Item-level stressor profiles provide informative, partly generalisable evidence on stressors linked to international students' mental health, with core stressors identified alongside context-specific factors that differ between UK and China settings.
Key Findings
Results
Random forest models explained approximately 22–35% of variance in perceived stress and 25–27% of variance in mental well-being across international student samples.
Six models total were evaluated across two sites and two outcomes (perceived stress and mental well-being).
Random forest outperformed elastic net models in explained variance.
Models were evaluated using nested cross-validation to reduce overfitting bias.
The sample consisted of 481 students across two sites: Chinese students in the UK and international students in China.
Results
Concerns about mental well-being, sleep difficulties, physical health, and family or self-imposed academic expectations consistently ranked as high-priority stressors across both sites.
These stressors were identified as consistently highly ranked across all six machine-learning models.
Rankings were derived from elastic net and random forest feature importance metrics.
The cross-site consistency of these stressors suggests they represent core, generalisable contributors to poor outcomes among international students.
Perceived stress was measured using the PSS-10 and mental well-being using the Warwick-Edinburgh Mental Wellbeing Scale (WEMWBS).
Results
Social isolation and restricted social life were strongly associated with lower mental well-being in international students.
These stressors were identified as particularly prominent predictors of the mental well-being outcome (WEMWBS).
Findings held across both the UK and China sites.
The cross-sectional design limits causal inference about the direction of these associations.
Results
Context-specific stressors differed between the UK and China settings, with time-management pressure, low confidence in interacting with others, and loneliness more strongly associated with poorer outcomes in the UK.
In the UK setting, time-management pressure, low confidence in interacting with others, and loneliness were more strongly associated with poorer mental health outcomes.
In the Chinese setting, difficulties balancing study and personal life and not knowing where to seek help played a larger role.
These context-specific differences were identified by comparing feature importance rankings between site-specific models.
The findings suggest that while core stressors generalise, locally salient challenges require context-sensitive intervention adaptations.
Results
Cross-site external validation indicated that a substantial portion of the item-level stressor signal generalised across the UK and China settings.
Cross-site transportability was evaluated by training models on one site and testing on the other.
The generalisability of stressor signals supports the use of item-level frameworks across different international student populations.
Some attenuation in predictive performance was expected and observed when models were applied to the external site, consistent with context-specific differences.
This is described as 'cross-site external testing' rather than a longitudinal or randomised design.
Background
International students experience a high mental health burden and face barriers to accessing and sustaining support, motivating the use of machine-learning models to prioritise specific item-level stressors.
The study notes that domain-level stress frameworks provide limited guidance on which specific stressors should be prioritised.
The survey was cross-sectional and conducted online across two sites.
Total sample size was 481 students.
The two populations studied were Chinese students in the UK and international students in China.
Conclusions
The study identified concrete, context-sensitive stressor targets that can inform how university mental health services prioritise support and prevention for international students.
Authors conclude that services should focus on core stressors while adapting provision to locally salient challenges.
High-impact stressors identified include mental well-being concerns, sleep difficulties, physical health, academic expectations, social isolation, and restricted social life.
The findings are intended to guide prevention and intervention design in university mental health services.
The authors frame item-level stressor profiling as offering more actionable guidance than domain-level frameworks.
What This Means
This research suggests that international students — whether Chinese students studying in the UK or international students studying in China — face a distinct set of mental health stressors that can be identified and prioritised using machine-learning methods. The study surveyed 481 students and used two types of predictive models to identify which specific worries and difficulties were most strongly linked to higher perceived stress and lower mental well-being. The models explained roughly a quarter to a third of the variation in students' mental health outcomes, highlighting key stressors including sleep problems, concerns about mental well-being, physical health issues, and pressure from academic expectations set by themselves or their families.
The study also found that social isolation and a restricted social life were consistently linked to lower mental well-being across both settings. However, some stressors were more prominent in particular locations: students in the UK were more affected by time-management pressure, low confidence in social interactions, and loneliness, while students in China were more affected by difficulty balancing academic and personal life, and uncertainty about where to find help. Importantly, the stressor patterns identified in one country's data were partially useful for predicting outcomes in the other country, suggesting that while some challenges are universal, local context still matters.
This research suggests that universities could benefit from moving beyond broad categories of student stress and instead focusing support and prevention efforts on these specific, high-impact stressors. Rather than a one-size-fits-all approach, the findings point toward tailoring mental health services to address both the common core challenges international students face and the locally specific difficulties that vary by setting.
Zhao J, Saunders R, Barnett P, Li S, Niu Y, Pilling S, et al.. (2026). Factors underpinning international student mental well-being: a UK and China survey.. BMJ mental health. https://doi.org/10.1136/bmjment-2026-302624