Problematic generative AI use and mental health risks among Chinese university students: latent profiles, network structure, and health literacy as a modifiable resource
Aug 2026· Frontiers in Public Health· 0 citations· 57 references
TL;DR
Findings may help universities identify students who report difficulty controlling AI use, academic worry, delayed bedtime, academic avoidance, or trouble setting limits on AI use and point to health literacy, self-regulation, and time management as candidate resources for future longitudinal and intervention studies.
Abstract
Generative artificial intelligence (GenAI) has become a routine academic tool for university students, but dysregulated use may co-occur with anxiety, sleep problems, psychological distress, and impaired academic functioning. This study examined GenAI-related digital behavior as a public health issue and assessed whether health literacy and self-regulated learning were associated with lower-risk profiles.
A cross-sectional survey was conducted among Chinese university students from six universities in eastern, central, and western China. A total of 2,140 questionnaires were submitted; after four-step data-quality screening, 1,872 valid responses were retained. Measures assessed GenAI-use dysregulation, academic anxiety, sleep problems, psychological distress, health literacy, self-regulated learning, academic engagement, and demographic and use-related covariates. Latent profile analysis identified GenAI-use and mental-health risk profiles. Profile differences were examined using omnibus and
post hoc
tests, multinomial logistic regression examined external factors associated with profile membership, and regularized network analysis identified central and bridge nodes.
A four-profile solution showed the best balance of fit, entropy, class size, and interpretability (entropy = 0.891). The profiles were low-risk adaptive users (39.9%), anxiety-prone dependent users (25.7%), sleep-disrupted overusers (19.8%), and high-risk dysregulated users (14.6%). Confirmatory factor analysis supported the measurement structure of the adapted and selected scales (CFI = 0.958, TLI = 0.951, RMSEA = 0.036, SRMR = 0.041), and Harman's single-factor test did not indicate serious common method bias. Higher health literacy and self-regulated learning were associated with greater odds of low-risk rather than high-risk membership. Network analysis identified loss of control, academic worry, delayed bedtime, and academic avoidance as central or bridge nodes, whereas credibility evaluation and time management showed negative bridge expected influence.
GenAI-related digital behavior among university students varied in ways that involved dysregulated use, academic anxiety, sleep problems, distress, and academic functioning. These findings may help universities identify students who report difficulty controlling AI use, academic worry, delayed bedtime, academic avoidance, or trouble setting limits on AI use. They also point to health literacy, self-regulation, and time management as candidate resources for future longitudinal and intervention studies.
Personal values may be important for university students’ mental health. However, most studies focus on single indicators such as anxiety, depression, or well-being, with limited attention to overall mental health profiles. How different value dimensions relate to mental health heterogeneity from a dual-factor perspective remains unclear. In this cross-sectional study, latent profile analysis and multinomial logistic regression were used to identify mental health profiles and examine their associations with personal values. A total of 14,929 Chinese university students completed self-report measures of personal values, anxiety, depression, life satisfaction, and psychological resilience. Three mental health profiles were identified: complete mental health, vulnerable, and troubled. The symptomatic but content profile was not found. Multinomial logistic regression showed that self-transcendence was linked to more favorable profiles, whereas self-enhancement was linked to less favorable profiles. Openness to change showed a conditional pattern: it was generally linked to lower likelihood of vulnerable or troubled membership, but among vulnerable and troubled students, higher openness to change was linked to higher likelihood of troubled membership. Conservation showed a weaker role and mainly distinguished the complete mental health and vulnerable profiles. This study provides a person-centered view of mental health heterogeneity among Chinese university students. The findings suggest that personal values are differentially associated with overall mental health profiles and support integrating symptom screening, value guidance, and positive psychological resource development in university mental health education.
BACKGROUND
Problematic social media use (PSMU) is a prevalent behavioral concern among adolescents and is closely linked to a range of adverse mental health outcomes. This study aimed to explore the heterogeneous structure of PSMU in Chinese adolescents using latent profile analysis (LPA), and to investigate differences in demographic characteristics, peer victimization, and psychological and sleep outcomes across latent profiles.
METHODS
A total of 51,219 middle and high school students participated in this cross-sectional study. All participants completed self-report questionnaires including the Bergen Social Media Addiction Scale (BSMAS), Patient Health Questionnaire-9 (PHQ-9), Generalized Anxiety Disorder-7 (GAD-7), Pittsburgh Sleep Quality Index (PSQI), and Multidimensional Peer Victimization Scale (MPVS). LPA was conducted to identify distinct PSMU subgroups. Chi-square tests and one-way analysis of variance (ANOVA) were applied to examine group differences.
RESULTS
A three-profile solution was selected as optimal based on model fit indices (AIC, BIC, SABIC), entropy, and bootstrapped likelihood ratio test (BLRT). The three profiles represented a gradient of PSMU severity: low-risk, moderate-risk, and high-risk (accounting for 10.50% of the sample). Depression, anxiety, peer victimization, and sleep disturbance all increased monotonically with rising PSMU risk, demonstrating a robust dose-response relationship. All between-profile differences were statistically significant (all p < 0.001).
CONCLUSIONS
Adolescent PSMU exhibits a continuous three-level risk spectrum rather than discrete categorical subtypes. Mental health burden increases consistently with the severity of PSMU. These findings support the continuous model of problematic online behaviors and provide empirical support for stratified screening and targeted intervention strategies for adolescent PSMU.
Jingyuan Mai, Huan Gao, R. He et al.· Journal of Affective Disorde...· 0 citations
BACKGROUND
Problematic social media use among adolescents is increasingly recognized as a multidimensional phenomenon with distinct symptom profiles. While aggregate scoring approaches are common, examining individual symptoms may offer a more efficient entry point for identifying adolescents at higher risk. This study examined the prevalence of individual problematic social media use symptoms and their associations with health, mental well-being, and lifestyle indicators among Lithuanian adolescents.
METHODS
A cross-sectional study was conducted using data from the 2022 Health Behaviour in School-aged Children survey in Lithuania (n = 6628; aged 11-17 years). Problematic social media use was assessed using the nine-item Social Media Disorder scale with binary response options. Fifteen outcomes were examined across three domains: health (subjective health, somatic complaints), psychological well-being (life satisfaction, psychological complaints, psychological well-being, anxiety, loneliness, happiness), and lifestyle (daily physical activity, vigorous physical activity, cigarette smoking, e-cigarette smoking, alcohol consumption, sleep duration, sleep quality). Logistic regression analyses were performed for each symptom-outcome pair, adjusting for gender, age, and socioeconomic status.
RESULTS
Escape-motivated use was the most prevalent symptom (51.2%), followed by loss of control (39.2%) and preoccupation (27.2%). Escape-motivated use was also the strongest predictor of adverse outcomes, showing strong evidence for associations across 14 out of 15 health and lifestyle indicators. Girls endorsed seven of nine symptoms at higher rates than boys, with the largest disparities observed for escape-motivated use and loss of control. An inverse age pattern was observed, with younger adolescents reporting higher symptom prevalence than older ones.
CONCLUSIONS
Symptom-level analysis reveals meaningful variation in the predictive relevance of individual problematic social media use criteria, with escape-motivated use emerging as the symptom most consistently associated with adverse outcomes.
Vladas Golambiauskas, T. Vaičiūnas, Justė Lukoševičiūtė-Barauskienė et al.· BMC Public Health· 0 citations
Mental health issues, especially depressive symptoms, among young adults represent a public health challenge. Conventional psychological assessment tools have limited sensitivity and specificity for identifying individuals at risk. This study aims to develop an explainable machine learning-based model to stratify concurrent depression risk in young adults. This study included 100,257 college students and collected mental health variables including depression, anxiety, resilience, parent-child relationship, and duration of mobile phone usage. The screening capabilities of 13 machine learning algorithms were systematically evaluated and compared. The SHapley Additive exPlanations (SHAP) framework was employed for the interpretability of the final model. The median scores for parent-child relationship, resilience, anxiety, and mobile phone usage time was 42.0, 28.0, 1.0 and 28.0, respectively. Among the 13 machine learning algorithms, the XGBoost model demonstrated superior performance. The final multivariate screening model achieved an area under the curve (AUC) of 0.887, a sensitivity of 0.787, a specificity of 0.830, and an accuracy of 0.816 in classifying young adults' concurrent depression risk. The SHAP analysis showed the importance of each variable: anxiety (2.303) > resilience (0.774) > parent-child relationship (0.708) > mobile phone usage time (0.411). The final multivariate model exhibited stable performance during cross-validation (AUC = 0.885 ± 0.032), significantly better than the single-variable model (P < 0.001) and better screening reliability (Brier score 0.153). The final multivariate XGBoost model provides a highly accurate and interpretable approach for young adults' depression risk stratification. As the model was developed using cross-sectional data collected during the COVID-19 campus lockdown, prospective validation is required before clinical deployment. Notably, anxiety level emerged as the most influential risk factor, and resilience demonstrated a significant protective effect.
BACKGROUND
While generative artificial intelligence (GenAI) has been rapidly adopted by college students, its relationship with mental health remains unclear. Most studies treat AI users as homogeneous, overlooking heterogeneity in GenAI use. This study identified latent profiles of GenAI use among Chinese college students and examined their associations with depression and anxiety.
METHODS
A cross-sectional survey of 5748 Chinese college students assessed AI usage, motivations, AI literacy, dependency, and symptoms of depression and anxiety using the Beck Depression Inventory-II and Beck Anxiety Inventory. Latent profile analysis identified usage patterns. An exploratory random-forest classifier with SHapley Additive exPlanations (SHAP) evaluated whether these profiles could be predicted from psychological and demographic correlates not used for profile construction and identified key distinguishing factors.
RESULTS
Four profiles were identified: Rational-Tool, Moderate-Recreational, Problem-Dependent, and Light-Exploratory. The Problem-Dependent profile (15%), characterized by high escapism motivation, low AI literacy, and high dependency, showed the highest depression and anxiety levels, whereas the Rational-Tool profile, characterized by high AI literacy and instrumental motivation, showed the most favorable outcomes. The profiles were recovered with good discrimination (accuracy = 0.81; macro-average AUC = 0.89), with the Problem-Dependent profile being the most distinguishable (AUC = 0.94). SHAP analysis identified depression, smartphone addiction, and executive-function difficulties as key correlates of the Problem-Dependent profile, whereas conscientiousness was central to the Rational-Tool profile.
CONCLUSIONS
GenAI use profiles were associated with different levels of depression and anxiety among college students. Person-centered prevention strategies should focus on usage motivation, AI literacy, and dependency rather than frequency alone.
Hao-Yang Shi, Tongyi Zhang, Zi-Hao Wang et al.· Journal of Affective Disorde...· 0 citations
Abstract Background Internet addiction (IA) has been consistently associated with adverse mental health outcomes, but less is known about whether adolescents with IA seek mental health support, and whether associations between help-seeking and mental health problems differ across pathways. Objective This study aimed to describe mental health help-seeking patterns across internet use and IA status, and examine the independent and interactive associations of IA and help-seeking with mental health problems. Methods Data were drawn from 2 repeated cross-sectional surveys conducted in Shandong Province in 2023 and 2024. A total of 171,970 junior middle school students were included. IA was assessed using a 9-item scale adapted from DSM-5 (Diagnostic and Statistical Manual of Mental Disorders [Fifth Edition]) Internet Gaming Disorder criteria to assess internet use–related symptoms. Students using the internet for at least 4 hours per day and endorsing at least 4 symptoms were classified as having IA. Help-seeking was categorized as no help-seeking, informal-only help-seeking, and any formal help-seeking. Mental health problems were defined as clinically significant depressive or anxiety symptoms. Pairwise logistic regression compared help-seeking patterns, and Poisson regression with school-clustered CR2 robust SEs estimated adjusted prevalence ratios (PRs). Results Overall, 1.4% (2406/171,970) met the definition of IA; 47.9% (82,402/171,970), 42.1% (72,461/171,970), and 9.9% (17,107/171,970) reported no help-seeking, informal-only help-seeking, and any formal help-seeking, respectively. Among adolescents with IA, 59.9% (1440/2406) reported no help-seeking, 33.8% (814/2406) reported informal-only help-seeking, and only 6.3% (152/2406) reported any formal help-seeking. Compared with nonaddictive internet users, adolescents with IA had lower odds of informal-only versus no help-seeking (OR [odds ratio] 0.690, 95% CI 0.631‐0.754), any formal versus no help-seeking (OR 0.540, 95% CI 0.455‐0.640), and any formal versus informal-only help-seeking (OR 0.786, 95% CI 0.659‐0.937). Compared with nonaddictive internet users, adolescents with IA had a higher prevalence of mental health problems (PR 3.662, 95% CI 3.428‐3.912). Informal-only (PR 0.581, 95% CI 0.550‐0.614) and any formal help-seeking (PR 0.461, 95% CI 0.420‐0.505) were associated with lower prevalence overall. Interaction terms for IA with informal-only help-seeking (PR 1.469, 95% CI 1.336‐1.616) and any formal help-seeking (PR 2.033, 95% CI 1.696‐2.437) indicated weaker inverse associations among adolescents with IA. Within the IA group, informal-only help-seeking was associated with lower prevalence (PR 0.848, 95% CI 0.784‐0.917), whereas any formal help-seeking was not clearly associated with lower prevalence (PR 0.945, 95% CI 0.814‐1.097). Sensitivity analyses supported these patterns. Conclusions Adolescents with IA had a substantially higher prevalence of clinically significant depressive and anxiety symptoms and were less likely to report mental health help-seeking, especially formal help-seeking. These findings highlight a high-burden, low help-seeking subgroup and suggest strengthening school-based mental health services, improving linkage between informal and formal support, reducing stigma, and developing low-threshold digital mental health pathways.
Afei Qin, Meiqi Wang, Lianlong Yu et al.· Journal of Medical Internet...· 0 citations