Interplay of academic stress, digital fatigue, and loneliness in predicting university students’ mental health in the post-COVID era Authors: Amjad Islam Amjad, Saira, Saba Firdos, Priyadarshini Muthukrishnan, Mohamad Ahmad Saleem Khasawneh, Thabet Bin Saeed Al-Kahlan Published: 2026 ABSTRACT AbstractThe COVID-19 pandemic created major challenges to students’ mental health. Research that university students continue to experience mental health challenges due to the changes in higher education and the increased technology integration in the post-COVID-19 era. The objective of this study was to examine association between academic stress related to technology integration and students’ mental health problems (MHPs), with digital fatigue as a mediating variable and loneliness as a moderating variable. The study's research design was a cross-sectional survey. This study was guided by Job-Demands Resources (JD-R) Theory. The data were collected online from 388 university students through the post-COVID-19 University Students’ Well-being Scale (PC-USWS). Academic stress was significantly associated with mental health problems (β = .639, SE = .04, 95% CI [.566, .713], t = 17.12, p < .001, R2 = .43). Digital fatigue partially explained the association between academic stress and mental health problems (ab = .18, SE = .04, 95% CI [.107, .261], Z = 4.64, p < .001, R2 = .51). The interaction between academic stress and loneliness was also statistically significant (βAS*Lo = .11, SE = .05, 95% CI [.008, .215], Z = 2.16, p = .03). The study suggests a more effective use of digital tools and greater socialization among students to improve their mental health. This study had several practical, policy, and research implications.Keywords: Academic stress, Digital fatigue, Loneliness, Mental health, Post-COVID era, Psychological harm Introduction The COVID-19 pandemic changed the course of the educational landscape, and it left a lasting effect on university students’ mental health problems (MHPs) [1, 2]. Social isolation and lockdowns were associated with higher anxiety and stress levels among students [3]. Students suffered a lot during the COVID pandemic, and it kept on influencing even when the universities reopened. COVID-19 severely disrupted the delivery of quality education, one of the important aspects of Sustainable Development Goal 4. A meta-analysis study revealed that COVID affected college students at a large scale, nearly one-third to two-fifths, who reported clinically relevant symptoms of anxiety and depression in the COVID or post-COVID era. Most of these reports were from the resource-constrained countries [4]. Another longitudinal empirical study indicated that the challenges posed by COVID are not persistent, and things are not yet returning to normal in the post-COVID era. They are still influencing students’ MHPs by disrupting their social lives and causing financial and academic uncertainty. After COVID, most of the higher education institutions introduced digital-enriched and hybrid learning models. These hybrid models bring uncertainty of time and facilities among students [5]. Thus, it is important to explore how such specific study-risk factors develop students’ post-pandemic MHPs at the university level. COVID-19 was associated with higher levels of academic academic stress (AS) among university students [6, 7]. It was one of the major factors affecting students’ MHPs during the pandemic [8, 9]. They were offered smart syllabi, urgently compressed curricula, continual assessment, and performance expectations, besides academic and future career uncertainties [10, 11]. These uncertainties were largely due to transitions between online, hybrid, and in-person formats introduced by various universities, depending on their resources. Research showed that academic stress is strongly associated with anxiety, depression symptoms, and psychological distress among university students [12, 13]. This shift to hybrid and remote learning during and after COVID-19 was associated with greater ambiguity around workloads and blurred study-home boundaries. Moreover, reduced access to informal academic support reinforces stress-distress cycles [14]. Academic stress is considered a chronic study burden in the post-pandemic era, affecting students’ emotional reserves and MHPs. In this post-pandemic and tech-driven era, digital fatigue (DF) is another emerging stressor for university students [15, 16]. It is defined as physical, emotional, cognitive, and social exhaustion stemming from prolonged and excessive use of technologies for academic tasks [17]. The expansion of online learning environments, learning management systems, and videoconferencing platforms, along with sudden switching between online and hybrid modes, has significantly linked with university students’ daily screen time [18]. DF comes up with several psychological issues, such as reduced motivation, concentration problems, anxiety, depression, and sleep disturbance [19]. This issue may become more severe when tech-driven education continued with insufficient facilities, inadequate digital infrastructure, and limited opportunities for offline recovery [17]. Excessive screen time has been associated with poorer mental health outcomes among students [20]. In the modern era of technology, students feel lonely even after joining several social networking or online study groups [21]. During and after the pandemic, university campuses remained closed, and this distancing adversely influenced students’ well-being [22]. A longitudinal study was carried out during the pandemic, which revealed that loneliness is linked with anxiety, depression, and stress among university students [5]. Similarly, a study from Middle Eastern and Asian regions indicated that a significant majority of the students reported about their loneliness, which adversely influenced their life satisfaction and MHPs [23]. Moreover, students who enrolled in university programs during the pandemic had less interaction with their peers due to online classes. This reduced their on-campus presence and disrupted the peer networks. Moreover, hybrid arrangements and re-establishment of supportive interactions and reopening of universities after the pandemic made those students lonelier in disengaged bonding with their peers, which can influence their MHPs [24]. Although there is a significant volume of research on students’ MHPs and pandemic effects, there are still research gaps. For example, several studies were conducted on MHPs, loneliness, academic stress, and digital fatigue in isolation [25–28]. However, these variables rarely been examined within an integrated framework. Similarly, several studies were conducted in high-income countries, where digital access, institutional support, and social norms differed from those in low-income or resource-constrained regions [12, 29]. To address these gaps, the present study was designed. The objective of this study was to examine how academic stress associates with the mental health problems of university students, with digital fatigue and loneliness as mediating and moderating factors in the post-COVID era. This study will provide empirical evidence from a low-income country, such as Pakistan, on how universities are managing students’ MHPs. Literature review Academic stress and mental health The literature suggests that academic stress is strongly associated with students’ mental. This issue became more prominent in the post-COVID era. For example, a recent systematic review study revealed that academic stress has strong correlations with depression, anxiety, and the MHPs of university students [30]. Another study on students’ MHPs collected data from 843 college-going students and revealed that after COVID-19, students reported poorer mental health that was associated with academic stress [31]. It suggests that COVID-related teaching and technical changes, and changing academic needs, impact students’ mental health. Another study revealed that academic stress is a significant predictor and contributor to undergraduate students’ psychological well-being [32]. Similarly, a study on upper secondary students revealed that academic stress is a significant predictor of students’ psychological distress. This suggests that academic pressure partly contributes to the mental health challenges of emerging adults [33]. In the post-COVID era, another study on university students revealed that returning to face-to-face classes, due to academic demands and emotional cost, increased students’ academic stress [34]. A study was carried out in Pakistan’s local context. It revealed a strong negative correlation between students’ academic stress and their MHPs. The researcher found that the students who experienced higher stress showed poorer mental health [35]. These studies clearly indicated that students’ emotional well-being and mental health are influenced by the academic stress they face during the transition period in the post-pandemic era. Thus, based on these studies, the following directional hypothesis was formulated. H1: University students’ academic stress significantly relates to their mental health problems in the post-COVID era. Academic stress, digital fatigue, and mental health Studies on digitalization in the post-COVID period emphasize that academic stress is not experienced through screens. Research reveals that digital burnout, or digital fatigue, results from academic pressures such as tech-based learning tasks, heavy online course loads, and digital connectivity. For example, a study on nursing students revealed that digital fatigue was shaped by digital stress, such as coursework, which predicted poor psychological and physical health [16]. Another study highlighted the academic demands and digital overload. After collecting data from 206 university students in China, the data revealed that university students who had digital fatigue and burnout showed a higher level of stress, affecting their mental health [36]. Another study supported the argument for the case of health-science students. The researchers found that students who spent more time in online learning environments using screens experience higher levels of academic burnout and fatigue [37]. Another recent study on university students revealed that digital screen time and online learning led students to poor mental health and a lack of sleep quality [38]. Use of digital media or online apps in the post-COVID period is not without challenges. For example, a study revealed that medical students who spent more time online were experiencing Zoom fatigue, which was leading to depression. They also identified that videoconferencing for academic purposes is harmful to students’ emotional well-being [39]. Studies in the field of technostress also showed similar results. For example, a study on higher education students revealed that technostress significantly influences students’ MHPs. They further supported the idea that this stress is linked to poor sleep, an indicator of psychological well-being. These studies suggest that academic stress is associated with excessive use of digital tools, which, in turn, was negatively associated with students’ mental health. However, no study directly measured the mediational effect of digital fatigue on the relationship between academic stress and MHPs among university students. Considering this knowledge gap and the directional nature of associations mentioned above, the following directional hypothesis was formulated. H2: Digital fatigue of university students significantly mediates the relationship between their academic stress and mental health problems in the post-COVID era. Academic stress, loneliness, and mental health During the COVID-19 period, loneliness was considered the major concern for university students. A study conducted in Australia during and after the pandemic revealed that university students reported higher levels of loneliness and a worse sense of belonging. This led to students’ poor mental health [40]. Similarly, a study in the UK revealed that 80% of the university students reported mild to severe loneliness, which led them to stress, anxiety, depression, and suicidal ideation. This shows that loneliness is a significant challenge to students’ mental health [41]. Similar findings were also reported in other countries of the world. For example, a study of university students in a resource-constrained country (Bangladesh) revealed that students felt lonely due to several factors, including their social and economic roots of loneliness. This, in turn, affected their psychological health [29]. A study in Spanish university students revealed that during the post-pandemic era, students felt very lonely, and this has a strong relationship with MHPs and poor self-rated health [42]. Research also supports that loneliness and academic stress are closely associated. For example, a study found that loneliness is strongly correlated with students’ academic stress [43]. It suggested that feeling socially disconnected can enhance the emotional cost of academic demands. In another study, which was conducted on Korean university students, it was found that loneliness is a predictor of school burnout and reduced life satisfaction. Lonely students experienced higher levels of burnout, and this lowered students’ mental health [44]. Similarly, a study among Spanish adults found that loneliness was a strong negative predictor of mental health [45]. These arguments indicate that, in the case of social connections, people have limited resources to cope with stress and anxiety. This shows that lonely students are not capable of handling academic stress and may experience a decline in their MHPs. After an extensive literature review, the authors found no studies that examined the moderating role of loneliness in the relationship between university students’ academic stress and their MHPs, particularly in the post-COVID era. Thus, the present study formulated the third directional hypothesis to address this research gap. H3: Loneliness of university students significantly moderates the relationship between their academic stress and mental health problems in the post-COVID era. Theoretical background The current study was guided by the Job-Demands Resources (JD-R) Theory. This theory is widely used in the literature to understand participants’ well-being and MHPs within the organizations [46]. This theory is suitable for the current research because it views the study environment as a mix of “demands” that drain students’ energy and “resources” that help them cope and stay stronger [47]. In JD-R, students’ demands are aspects of study that demand sustained effort and are therefore psychologically or physically costly. Study resources are the things that support learning, protect against stress, and promote well-being with support and feedback. This theory has become an important framework for understanding students’ well-being in higher education in post-COVID contexts [46]. In the context of the current study, academic stress is a core study demand. It includes perceived overload, time pressure, and fear of failure that come from several academic tasks, such as coursework, exams, and uncertainty about academic progress. The literature indicates that academic stress intensified during and after COVID-19, with implications for students’ mental health and well-being [48, 49]. Similarly, digital fatigue is another demand arising from heavy, prolonged use of digital platforms for learning. This condition may worsen because of concentration problems and a loss of motivation after long hours of screen-based study in online classes. These factors also influence students’ mental health [17]. Furthermore, loneliness is the opposite of social resources. Instead of feeling supported and connected, lonely students feel emotionally disconnected from peers and teachers. In post-pandemic studies, it is evident that loneliness is closely associated with anxiety, depression, and lower satisfaction, affecting students’ mental health [50]. Within the JD-R framework, loneliness is a lack of social support, which hinders students’ ability to cope with higher demands, such as academic stress and digital fatigue. The research model based on JD-R is presented in Fig. 1. Research model Methods Research design This study was carried out in the post-pandemic era. The purpose was to examine how the restructuring of educational institutions influences students’ academic tasks in the presence of digital tools. These tools were the product of COVID-19 and evolved to help university students throughout the pandemic. The objective was to examine how academic stress associates with university students’ MHPs in the post-COVID era, with digital fatigue as a mediating factor and students’ loneliness as a moderating factor. The literature supports the view that a cross-sectional survey design is most suitable for exploring the relationships between variables [51]. This design is also useful for understanding relationships among variables in natural settings [52]. Moreover, the literature argues that this design is easy to implement when data are collected from a large group of participants [53]. These characteristics made this design more suitable for the current study. Participants The participants were final-semester students enrolled in four-year Bachelor's programs in top-ranked Pakistani universities. The researchers developed inclusion and exclusion. Only those students were selected who were enrolled during the pandemic years and had the experience of online classes. Later, they returned to their on-campus academic routine during the post-pandemic era. They had both experiences. The selected students used digital tools in both pre- and post-pandemic eras. Moreover, students who got enrolled during the post-COVID era were excluded. The researcher considered the top 12 public and private-sector universities in Pakistan [54]. These universities were selected from the QS-Ranking list of 2025. QS rankings were used as an indicator of institutional quality. The idea was to target only the best-equipped universities to examine how they prepared for students’ MHPs after COVID-19. To select a representative sample of participants, the researchers used multi-method and multi-stage sampling techniques. At the first stage, universities were selected according to the predefined criteria. These universities were contacted through personal contact with the Deans of social and natural sciences via email and WhatsApp. At the second stage, a stratified random sampling was used to identify strata based on university sector (public/private), gender (male/female), and discipline (social/natural). At the third stage, simple random sampling was used to select participants from the already selected universities. The sample was selected through a rigorous process. Overall, 400 participants were invited, and 388 participants (97% response rate) responded to the questionnaire. A power analysis was conducted using G*Power 3.1 to determine the required sample size for detecting a medium effect size (f2 = 0.15) at α = 0.05 and power = 0.95 in a multiple regression model with three predictors. The analysis indicated a minimum required sample of 119 participants. Therefore, the final sample size of 388 exceeded this requirement, ensuring adequate statistical power. Male participants (202, 52%) led the study than their counterparts (186, 48%). Similarly, students from social sciences (210, 54%) participated more than students from the natural sciences (178, 46%). However, there were almost equal numbers of students from public-sector (198, 51%) and private-sector (190, 49%) universities. Instrumentation For the data collection, no such tool was found in the literature that could be used in the current study’s context and variables. The researchers developed a questionnaire after an extensive literature review on these variables in the pre- and post-COVID eras. The researchers collected information regarding their gender, discipline, and university sector. Students’ responses against the study’s four variables were measured on a five-point Likert scale with options ranging from 1 (strongly agree) to 5 (strongly disagree). All the items were developed on this scale, and none of the items were reverse-coded. Moreover, all the interpretations match the coding. Initially, the researchers developed 28 items of PC-USWS (Post-COVID-19 University Students’ Well-being Scale). However, after expert and pilot-tester feedback, they remained 24 (AS = 7, DF = 6, Lo = 5, MHPs = 6). The opinion was taken from three experts in the field and two practitioners. They were all PhDs. Pilot testing was conducted with 40 respondents. The reliability analysis indicated that Cronbach's Alpha values for all sub-scales were above 0.70, the minimum threshold for questionnaire acceptability. The final data were collected online via Google Forms. The Google Form link was shared via WhatsApp and email. The data were collected during July and August 2025. The reliability analysis after final data collection and the tool development process is explained in Table 1. Table 1Development of TP-SGSWB and reliability analysisSr nSub-scalesNαCRSample itemsLiterature review1Academic stress (AS)70.720.73The changes in teaching and assessment after COVID-19 have increased my academic stressI often worry that I will fall behind in my studies if I take even a short break[32, 33, 55–58]2Digital fatigue (DF)60.730.74I find it hard to concentrate during screen-based lectures or tutorialsSpending too much time studying on screens makes me lose motivation to engage with my coursework[17, 19]3Loneliness (Lo)50.760.78Since the COVID-19 pandemic, I feel less connected to my classmates than beforeI rarely have someone from my university community to share my feelings with[40, 42, 43, 45, 59]4Mental health problems (MHPs)60.710.73Worries about my academic future interfere with my sleep or restI feel that stress related to university is affecting my emotional well-being[38, 45, 60, 61] Development of TP-SGSWB and reliability analysis The changes in teaching and assessment after COVID-19 have increased my academic stress I often worry that I will fall behind in my studies if I take even a short break I find it hard to concentrate during screen-based lectures or tutorials Spending too much time studying on screens makes me lose motivation to engage with my coursework Since the COVID-19 pandemic, I feel less connected to my classmates than before I rarely have someone from my university community to share my feelings with Worries about my academic future interfere with my sleep or rest I feel that stress related to university is affecting my emotional well-being To ensure the questionnaire's construct validity, an exploratory factor analysis (EFA) and confirmatory factor analysis (CFA) were conducted on the same dataset due to sample size constraints. This may inflate model fit and is acknowledged as a limitation. EFA was performed using principal axis factoring with Promax rotation, and factors were retained based on eigenvalues greater than 1. CFA was conducted using maximum likelihood estimation. The model showed acceptable fit (χ2[276] = 3754, p < 0.001, CFI = 0.91, TLI = 0.91, RMSEA = 0.05, 95% CL [0.04, 0.06], SRMR = 0.04). All standard factor loadings were statistically significant. The AVE values for each factor were greater than 0.50 and ranged from 0.52 to 0.68, and the CR values ranged from 0.73 to 0.78. The AVE for each variable was higher than its corresponding inter-construct correlations. This satisfies the Fornell-Larcker criterion. Moreover, all HTMT values were below the threshold of 0.85, indicating adequate discriminant validity among the constructs. The factor loading is shown in Table 2. Table 2Factor loadingFactorIndicatorEstimateSEZpASAS10.3690.05386.86 < 0.001AS20.4360.05597.81 < 0.001AS30.6640.053412.44 < 0.001AS40.4880.05818.41 < 0.001AS50.4520.05098.87 < 0.001AS60.3610.05047.16 < 0.001AS70.602051411.70 < 0.001DFDF10.6410.058111.03 < 0.001DF20.6680.048513.77 < 0.001DF30.5310.044811.85 < 0.001DF40.4250.05927.18 < 0.001DF50.4470.05318.41 < 0.001DF60.7030.047014.95 < 0.001LoLo10.5700.045312.59 < 0.001Lo20.4330.05577.79 < 0.001Lo30.5860.05899.95 < 0.001Lo40.2230.05344.16 < 0.001Lo50.6160.049812.38 < 0.001MHPsMH10.4530.04719.62 < 0.001MH20.4790.044610.75 < 0.001MH30.4560.04729.66 < 0.001MH40.2410.06083.96 < 0.001MH50.5910.046412.76 < 0.001MH60.4710.05348.81 < 0.001 Factor loading Table 2 presents the factor loadings for all study constructs. The results indicated that most items had statistically significant loadings (p < 0.001) and met the acceptable retention threshold. For the AS construct, factor loadings ranged from 0.361 to 0.664, and all six items (AS1-AS6) were retained despite relatively lower loadings for AS1 and AS6. For DF, the loadings ranged from 0.425 to 0.703, and all six items (DF1-DF6) were retained. In the Lo construct, the factor loadings ranged from 0.223 to 0.616. Item Lo4 showed a substantially low loading (0.223) and was therefore dropped, while the remaining items were retained. Similarly, for the MHPs construct, factor loadings ranged from 0.241 to 0.591. Item MH4 demonstrated a weak loading (0.241) and was removed from further analysis, whereas the remaining items were retained. Overall, the retained items showed factor loadings ranging from 0.361 to 0.703, indicating acceptable convergent validity for the measurement model. Data analysis The data were analyzed using Jamovi software (version 2.7.26). After collecting data, the data were entered into an Excel sheet and screened for missing values. Then, the researcher used the Shapiro–Wilk test for normality. It indicated that academic stress (W = 0.95, p = 0.06), digital fatigue (W = 0.97, p = 0.07), loneliness (W = 0.98, p = 0.05), and mental health problems (W = 0.96, p = 0.08) did not significantly deviate from normality. Levene’s test confirmed homogeneity of variance across variables (p > 0.05). Multicollinearity diagnostics showed that VIF values ranged from 1.45 to 2.10, indicating acceptable levels. Furthermore, inspection of standardized residual plots suggested that the assumptions of linearity and homoscedasticity were met. The data met the assumptions required for parametric statistics. Then, Pearson correlations, simple linear regression, mediation, and moderation analysis were carried out. As the data were self-reported, there was a possibility of common method bias. To manage this bias, we applied a Harman’s single-factor test using an exploratory factor analysis (EFA). All 24 items from the primary study variables (Academic Stress, Digital Fatigue, Loneliness, and Mental Health Problems) were loaded into a single-factor solution using the minimum residual extraction method with no rotation. The results indicated that the single factor accounted for only 24.90% of the total variance. Because this value is significantly below the 50% threshold, common method bias was not considered a significant threat to the validity or interpretability of the study findings. In addition to Harman’s single-factor test, a single-factor CFA model was estimated to further examine common method bias. The single-factor model showed poor fit to the data (χ2/df = 5.81, CFI = 0.61, TLI = 0.57, RMSEA = 0.14, SRMR = 0.12), compared with the proposed four-factor measurement model. These results suggest that common method bias was unlikely to substantially influence the findings. Research ethics The current study was approved by the ethics committee of the School Education Department vide number (RC-1452/HM/2025). The researcher adhered to the ethical principles outlined in the Declaration of Helsinki and local regulations. Informed consent was obtained in writing. The first section of the questionnaire was about volunteer participation. By ticking the “agree” box, they proceeded with data collection. The data were protected using a password, and only the principal investigator had it. So, the data remained confidential. The participants were not asked any such question that could reveal their anonymity. The questionnaire link was shared online, and university students responded at their convenience. Thus, no physical or psychological harm was involved. The researchers gave students the option to withdraw from the data collection process if they felt any inconvenience. The consent to publication does not apply to the current study, as no participant names or personal information were used. Results After assumption check for parametric statistics, Pearson correlation was used to assess the nature of relationships among the study’s sub-scales. Moreover, descriptive statistics were also applied to examine the role of these sub-scales in data collection. The results are presented in Table 3. Table 3Correlation among constructs (N = 388)Variables1234MSDAcademic stress (AS)–3.690.59Mental health problems (MHPs)0.657***95% CI [0.596, 0.710]-3.680.57Digital fatigue (DF)0.587***95% CI [0.517, 0.648]0.592***95% CI [0.523, 0.653]–3.570.66Loneliness (Lo)0.696***95% CI [0.641, 0.744]0.646***95% CI [0.584, 0.701]0.645***95% CI [0.583, 0.700]–3.640.63***p < 0.001 Correlation among constructs (N = 388) 0.657*** 95% CI [0.596, 0.710] 0.587*** 95% CI [0.517, 0.648] 0.592*** 95% CI [0.523, 0.653] 0.696*** 95% CI [0.641, 0.744] 0.646*** 95% CI [0.584, 0.701] 0.645*** 95% CI [0.583, 0.700] Table 3 shows that academic stress had the highest score (M = 3.69, SD = 0.59), and digital fatigue had the lowest score (M = 3.57, SD = 0.66) on this scale. Higher scores (M = 3.68, SD = 0.57) on the mental health problems scale indicate greater levels of psychological distress. Moreover, all the relationships among the study’s constructs were positive and statistically significant. Academic stress was positively associated with mental health problems (r = 0.657, 95% CI [0.596, 0.710], p < 0.001). It provided the foundation for simple linear regression between predictor and dependent variables. Similarly, digital fatigue was associated with mental health problems (r = 0.592, 95% CI [0.523, 0.653], p < 0.001). This finding supported the mediation analysis. Moreover, digital fatigue was also positively associated with participants’ loneliness (r = 0.645, 95% CI [0.583, 0.700], p < 0.001). The graphical representation of the correlations is presented in Fig. 2. Correlation matrix The first hypothesis was formulated to examine the contribution of university students’ academic stress to their MHPs. After collecting data on both of the constructs, H1 was tested by applying simple linear regression, and the results are provided in Table 4. Table 4Role of academic stress in mental health problemsPredictorEstimateSE95% of confidence intervaltpLowerUpperIntercept1.3210.141.0461.5969.46 < 0.001Academic Stress0.6390.040.5660.71317.12 < 0.001R = 0.66 R2 = 0.43, F(1, 386) = 293, p < 0.001Standardized coefficients (β) are reported for predictors. The intercept is reported as an unstandardized (B) Role of academic stress in mental health problems R = 0.66 R2 = 0.43, F(1, 386) = 293, p < 0.001 Standardized coefficients (β) are reported for predictors. The intercept is reported as an unstandardized (B) Table 4 provides the results for the simple linear regression analysis. The analysis revealed that the overall model was significant (F[1, 386] = 293, p < 0.001). The academic stress explained 43% of the variance in university students’ MHPs (R = 0.66, R2 = 0.43). The effect size was large (f2 = 0.75), indicating that academic stress explains a substantial proportion of variance in mental health problems. Supporting H1, the analysis showed that academic stress significantly relates to students’ MHPs. The intercept was statistically significant, indicating that when academic stress was zero, the predicted MHPs score was (B = 1.321, SE = 0.14, 95% CI [1.046, 1.596], t[386] = 9.46, p < 0.001). Academic stress significantly predicted mental health problems among university students in the post-COVID era (β = 0.639, SE = 0.04, 95% CI [0.566, 0.713], t[386] = 17.12, p < 0.001). A one-unit increase in academic stress was associated with a 0.64-unit increase in students’ mental health problem score. The second hypothesis of the study was designed to examine the mediating role of students’ digital fatigue, resulting from longer screen time, between their academic stress and MHPs. After collecting data on these constructs, H2 was tested using simple mediation with a 95% confidence interval (CI) based on a bootstrap sample of 5000. The results are presented in Table 5. Table 5Mediation estimatesEffectLabelsβSE95% CIZp% MediationLowerUpperIndirecta x b (AS → DF → MH)0.180.040.1070.2614.64 < 0.00130Directc (AS → MH)0.460.060.3330.5807.63 < 0.00170Totalc + a x b0.640.050.4730.73113.53 < 0.001100R2 = 0.51, Standardized coefficients (β) are reported for predictors. The intercept is reported as an unstandardized (B) Mediation estimates R2 = 0.51, Standardized coefficients (β) are reported for predictors. The intercept is reported as an unstandardized (B) Table 5 provides the results for the mediation analysis. Mediation analysis was conducted using the jAMM module. Indirect effects were estimated using bootstrapping with 5000 resamples and bias-corrected (BCa) 95% confidence intervals. Statistical significance was determined based on whether the confidence interval excluded zero. Standardized and unstandardized coefficients were both examined to interpret effect sizes. The indirect effect was significant, ab = 0.18, 95% CI [0.107, 0.261], indicating partial mediation. This indicated that part of the effect of academic stress on students’ MHPs was mediated through their digital fatigue, and it was 30% of the total effect. The direct effect of academic stress on students’ MHPs remained significant after controlling for the effect of digital fatigue (c′: β = 0.46, SE = 0.06, 95% CI [0.333, 0.580], Z = 7.63, p < 0.001). The direct effect was 70% of the total effect. The overall mediation model explained 51% of the variance in mental health problems. The standardized indirect effect (β = 0.18) indicates a moderate mediating effect of digital fatigue between academic stress and mental health problems. The results partially support H2 and show that digital fatigue partially mediated the relationship between students’ academic stress and MHPs in the post-COVID era. The path estimates of structural model is presented in Table 6. Table 6Path estimatesEstimateSE95% Confidence intervalZpLowerUpperAS → DF0.6600.04620.5690.75114.27 < 0.001DF → MH0.2720.03850.1970.3487.07 < 0.001AS → MH0.4600.04330.3750.54510.62 < 0.001 Path estimates Table 6 presents the path estimates for the structural model. The results indicated that AS had a significant positive association with DF (β = 0.660, SE = 0.046, 95% CI [0.569, 0.751], Z = 14.27, p < 0.001). DF also showed a significant positive relationship with MH (β = 0.272, SE = 0.039, 95% CI [0.197, 0.348], Z = 7.07, p < 0.001). In addition, AS showed a significant positive relationship with MH (β = 0.460, SE = 0.043, 95% CI [0.375, 0.545], Z = 10.62, p < 0.001). Since all confidence intervals excluded zero and all p-values were below 0.001, the findings supported the statistical significance of all proposed paths in the model. The pictorial representation of the mediational role of digital fatigue in the relationship between academic stress and MHPs is presented in Fig. 3. Mediation model with standardized coefficients (β) The third hypothesis was created to examine the moderating role of students’ loneliness in classes in the post-COVID era between their academic stress and MHPs. After collecting data on these variables, H3 was tested using a simple moderation analysis with a 95% confidence interval and a bootstrap sample of 5000. The results are provided in Table 7. Table 7Moderation EstimateVariablesBSE95% CIZpLowerUpperAcademic STRESS0.380.070.2530.5175.68 < 0.001Loneliness (Lo)0.340.060.2220.4465.91 < 0.001AS*Lo0.110.050.0080.2152.16 < 0.05 Moderation Estimate Table 7 provides the analysis for H3. All variables were mean-centred before analysis. The results indicated that academic stress has a significant effect on students’ MHPs (βAS = 0.38, SE = 0.07, 95% CI [0.253, 0.517], Z = 5.68, p < 0.001). Similarly, students’ loneliness also has a significant effect on students’ MHPs (βLo = 0.34, SE = 0.06, 95% CI [0.222, 0.446], Z = 5.91, p < 0.001). Moreover, their interaction effect (AS*Lo) also showed a significant effect on university students’ MHPs (βAS*Lo = 0.11, SE = 0.05, 95% CI [0.008, 0.215], Z = 2.16, p < 0.05). Although the interaction effect was statistically significant, its magnitude was small. This suggests that loneliness plays a limited moderating role in practical terms. The interaction term contributed a small increase in explained variance (ΔR2 = 0.01), indicating a modest moderation effect. The positive combined effect shows that the effect of academic stress on students’ MHPs becomes stronger at higher levels of loneliness. This was consistent with the idea that students experiencing greater loneliness are more vulnerable to the negative impact of academic stress on students’ MHPs. The simple slope analysis is presented in Table 8. Table 8Simple slope analysisEstimateSE95% of confidence intervalZpLowerUpperAverage0.380.070.2500.5155.63 < 0.001Low (−1SD)0.310.070.1660.4544.31 < 0.001High (+ 1SD)0.450.080.3050.6087.70 < 0.001 Simple slope analysis Table 8 shows the simple slope analysis and shows that the relationship between academic stress and mental health problems was stronger at higher levels of loneliness (+ 1SD), B = 0.45, SE = 0.08, 95% CI [0.305, 0.608], compared to the mean level (average), B = 0.38, SE = 0.07, 95% CI [0.250, 0.515]. it is weaker at the lower levels of loneliness (-1SD), B = 0.31, SE = 0.07, 95% CI [0.166, 0.454]. These findings suggest that students experiencing higher loneliness show a stronger positive association between academic stress and mental health problems. To further probe the significant interaction between academic stress and loneliness, a Johnson–Neyman (J–N) technique was applied to identify the regions of significance. The analysis revealed that the relationship between academic stress and mental health problems was conditioned by levels of loneliness. Specifically, the positive association between academic stress and mental health problems was statistically significant only when loneliness scores were greater than 1.92 (see Fig. 4). For individuals with loneliness scores below this threshold, academic stress did not significantly predict mental health problems (p ≥ 0.05). The Johnson-Neyman plot is presented in Fig. 4. Johnson–Neyman Plot Discussion The present study provides empirical evidence on factors affecting students’ MHPs in the higher education system of a developing country in the post-COVID era. This study was guided by a research objective that had three hypotheses. The first hypothesis aimed to investigate the association of students’ academic stress with their MHPs. The results revealed that students’ MHPs were significantly associated with academic stress related to various academic tasks. These results support the previous studies, which mentioned that changes in higher education after COVID-19 continue to be associated with students’ mental health problems, with the introduction of digital gadgets and online systems for various academic tasks [30–32]. They emphasized that the integration of online tools has significantly associated with students’ MHPs, which were further linked to their overall physical and psychological health [34]. These results imply that the academic stressors are high job demand factors, which, according to the JD-R Theory, put extra stress on students. This situation demands that students be trained in the moderate use of online systems. It can also be used for recreational activities to improve students’ MHPs. The second hypothesis examined the mediating role of digital fatigue in the relationship between university students’ academic stress and their MHPs. The analysis revealed that digital fatigue partially mediated the mentioned relationship and accounted for 30% of the total variance. However, this effect was statistically significant in nature. This shows that digital fatigue showed a significant association with students’ MHPs. These results were consistent with the findings of the previous studies that explored that digital fatigue and burnout are associated with students’ poor mental and physical health [16, 36–38]. This shows that if students keep on continuously using screens, it will increase their mental health challenges. Thus, students need to use the digital tools wisely with sufficient breaks. This study extended the implications of JR-D theory in a technological strain demand of digital fatigue. The study also explored the moderating effect of loneliness on the relationship between students’ academic stress and MHPs. The third hypothesis was generated because students enrolled in COVID years had little interaction and were not social even with their classmates. Education is a social process, and students felt lonely, which increased their MHPs. The analysis revealed that loneliness was significantly associated with students’ MHPs at both independently and in and in interaction with academic stress. These findings support the results of previous studies, which concluded that during COVID-19, students’ loneliness adversely influenced their well-being, sense of belonging, and led them to anxiety, depression, and suicidal ideation [29, 40, 41]. This implies that if students are left alone to cope with their MHPs, this will adversely affect their lifestyle and well-being. JD-R theory also highlights similar challenges faced students by demanding them to be up to the mark. At the university level, teachers and peers can assist such students by fostering their participation in social events. Universities are also supported to provide entertaining opportunities to improve their well-being by reducing the effects of loneliness. Conclusion This study aimed to understand the various challenges associated with students’ mental health in the post-COVID era at the university level. The focus was to investigate the role of academic stress on students’ MHPs in post-pandemic years, with the mediating role of digital fatigue and the moderating role of loneliness on said relationship. The current survey study was guided by Job-Demands Resources (JD-R) theory. The analysis revealed that students’ academic stress significantly relates to MHPs in the post-pandemic era, too. This indicates that high study demands can be a significant challenge to students’ overall well-being and MHPs. Students who reported higher academic stress reported more anxiety, low mood, overload, depression, and emotional drain. Digital fatigue is another construct that relates to students’ MHPs. The analysis further supported that it partially mediates, with 30% of the total effect on the relationship between university students’ academic stress and MHPs. This shows that greater screen time and technology-related fatigue were associated with poorer mental health outcomes. From the JD-R theory’s perspective, digital fatigue is an agent that drains students’ energy and influences their mental processes. Students’ loneliness was found to be a significant predictor of their MHPs at the individual level and in combination with their academic stress as a moderator. This also aligns with the JD-R theory, which argues that when social resources are weak, demands are more damaging. The present findings indicates that academic stress alone cannot predict students’ MHPs in the post-COVID era. However, digitally mediated fatigue and the absence of social connection together shape how academic demands influence students’ psychological well-being. Implications The present study had several implications for policy, practice, research, and theory. Academic stress, digital fatigue, and loneliness are closely interconnected constructs, and this must be understood at the university level. Thus, universities are suggested to design policies with a reasonable set of expectations for workload, assessment, and online class participation to minimize students’ mental health and well-being issues. University administrations are also suggested to design a policy for uniform support and counselling facilities, ensuring peer support programs. For practical implications, the analysis indicates that students’ support services, academic advisors, and faculty need to monitor students’ academic and digital workload. A better balance between academic and digital demands is needed. Digital fatigue can be decreased by reducing online sessions, webinars, and videoconferences. This can be done by providing flexible assessment schedules. The current study examined a snapshot of the situation through a cross-sectional study. However, the long-lasting impact can be seen through longitudinal studies. Thus, longitudinal studies are recommended to examine the interplay among the variables in the current study and university students’ mental health. The present study extended the implications of JD-R theory by investigating the digitally mediated demands and deficits in social resources of a developing country’s higher education system. We suggest that this theory must also be tested in the context of technology-related demands and appropriate resources in developed countries’ higher education systems. Limitations and future directions Alongside these significant implications, the current study also had several limitations. For example, the researchers collected cross-sectional data using a self-reported questionnaire, which limits the ability to draw strong conclusions and may be associated with common method bias. However, this limitation can be addressed by designing longitudinal studies that collect data using various other tools as well. Another limitation was that the data were collected from university students, which may not reflect results for other student groups. This limitation can be addressed by designing future studies that include participants from schools and colleges. The third limitation was that the model focused on three predictors of MHPs by neglecting others, such as financial strain, sleep quality, and social support. This limitation can be addressed by designing future studies that also explain the JD-R theory by integrating other variables. Another limitation was that the study used a voluntary online survey. Therefore, self-selection bias cannot be completely ruled out. Students who were more interested in mental health issues or who experienced greater psychological stress may have been more likely to participate. This may limit the generalizability of the findings. Future research may use additional data collection methods to overcome self-selection bias. Acknowledgements The study was supported by the Deanship of Scientific Research at King Khalid University, which funded this work through Large Research Groups under grant number (RGP 2/412/46) Author contribution A.I.A. conceived and designed the study, performed the data analysis, and wrote the main manuscript text. S. assisted in data interpretation and manuscript editing. S.F. carried out data analysis and addressed the additional concerns raised by reviewers at the revision stage. P.M. contributed to the research methodology. M.A.S.K. contributed to methodological refinement and statistical validation. T.B.S.A-K. reviewed the theoretical framework and revised the manuscript for intellectual content. All authors reviewed and approved the final version of the manuscript. Funding The study was supported by the Deanship of Scientific Research at King Khalid University, which funded this work through Large Research Groups under grant number (RGP 2/412/46). Data availability The data is available and can be provided upon reasonable request. Declarations Ethical approval and consent to participate The ethics protocol was approved by the ethics committee of the School Education Department, Punjab, Pakistan, in accordance with the Declaration of Helsinki. The protocol was approved vide number (RC-1452/HM/2025). Consent for publication Not applicable. Informed consent Informed written consent was obtained from participants before data collection. This participation was voluntary in nature. Competing interests The authors declare no competing interests. Footnotes Publisher's Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The data is available and can be provided upon reasonable request.