n\u003ch2\u003e\u003cstrong\u003eConflict of interest\u003c/strong\u003e\u003c/h2\u003e\n\u003cp\u003eThe authors declare no conflict of interest to this work.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\u003cli\u003e\u003cspan\u003eHan, L., Chan, M. Y., Agres, P. F., Winter-Nelson, E., Zhang, Z., \u0026amp; Wig, G. S. (2024). Measures of resting-state brain network segregation and integration vary in relation to data quantity: Implications for within and between subject comparisons of functional brain network organization. \u003cem\u003eCerebral Cortex\u003c/em\u003e, \u003cem\u003e34\u003c/em\u003e(2), bhad506.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang, R., Liu, M., Cheng, X., Wu, Y., Hildebrandt, A., \u0026amp; Zhou, C. (2021). Segregation, integration, and balance of large-scale resting brain networks configure different cognitive abilities. \u003cem\u003eProceedings of the National Academy of Sciences\u003c/em\u003e, \u003cem\u003e118\u003c/em\u003e(23), e2022288118.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eSong, Z., Jiang, Z., Zhang, Z., Wang, Y., Chen, Y., Tang, X., \u0026amp; Li, H. (2024). Evolving brain network dynamics in early childhood: Insights from modular graph metrics. \u003cem\u003eNeuroImage\u003c/em\u003e, \u003cem\u003e297\u003c/em\u003e, 120740.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eGavilan, D., Fern\u0026aacute;ndez-Lores, S., \u0026amp; Martinez-Navarro, G. (2020). Vividness of news push notifications and users\u0026rsquo; response. \u003cem\u003eTechnological Forecasting and Social Change\u003c/em\u003e, \u003cem\u003e161\u003c/em\u003e, 120281.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDekker, C. A., Baumgartner, S. E., Sumter, S. R., \u0026amp; Ohme, J. (2025). Beyond the Buzz: Investigating the Effects of a Notification-Disabling Intervention on Smartphone Behavior and Digital Well-Being. \u003cem\u003eMedia Psychology\u003c/em\u003e, \u003cem\u003e28\u003c/em\u003e(1), 162\u0026ndash;188.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMayerhofer, D., Haider, K., Amon, M., G\u0026auml;chter, A., O\u0026rsquo;Rourke, T., Dale, R., Humer, E., Probst, T., \u0026amp; Pieh, C. (2024). The Association between Problematic Smartphone Use and Mental Health in Austrian Adolescents and Young Adults. \u003cem\u003eHealthcare\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(6), 600.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eNdayambaje, E., \u0026amp; Okereke, P. U. (2025). The Psychopathology of Problematic Smartphone Use (PSU): A Narrative Review of Burden, Mediating Factors, and Prevention. \u003cem\u003eHealth Science Reports\u003c/em\u003e, \u003cem\u003e8\u003c/em\u003e(5), e70843.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePoulain, T., Meigen, C., Kiess, W., \u0026amp; Vogel, M. (2025). Smartphone use, wellbeing, and their association in children. \u003cem\u003ePediatric Research\u003c/em\u003e, 1\u0026ndash;7.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAhmed, O., Walsh, E. I., Dawel, A., Alateeq, K., Espinoza Oyarce, D. A., \u0026amp; Cherbuin, N. (2024). Social media use, mental health and sleep: A systematic review with meta-analyses. \u003cem\u003eJournal of Affective Disorders\u003c/em\u003e, \u003cem\u003e367\u003c/em\u003e, 701\u0026ndash;712.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eKim, S.-G., Park, J., Kim, H.-T., Pan, Z., Lee, Y., \u0026amp; McIntyre, R. S. (2019). The relationship between smartphone addiction and symptoms of depression, anxiety, and attention-deficit/hyperactivity in South Korean adolescents. \u003cem\u003eAnnals of General Psychiatry\u003c/em\u003e, \u003cem\u003e18\u003c/em\u003e, 1.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhang, R., Wang, Y., Womer, F., Yang, W., Wang, X., Xu, X., Su, Z., Wang, L., Zhou, J., Qin, S., Zhou, K., Liu, F., Zhang, X., Lin, Y., Zhang, X., Yang, J., \u0026amp; Wang, F. (2023). School-based Evaluation Advancing Response for Child Health (SEARCH): A mixed longitudinal cohort study from multifaceted perspectives in Jiangsu, China. \u003cem\u003eBMJ Mental Health\u003c/em\u003e, \u003cem\u003e26\u003c/em\u003e(1).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZheng, J., Zong, X., Tang, L., Guo, H., Zhao, P., Womer, F. Y., Zhang, X., Tang, Y., \u0026amp; Wang, F. (2024). Characterizing the distinct imaging phenotypes, clinical behavior, and genetic vulnerability of brain maturational subtypes in mood disorders. \u003cem\u003ePsychological Medicine\u003c/em\u003e, \u003cem\u003e54\u003c/em\u003e(10), 2774\u0026ndash;2784.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLe\u0026oacute;n M\u0026eacute;ndez, M., Padr\u0026oacute;n, I., Fumero, A., \u0026amp; Marrero, R. J. (2024). Effects of internet and smartphone addiction on cognitive control in adolescents and young adults: A systematic review of fMRI studies. \u003cem\u003eNeuroscience \u0026amp; Biobehavioral Reviews\u003c/em\u003e, \u003cem\u003e159\u003c/em\u003e, 105572.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLee, S., Cheong, Y., Ro, J., Bae, J., \u0026amp; Jung, M. (2024). Alterations in functional connectivity in the salience network shared by depressive symptoms and smartphone overuse. \u003cem\u003eScientific Reports\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(1), 28679.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLiu, D., Liu, X., Long, Y., Xiang, Z., Wu, Z., Liu, Z., Bian, D., \u0026amp; Tang, S. (2022). Problematic smartphone use is associated with differences in static and dynamic brain functional connectivity in young adults. \u003cem\u003eFrontiers in Neuroscience\u003c/em\u003e, \u003cem\u003e16\u003c/em\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003e\u0026Aacute;fra, E., Janszky, J., Perlaki, G., Orsi, G., Nagy, S. A., Arat\u0026oacute;, \u0026Aacute;., Szente, A., Alhour, H. A. M., Kis-Jakab, G., \u0026amp; Darnai, G. (2024). Altered functional brain networks in problematic smartphone and social media use: Resting-state fMRI study. \u003cem\u003eBrain Imaging and Behavior\u003c/em\u003e, \u003cem\u003e18\u003c/em\u003e(2), 292\u0026ndash;301.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eOng, R. H. S., Sim, H. S., Bergman, M. M., How, C. H., Png, C. A. L., Lim, C. S., Peh, L. H., \u0026amp; Oh, H. C. (2024). Prevalence and associations of problematic smartphone use with smartphone activities, psychological well-being, and sleep quality in a household survey of Singapore adults. \u003cem\u003ePLOS ONE\u003c/em\u003e, \u003cem\u003e19\u003c/em\u003e(12), e0315364.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMo, X., Li, X., Liu, M., Hu, L., Li, Q., Wang, J., Deng, H., Lv, F., Zhou, X., Mao, Y., \u0026amp; Huang, Y. (2025). Resting-state fMRI graph theory analysis for predicting selective serotonin reuptake inhibitors treatment response in adolescent major depressive disorder. \u003cem\u003eFrontiers in Psychiatry\u003c/em\u003e, \u003cem\u003e16\u003c/em\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHan, Y.-K., Zhang, H.-J., Chen, Y.-J., Liu, C., Zhang, Y.-H., Zhang, Z.-J., Jing, R.-T., Guo, L., Li, D., Chu, W.-Y., Wu, W.-J., Zhang, K., \u0026amp; Cui, L.-B. (2025). Small-world network and neuroscience. \u003cem\u003eBrain-X\u003c/em\u003e, \u003cem\u003e3\u003c/em\u003e(1), e70025.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHallquist, M. N., \u0026amp; Hillary, F. G. (2018). Graph theory approaches to functional network organization in brain disorders: A critique for a brave new small-world. \u003cem\u003eNetwork Neuroscience\u003c/em\u003e, \u003cem\u003e3\u003c/em\u003e(1), 1\u0026ndash;26.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eUpshaw, J. D., Stevens, C. E., Ganis, G., \u0026amp; Zabelina, D. L. (2022). The hidden cost of a smartphone: The effects of smartphone notifications on cognitive control from a behavioral and electrophysiological perspective. \u003cem\u003ePLOS ONE\u003c/em\u003e, \u003cem\u003e17\u003c/em\u003e(11), e0277220.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBassett, D. S., \u0026amp; Sporns, O. (2017). Network neuroscience. \u003cem\u003eNature Neuroscience\u003c/em\u003e, \u003cem\u003e20\u003c/em\u003e(3), 353\u0026ndash;364.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCai, H., Bai, W., Liu, H., Chen, X., Qi, H., Liu, R., Cheung, T., Su, Z., Lin, J., Tang, Y., Jackson, T., Zhang, Q., \u0026amp; Xiang, Y.-T. (2022). Network analysis of depressive and anxiety symptoms in adolescents during the later stage of the COVID-19 pandemic. \u003cem\u003eTranslational Psychiatry\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(1), 98.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eCushnie, A. K., Tang, W., \u0026amp; Heilbronner, S. R. (2023). Connecting Circuits with Networks in Addiction Neuroscience: A Salience Network Perspective. \u003cem\u003eInternational Journal of Molecular Sciences\u003c/em\u003e, \u003cem\u003e24\u003c/em\u003e(10), 9083.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eLin, Y.-H., Chang, L.-R., Lee, Y.-H., Tseng, H.-W., Kuo, T. B. J., \u0026amp; Chen, S.-H. (2014). Development and Validation of the Smartphone Addiction Inventory (SPAI). \u003cem\u003ePLOS ONE\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e(6), e98312.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePan, C., Ma, Y., Wang, L., Zhang, Y., Wang, F., \u0026amp; Zhang, X. (2024). From Connectivity to Controllability: Unraveling the Brain Biomarkers of Major Depressive Disorder. \u003cem\u003eBrain Sciences\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(5), 509.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eRolls, E. T., Huang, C.-C., Lin, C.-P., Feng, J., \u0026amp; Joliot, M. (2020). Automated anatomical labelling atlas 3. \u003cem\u003eNeuroImage\u003c/em\u003e, \u003cem\u003e206\u003c/em\u003e, 116189.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eChen, Y., Zhao, P., Pan, C., Chang, M., Zhang, X., Duan, J., Wei, Y., Tang, Y., \u0026amp; Wang, F. (2024). State- and trait-related dysfunctions in bipolar disorder across different mood states: A graph theory study. \u003cem\u003eJournal of Psychiatry and Neuroscience\u003c/em\u003e, \u003cem\u003e49\u003c/em\u003e(1), E11\u0026ndash;E22.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBuchanan, C. R., Bastin, M. E., Ritchie, S. J., Liewald, D. C., Madole, J., Tucker-Drob, E. M., Deary, I. J., \u0026amp; Cox, S. R. (2020). The effect of network thresholding and weighting on structural brain networks in the UK Biobank. \u003cem\u003eNeuroImage\u003c/em\u003e, \u003cem\u003e211\u003c/em\u003e, 116443.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eTheis, N., Rubin, J., Cape, J., Iyengar, S., \u0026amp; Prasad, K. M. (2023). Threshold Selection for Brain Connectomes. \u003cem\u003eBrain Connectivity\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e(7), 383\u0026ndash;393.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePan, C., Zhang, X., Zheng, H., Zhang, Y., Su, Z., Zhang, C., \u0026amp; Zhang, W. (2025). Adaptive Control of Dynamic Networks. \u003cem\u003eIEEE Transactions on Network Science and Engineering\u003c/em\u003e, 1\u0026ndash;15.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eHassett, J. D., Craig, B. T., Hilderley, A., Kinney-Lang, E., Yeates, K. O., MacMaster, F. P., Miller, J., Noel, M., Brooks, B. L., Barlow, K., Lebel, C., Kirton, A., \u0026amp; Carlson, H. L. (2024). Development of the whole-brain functional connectome explored via graph theory analysis. \u003cem\u003eAperture Neuro\u003c/em\u003e, \u003cem\u003e4\u003c/em\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eFortunato, S., \u0026amp; Hric, D. (2016). Community detection in networks: A user guide. \u003cem\u003ePhysics Reports: Review Section of Physics Letters\u003c/em\u003e, \u003cem\u003e659\u003c/em\u003e, 1\u0026ndash;44.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003ePan, C., Zhang, Q., Zhu, Y., Kong, S., Liu, J., Zhang, C., Wang, F., \u0026amp; Zhang, X. (2024). Module control of network analysis in psychopathology. \u003cem\u003eiScience\u003c/em\u003e, \u003cem\u003e27\u003c/em\u003e(7).\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhou, L., Xie, Y., Wang, R., Fan, Y., \u0026amp; Wu, Y. (2023). Dynamic segregation and integration of brain functional networks associated with emotional arousal. \u003cem\u003eiScience\u003c/em\u003e, \u003cem\u003e26\u003c/em\u003e(5), 106609.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eMontag, C., \u0026amp; Becker, B. (2023). Neuroimaging the effects of smartphone (over-)use on brain function and structure\u0026mdash;A review on the current state of MRI-based findings and a roadmap for future research. \u003cem\u003ePsychoradiology\u003c/em\u003e, \u003cem\u003e3\u003c/em\u003e, kkad001.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eAhn, J., Lee, D., Namkoong, K., \u0026amp; Jung, Y.-C. (2021). Altered Functional Connectivity of the Salience Network in Problematic Smartphone Users. \u003cem\u003eFrontiers in Psychiatry\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e, 636730.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWolf, N. D., Henemann, G. M., Schmitgen, M. M., Koenig, J., Bach, P., \u0026amp; Wolf, R. C. (2025). Screens and brains: Multimodal neuroimaging insights into mechanisms of problematic smartphone use. \u003cem\u003eProgress in Neuro-Psychopharmacology and Biological Psychiatry\u003c/em\u003e, \u003cem\u003e142\u003c/em\u003e, 111531.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eEinh\u0026auml;user, W., Neubert, C. R., Grimm, S., \u0026amp; Bendixen, A. (2024). High visual salience of alert signals can lead to a counterintuitive increase of reaction times. \u003cem\u003eScientific Reports\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e(1), 8858.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eZhou, J., Duan, J., Liu, X., Wang, Y., Zheng, J., Tang, L., Zhao, P., Zhang, X., Zhu, R., \u0026amp; Wang, F. (2024). Functional network characteristics in adolescent psychotic mood disorder: Associations with symptom severity and treatment effects. \u003cem\u003eEuropean Child \u0026amp; Adolescent Psychiatry\u003c/em\u003e, \u003cem\u003e33\u003c/em\u003e(7), 2319\u0026ndash;2329.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eWang, S.-B., Xu, W.-Q., Gao, L.-J., Tan, W.-Y., Zheng, H.-R., Hou, C.-L., \u0026amp; Jia, F.-J. (2023). Bridge connection between depression and anxiety symptoms and lifestyles in Chinese residents from a network perspective. \u003cem\u003eFrontiers in Psychiatry\u003c/em\u003e, \u003cem\u003e14\u003c/em\u003e.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eDoll, A., H\u0026ouml;lzel, B. K., Boucard, C. C., Wohlschl\u0026auml;ger, A. M., \u0026amp; Sorg, C. (2015). Mindfulness is associated with intrinsic functional connectivity between default mode and salience networks. \u003cem\u003eFrontiers in Human Neuroscience\u003c/em\u003e, \u003cem\u003e9\u003c/em\u003e, 461.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eBremer, B., Wu, Q., Mora \u0026Aacute;lvarez, M. G., H\u0026ouml;lzel, B. K., Wilhelm, M., Hell, E., Tavacioglu, E. E., Torske, A., \u0026amp; Koch, K. (2022). Mindfulness meditation increases default mode, salience, and central executive network connectivity. \u003cem\u003eScientific Reports\u003c/em\u003e, \u003cem\u003e12\u003c/em\u003e(1), 13219.\u003c/span\u003e\u003c/li\u003e\u003cli\u003e\u003cspan\u003eYue, W. L., Ng, K. K., Koh, A. J., Perini, F., Doshi, K., Zhou, J. H., \u0026amp; Lim, J. (2023). Mindfulness-based therapy improves brain functional network reconfiguration efficiency. \u003cem\u003eTranslational Psychiatry\u003c/em\u003e, \u003cem\u003e13\u003c/em\u003e, 345.\u003c/span\u003e\u003c/li\u003e\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Problematic smartphone use, Adolescents, Mood disorders, Resting-state fMRI, Functional brain networks","lastPublishedDoi":"10.21203/rs.3.rs-8256165/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-8256165/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground and purpose\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eProblematic smartphone use (PSU) is highly prevalent in mood-disordered adolescents, but it is unclear whether it reflects global disruption of functional architecture or specific circuits, and how it relates to depressive symptom.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eA total of 199 adolescents with mood disorders (99 major depressive disorder, 100 bipolar disorder) were stratified into PSU (n = 86) and non-PSU (n = 113) groups using DSM-5–adapted criteria. Resting-state fMRI networks were analyzed with multi-scale graph measures across proportional densities. Hamilton Depression Rating Scale items were modelled as symptom networks to derive symptom communities and brain–symptom associations.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eGlobal small-world indices did not differ between groups. PSU was instead associated with selective reweighting at modular and nodal levels, including migration of attention regions into a hyper-cohesive reward–limbic core and reduced broadcast roles of executive hubs. Only in the PSU group did greater executive-network cohesion predict more severe anxiety-somatic symptoms.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eIn mood-disordered adolescents, PSU is instantiated as selective network reweighting, not global connectome breakdown, and alters the coupling between executive control and anxiety-somatic symptom clusters. These multi-level network signatures suggest mechanistic targets for interventions that rebalance communication among reward, salience and executive systems.\u003c/p\u003e","manuscriptTitle":"Selective Network Reweighting in Adolescents with Mood Disorders and Problematic Smartphone Use","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-12-03 07:38:24","doi":"10.21203/rs.3.rs-8256165/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"cef7f445-b232-4a2d-8350-db5f7e7d2309","owner":[],"postedDate":"December 3rd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":58993031,"name":"Health sciences/Diseases/Psychiatric disorders/Addiction"},{"id":58993032,"name":"Health sciences/Diseases/Psychiatric disorders/Depression"},{"id":58993033,"name":"Health sciences/Diseases/Psychiatric disorders/Bipolar disorder"}],"tags":[],"updatedAt":"2026-02-05T15:21:25+00:00","versionOfRecord":[],"versionCreatedAt":"2025-12-03 07:38:24","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-8256165","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-8256165","identity":"rs-8256165","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}