Within the COM-B model, capability includes physical/psychological status and knowledge/skills (18), serving as the key to behavioral execution. College students are the main group accessing online health information, which is a key factor associated with their healthy lifestyles. eHealth literacy—defined as the core capability to evaluate and apply such information—is directly correlated with the scientific validity and sustainability of their healthy lifestyles (18). The operating mechanism of this capability can be explained by the information processing process revealed by the Bayesian Mind sponge Framework (BMF) (19). This framework points out that when individuals encounter online health information, they evaluate it against their existing health-related cognitive-behavioral schemata, subsequently engaging in selective processing: information consistent with scientific evidence is “absorbed” and integrated, whereas conflicting or unsubstantiated information is “filtered out” (19). The level of eHealth literacy determines the efficiency of this information processing. Individuals with high eHealth literacy can establish scientific cognitive benchmarks, accurately identify valuable content during information screening, and translate it into health behaviors, with research showing significantly higher correct health behavior rates among this group (20). Therefore, as the core manifestation of the capability dimension in the COM-B model, the role of eHealth literacy is realized precisely through the mind sponge mechanism of the BMF. Together, they construct a comprehensive capability system that bridges cognitive processing and behavioral transformation.

Opportunity includes the environmental and social factors that facilitate a behavior (21). Social support, as a key opportunity factor, systematically creates favorable conditions for college students to practice healthy behaviors through three levels: interpersonal networks, learning and living environments, and information environments (22). At the interpersonal network level, emotional support from family members, friends and teachers is the foundation. The care from family members and the encouragement from peers can enhance students’ psychological motivation to adhere to healthy behaviors (23, 24). At the level of the learning and living environment, as the main venue for daily activities, the institutional and organizational support provided by universities is of vital importance. A supportive campus environment can significantly enhance students’ willingness and possibility to adopt a healthy lifestyle. Studies have shown that a well-planned campus-built environment is conducive to college students’ physical exercise behaviors (25). At the information environment level, health information platform services help students form scientific cognition and promote the generation of healthy behaviors (26).

Motivation involves the brain processes that energize and direct behavior, encompassing emotional responses, habitual patterns, and analytical decision-making (21). To gain a deeper understanding of the formation mechanism of motivation, this study introduces the value theory and draws on the Granular Interaction Thinking Theory (GITT). Value theory (27) holds that the formation of value is an information processing process aimed at reducing psychological uncertainty. This process depends on GITT, which is developed with the psychological mechanism described by MT as its most fundamental conception and starting point, and MT is also the core mechanism on which BMF relies (28). GITT believes that individuals transform various influencing factors into different attributes of “information granularity” in behavioral decision-making (29). Through the dynamic interaction of multi-granularity information, they re-evaluate and integrate their own values, ultimately forming stable and behavior-oriented motivations. Social responsibility, including individual health responsibility and public health responsibility, plays an important role in promoting healthy lifestyle. Individual health responsibility is the basis and premise of healthy behavior (30), while public health responsibility contributes to the formation of individual and collective health behavior and the formation of healthy lifestyle (31). This motivation perspective based on value theory reveals the dynamics and situational dependence of motivation formation, providing a new theoretical perspective for understanding the persistence mechanism of healthy behaviors.

In the above theoretical framework, in addition to the core components, factors including students’ academic major, mental health, and sense of hope also play a key moderating role, and are closely correlated with the transformation efficiency among capability, opportunity, and motivation. Academic major functions as a moderator in the relationship between capability and behavior. The knowledge structure and thinking mode shaped by students’ academic majors form a cognitive framework that is directly linked to the efficiency of translating eHealth literacy into practical behaviors (32). Mental health status regulates the relationship between opportunity and behavior. Mental health levels can promote the formation of self-efficacy (33), enabling individuals to independently cope with various challenges. Even in the absence of social support, they can maintain a healthy lifestyle by leveraging their psychological advantages (34). Sense of hope is an important moderating variable between motivation and behavior. As a positive psychological trait facing the future, high hope can enhance individuals’ confidence and motivation belief in the path to achieve health goals, thus significantly strengthening the driving effect of reflective motivation on health behavior (35). When individuals have firm expectations for future health outcomes, their willingness and perseverance to adhere to a healthy lifestyle will be significantly improved.

Therefore, this study adopts the COM-B model as the core framework and integrates BMF and value theory for theoretical enhancement, constructing a comprehensive model of the formation mechanism of healthy lifestyles among college students (Figure 1). This model not only systematically examines the driving effects of eHealth literacy (capability), social support (opportunity), and social responsibility (motivation) on college students’ healthy lifestyles, but also reveals the moderating mechanisms of major, sense of hope, and mental health in the above pathways. This research is expected to provide an integrated theoretical perspective for understanding the complex causes of college students’ health behaviors and offer empirical evidence for developing precise health promotion intervention strategies.

A cross-sectional survey design was adopted. Methodologically, this study aimed to be an observational study that describes the current state of healthy lifestyles among college students and, more importantly, tests the specific mediating and moderating mechanisms proposed within the COM-B framework.

The participants were 4,036 undergraduate students from 6 universities in Shandong Province. The inclusion criteria were: (1) full-time undergraduate students (2) voluntary participation. Exclusion criteria comprised individuals with serious health conditions that prevented engagement in normal physical exercise and other basic daily activities.

From May to June 2022, by using the stratified cluster sampling method, six universities were selected in Shandong Province based on the nature of their disciplines: 2 comprehensive universities (Qingdao University and Ludong University), 2 science and engineering universities (Qingdao University of Technology and Weifang Institute of Technology), and 2 medical universities (Weifang Medical University and Binzhou Medical University). First, randomly select four majors from each university; Secondly, students are stratified by grade. One class is randomly selected from each grade, and a questionnaire survey is conducted among all the undergraduate students in the selected class (with a total of 127,985 students from 6 universities, and a sampling ratio of 3% is adopted (53), the overall sample size should be no less than 3,840 people). A total of 4,320 questionnaires were distributed, among which 4,036 were valid, with an effective rate of 93.4%.

The survey was conducted anonymously through an online platform (wenjuanxing), with each IP address restricted to a single submission. Additionally, questionnaires with missing basic respondent information, a response time of less than 120 s, or an item missing rate of ≥5% were excluded (61).

Data processing and analysis were performed using SPSS version 27.0. The normality test of the data showed that the eHealth literacy, social support, social responsibility, mental health, sense of hope and healthy lifestyles failed to pass the normality test, but the overall distribution was relatively symmetrical, the degree of deviation from the normal distribution was relatively small, and the sample size in this study was large (62, 63). Therefore, in this study, the data were analyzed as if they were in accordance with normal distribution. Continuous variables conforming to normal distribution were described by Mean ± SD, and categorical variables were described by N (%). Spearman’s correlations were used to evaluate the associations between social support, social responsibility, eHealth literacy, mental health, sense of hope and college students’ healthy lifestyles, with the correlation coefficient denoted by the symbol ρ (rho). We utilized the SPSS PROCESS macro developed by Hayes (64) to examine the mediating role of social responsibility in two pathways: (1) between eHealth literacy and healthy lifestyle, and (2) between social support and healthy lifestyle. For this mediation analysis, we employed the bootstrap method with 5,000 repeated samplings and calculated 95% confidence intervals. Results were deemed statistically significant if the 95% CI did not include zero. Subsequently, we used the same SPSS PROCESS macro (64) to assess the moderating effects of major, mental health, and hope. A statistically significant regression coefficient for the interaction term (p < 0.05) was taken as evidence of a moderation effect.

This study was approved by the Medical Research Ethics Committee of Shandong Second Medical University (IRB approval no. 2023YX116). Informed consent has been obtained from all subjects and their legal guardians. This study adhered to the Declaration of Helsinki to this effect.

Table 1 provides the basic information about participants. Among the 4,036 respondents included in this study, 1,811 (44.9%) were male and 2,225 (55.1%) were female. There were 2,216 (55.0%) lower-grade students and 1,820 (45.0%) upper-grade students. In terms of major, 991 (24.6%) were medical majors and 3,045 (75.4%) were non-medical majors. Additionally, 1,312 (32.5%) participants were single-child, and 2,724 (67.5%) were not. The average scores of eHealth literacy, social support, social responsibility mental health, sense of hope and healthy lifestyles were 4.25 ± 0.83, 3.92 ± 0.76, 4.37 ± 0.76, 3.36 ± 0.76, 4.25 ± 0.74, 3.95 ± 0.74, respectively. Female, lower-grade, non-medical major, and only-child students had higher healthy lifestyle scores. To clarify the distribution characteristics of the sample across core research variables, this study adopted the equidistant trisection method with balanced adjustment to divide the total scores of the six scales into three levels (low, medium, and high), and calculated their frequencies and percentages (see Table 2). The proportion of participants with a high level of Healthy Lifestyles reached 71.2%, indicating that the majority of respondents had a relatively high level of healthy lifestyles.

The results of the correlation analysis are shown in Table 3. Spearman correlation analysis showed that eHealth literacy, social support, social responsibility was all positively correlated with healthy lifestyles (ρ = 0.417, p < 0.01; ρ = 0.709, p < 0.01; ρ = 0.755, p < 0.01, respectively). A positive correlation was also observed between eHealth literacy and social responsibility (ρ = 0.431, p < 0.01). Furthermore, social support showed a significant positive correlation with social responsibility (ρ = 0.754, p < 0.01).

Covariance diagnostics for the four variables (eHealth literacy, social support, social responsibility, and healthy lifestyles) indicated that the inflation factors were all below 10 (VIF: 1.390–2.367), and the tolerances ranged from greater than 0.1 to less than1 (Tol: 0.423–0.720), confirming the Absence of serious multicollinearity among the variables (see Table 4).

The multiple linear regression analysis revealed that eHealth literacy (β = 0.032, p < 0.01), social support (β = 0.380, p < 0.001), and social responsibility (β = 0.449, p < 0.001) were all significant positive predictors of healthy lifestyles. Among these, social responsibility demonstrated the highest standardized regression coefficient (β = 0.449).

To elucidate the underlying relationships among social responsibility, eHealth literacy, social support, and healthy lifestyles, two separate mediation analyses were conducted, controlling for demographic characteristics including gender, grade, and single-child status. The first model examined social responsibility as a mediator between eHealth literacy (independent variable) and healthy lifestyles (dependent variable). The second model tested social responsibility as a mediator between social support (independent variable) and healthy lifestyles (dependent variable). The mediation analysis was conducted using the PROCESS macro for SPSS.

The mediating model of social responsibility between eHealth literacy and healthy lifestyle shows that the total effect of eHealth literacy on healthy lifestyle is 0.356 (p < 0.01), and the 95% confidence interval is (0.331, 0.382). eHealth literacy can positively predict social responsibility, with a path coefficient of 0.340 (p < 0.01). Social responsibility can positively predict a healthy lifestyle, with a path coefficient of 0.660 (p < 0.01). The indirect effect of eHealth literacy on a healthy lifestyle is 0.225, with a 95% confidence interval of (0.202, 0.247), indicating that the mediating effect is valid. After adding the mediating variable (social responsibility), the direct effect of eHealth literacy on a healthy lifestyle was 0.131 (p < 0.01), with a 95% confidence interval of (0.111, 0.151), indicating that social responsibility plays a partial mediating role in the relationship between eHealth literacy and a healthy lifestyle. Please refer to Tables 5, 6 and Figure 2 for details.

The mediation model of social responsibility between social support and a healthy lifestyle shows that the total effect of social support on a healthy lifestyle is 0.694 (p < 0.01), and the 95% confidence interval is (0.672, 0.715). Social support can positively predict social responsibility, with a path coefficient of 0.697 (p < 0.01). Social responsibility can positively predict a healthy lifestyle, with a path coefficient of 0.444 (p < 0.01). The indirect effect of social support on a healthy lifestyle is 0.309, with a 95% confidence interval of (0.281, 0.337), indicating that the mediating effect holds. After adding the mediating variable (social responsibility), the direct effect of social support on a healthy lifestyle was 0.384 (p < 0.01), with a 95% confidence interval of (0.358, 0.411), indicating that social responsibility plays a partial mediating role in the relationship between social support and a healthy lifestyle. Please refer to Tables 5, 6 and Figure 3 for details.

The moderation analysis was conducted using the PROCESS macro for SPSS, controlling for gender, grade, and single-child status.

In the relationship between eHealth literacy and healthy lifestyles (Table 7; Figure 4), eHealth literacy was a significant positive predictor of healthy lifestyles. The interaction term between eHealth literacy and major was statistically significant (β = −0.127, p < 0.001), indicating a significant moderating effect of major. Simple slope analysis revealed that while eHealth literacy positively predicted healthy lifestyles among both non-medical majors (β = 0.385, p < 0.001) and medical majors (β = 0.259, p < 0.001), this positive effect was significantly stronger in the non-medical student group.

Social Support and Healthy Lifestyles (Table 8; Figure 5) demonstrated that social support was a significant positive predictor of healthy lifestyles The interaction term between social support and mental health was statistically significant (β = −0.087, p < 0.001). Simple slope analysis indicated that while social support positively predicted healthy lifestyles among both students with low (β = 0.763, p < 0.001) and high (β = 0.615, p < 0.001) levels of positive mental health, this effect was significantly stronger in the group with lower psychological well-being.

Regarding the relationship between Social Responsibility and Sense of Hope on Healthy Lifestyles (Table 9; Figure 6), both variables emerged as significant positive predictors. A statistically significant interaction was found between social responsibility and sense of hope (β = 0.040, p < 0.001). Simple slope analysis revealed that the positive effect of social responsibility on healthy lifestyles was significantly stronger among students with a high sense of hope (β = 0.717, p < 0.001) compared to those with a low sense of hope (β = 0.602, p < 0.001).

This study employed a revised scale developed by Jiao et al. for measurement purposes (54). The scale encompasses multiple dimensions, systematically and comprehensively covering core domains of healthy living, thereby providing a holistic reflection of individuals’ overall healthy lifestyle levels. The average healthy lifestyles score among the 4,036 participants was 3.95 ± 0.74, indicating a moderately high level overall. Significant disparities were observed across different demographic characteristics. Specifically, female students, those in lower grades, single-child, and students from urban areas generally reported higher levels of healthy lifestyles. Female students scored significantly higher than male students, likely because they pay more attention to diet management and regular sleep schedules, consistent with previous research (65). Lower-grade students demonstrated significantly better healthy lifestyles than higher-grade students, a finding that contrasts with some previous studies (66). This observed difference may be attributed to the heightened academic and future-oriented pressures encountered by upper-grade students in Shandong Province. Renowned for its competitive educational environment, Shandong places a strong emphasis on academic achievement and post-graduation prospects. As college students progress to higher grades, they face escalating demands from major-specific coursework, internships, thesis writing, and career planning, which can diminish the time and energy available for health-promoting behaviors. Consequently, the decline in healthy lifestyles among upper-grade students appears to be shaped by the distinctive educational culture and structural pressures characteristic of this region. The single-child tend to have superior healthy lifestyles, likely because they often receive more concentrated resource investment and emotional concern within their families, fostering better health cognition and behavioral patterns (12). There are differences in the healthy lifestyles of college students with different gender, grade, and family type, so we should pay more attention to male students, upper-grade students, and non-only-child students.

This study indicates that eHealth literacy, social support, and social responsibility all show a significant positive correlation with college students’ healthy lifestyles, among which social responsibility exhibits the strongest correlation, thus providing support for Hypothesis 1.

Under the framework of the COM-B model, eHealth literacy can be categorized as a capability factor. According to Guo et al., such capability involves the access, appraisal, and application of online health information (55). For college students, a higher level of eHealth literacy is reflected in their ability to effectively obtain health information from electronic media, to scientifically evaluate its credibility and applicability, and to a certain extent, effectively apply such information in addressing practical health problems (67). Under the framework of the COM-B model, social support belongs to the opportunity factor. According to Lou et al., it encompasses three dimensions: interpersonal networks, learning and living environments, and information environments (57). From our perspective, these elements create favorable conditions for college students to engage in health behaviors. Care and support from family members and peers are correlated with college students’ adherence to healthy behaviors (23, 24). A well-planned campus built environment is conducive to college students’ physical activity participation, while health information platform services are associated with students’ development of scientific health cognition, both of which are associated with the formation of healthy lifestyles (25, 26). Under the framework of the COM-B model, social responsibility is categorized as a motivation factor. Particularly within the Chinese sociocultural context, its connotation often extends beyond the scope of individual health management and is deeply integrated with traditional collectivist values. It encompasses individuals’ health responsibilities and public health responsibilities toward their families, communities, and even the nation. Zhu notes that increased awareness of social responsibility is associated with greater engagement in responsible practices, and our study also found that social responsibility is correlated with the adoption of healthy lifestyle behaviors (58). Unlike Western individualistic culture, the sense of social responsibility nurtured by collectivist education in China carries a deeper meaning: safeguarding one’s own health and that of others, as well as upholding collective interests. Based on current evidence, it is plausible that individuals with a strong sense of social responsibility may show a greater tendency to adopt preventive health behaviors and other relevant measures (31, 32).

Our study found that among the various factors linked to college students’ adoption of healthy lifestyle behaviors, social responsibility plays a relatively central and prominent role. From our perspective, its essence can be illuminated through Value Theory (27) and GITT. Social responsibility represents a final product of the entropy reduction process within an individual’s psychological world—a transition from “chaos and disorder” to “stability and order”—which is achieved by absorbing and integrating diverse external information through dynamic interactions (29). Interpretatively, this may suggest that eHealth literacy and social support’s regulation of information flow in digital and social environments correlates with the facilitation of this entropy reduction process. Specifically, eHealth literacy can serve as an accurate tool for information screening and integration, effectively identifying scientific health information and preventing cognitive confusion. On the other hand, social support establishes external resources via interpersonal networks, campus environments, and other channels, which appears to be conducive to accelerating the internalization of a sense of social responsibility. Collectively, these two factors are correlated with the reduction of psychological entropy, endowing social responsibility with a relatively stable and sustainable capacity that is associated with the facilitation of healthy behaviors, thereby positioning it as a core driver related to guiding college students to adopt healthy lifestyles.

The results showed that social responsibility played a mediating role between eHealth literacy, social support, and college students’ healthy lifestyles, forming two synergistic pathways and hypotheses 2a and 2b were supported. High eHealth literacy not only enriches students’ health knowledge but also strengthens their awareness of health obligations (68), enabling them to recognize the potential relevance of personal health to their families, schools, and society at large. Our study shows that eHealth literacy is associated with positive correlations on healthy lifestyle behaviors by enhancing students’ sense of social responsibility, which partially corroborates the pathway of “effectiveness of capability → motivation → behavior.” Social support not only provides direct material and emotional assistance (69) but also fosters a sense of belonging and collective identity (70). Our study also shows that social support is correlated with the facilitation of healthy lifestyle formation by stimulating students’ sense of social responsibility, verifying the pathway of “effectiveness of opportunity → motivation → behavior.”

The results indicated that the positive correlation between eHealth literacy and college students’ healthy lifestyles varies by academic major, meaning this association is moderated by students’ academic disciplines. Specifically, the correlation between eHealth literacy and healthy lifestyles is more pronounced among non-medical students, and this finding provides empirical support for Hypotheses 3a and 3b. The underlying reason is that medical students, having received systematic curriculum training, typically possess a relatively comprehensive health knowledge system and practical skill set (71). Consequently, as a supplementary source of external information, eHealth literacy has a relatively limited incremental value for medical students, with its marginal utility showing a diminishing trend.

The study indicated that mental health level significantly moderates the relationship between social support and healthy lifestyles: the positive correlation between social support and healthy lifestyles shows a decreasing trend with the improvement of mental health level, which provides robust empirical support for Hypotheses 4a and 4b. College students with a high level of mental health exhibit strong emotion regulation ability, and their positive attitudes, psychological resilience, subjective well-being, and harmonious interpersonal relationships form a stable internal support system (59). Such students demonstrate a high level of autonomy in maintaining healthy behaviors and a low dependence on external support (33, 34). In contrast, those with a low level of mental health have vague health cognition and poor behavioral execution, and they are more reliant on external social support. When external resources are insufficient, they are more likely to engage in health risk behaviors (72).

The study demonstrated that as the level of hope increases, the positive correlation between social responsibility and healthy lifestyles becomes increasingly stronger, which partially validates Hypotheses 5a and 5b. Hope consists of two components: Dynamic thinking drives individuals to persistently pursue their goals, while pathway thinking involves the ability to formulate specific plans and strategies for achieving these goals (51). Individuals with high hope typically possess clear action pathways and strong intrinsic motivation, enabling them to internalize social responsibility as a motivation for healthy behaviors and amplify the positive correlation between social responsibility and healthy lifestyles (35). In contrast, those with low hope often lack clear health goals and confidence in achieving them (73). In such cases, social responsibility serves as an important external driver for maintaining healthy habits.

This study has several limitations. First, owing to human resource and geographical constraints, participants were primarily undergraduate students in Shandong Province, which may compromise the generalizability of the findings to some extent. Second, given the cross-sectional design adopted in this study, it is difficult to establish definitive causal relationships among the research variables. Third, as all data were collected via self-report at a single time point, common issues of methodological measurement bias may exert a certain association with on the research results.

The research results provide the following targeted intervention paths for the construction of this system: Firstly, with motivation as the central focus, cultivating social responsibility should be positioned as the core strategy. Values-based education should guide students to internalize health behaviors as conscious actions benefiting both themselves and others. Secondly, when enhancing eHealth literacy, the moderating effect of academic discipline must be considered, with tailored foundational courses on health information discernment and application designed specifically for non-medical students. Furthermore, institutions should proactively develop and optimize campus social support networks while intentionally guiding students to gain positive experiences within these systems. This approach strengthens their sense of social responsibility and transforms supportive environments into sustainable drivers of health behaviors. Finally, the fundamental roles of sense of hope and mental health should be emphasized. Through positive psychology interventions and mental health services, students’ psychological capital can be enhanced, ensuring they possess the intrinsic resilience to maintain healthy behaviors when confronting challenges.

The core theoretical contribution of this study lies in its introduction of the Bayesian Mind sponge Framework and the Granular Interaction Thinking Theory, which significantly deepens and expands the theoretical connotation and explanatory power of the COM-B model. Specifically, the introduction of BMF [20]reveals the intrinsic mechanism of capability (eHealth literacy) from the perspective of information processing, elucidating how individuals filter, absorb, and integrate health information within complex online information environments based on their cognitive frameworks. The integration of Value Theory (27) and GITT (29), furthermore, explains the dynamic construction process of values from the perspective of motivation formation, emphasizing that through continuous interaction with multi-granular information in the environment, individuals constantly reassess and integrate their own values, ultimately forming stable, behavior-oriented motivations (social responsibility). This integrated framework connects the macro-structure of the COM-B model with micro-level cognitive-value mechanisms, providing a more comprehensive and dynamic analytical tool for systematically understanding and precisely intervening in complex health behaviors.

Based on the systematic findings of this study, universities and public health policy makers should shift from the current single-dimensional intervention model to a multi-level and systematic health promotion system construction.

Future research could be further developed in the following directions: First, longitudinal study designs could be adopted to capture the causal relationships in the formation process of healthy lifestyles. Second, in terms of research content, further exploration could be conducted on the moderating effects of other psychological traits within the COM-B model, while examining the generalizability of the integrated theoretical model proposed in this study to other populations. Third, regarding theoretical integration, attempts could be made to engage the integrated framework of this study in dialogue with more diverse theoretical perspectives, thereby further enriching our understanding of the complexity of healthy lifestyles.