Typically, human behavior follows a systematic and purposeful pattern. Prior to decision-making, individuals frequently engage in contemplation and planning (22). Within the C-A-C model, the interplay and influence between individuals’ cognition and emotions in the course of an activity are well elucidated, along with how this relationship culminates in the eventual occurrence of behavior (23). For instance, the process of consumer behavior entails three stages: cognitive (thinking), affective (feeling), and conative (doing). When consumers engage in purchasing goods, they actively seek information about the products (advertisements, store descriptions, etc.). This product information, in turn, influences their anticipations and emotional shifts, ultimately steering their purchasing decisions (24). Over time, the C-A-C framework continuously validates the process of individuals’ cognition, emotions, and behavior. Numerous scholars have expanded upon the C-A-C theory to encompass related theoretical models, such as the ECM (21), the Technology Acceptance Model (TAM) (25), and the American Customer Satisfaction Index Model (26). These expansions seek to elucidate the relationships among cognition, emotions, and behavior across various life contexts. Numerous studies have confirmed the applicability of the C-A-C framework in diverse industry domains. For instance, in the realms of business and service, the model verifies the interplay between consumer-perceived brand value and consumption behavior (27). The psychological and consumption impacts of IT product factors on consumers have been underscored (28). Perceived value, attitudes (satisfaction and trust), and their influences on consumer purchasing intentions have been explored in the context of tourist destinations (29). In the domains of online information and social interactions, the C-A-C framework has similarly demonstrated associations between users’ perceived value, loyalty, and continuous usage intentions toward information (30).

Building upon these foundations, this study adopts the C-A-C framework as its base theoretical foundation to construct a theoretical framework that delineates the sustained usage intentions of users toward health science popularization videos. This framework aims to illuminate the inherent mechanisms linking the content and format of health science popularization videos with users’ perceived value and their intentions for continued usage.

The expectation confirmation theory (ECT) was originally employed to scrutinize and comprehend consumer behavior, effectively delving into the outcomes arising from the alignment of consumers’ pre-purchase expectations with post-purchase performance. It delineates how this alignment influences the relationship between consumer satisfaction and consumption behavior (31). As a result, the theory has found extensive application in diverse marketing domains, including the fast-moving consumer goods industry (32, 33) and service sector (34, 35). As the theory extended its scope to the realm of information systems, the ECM was formulated and applied to explore users’ intentions for continued usage of information system products/services (36, 37).

The ECM is a useful approach for examining intricate information systems, especially for understanding how new factors impact users. It emphasizes the perceived utility following the utilization of products/services and aligns with the degree of expectation confirmation. This alignment facilitates a more comprehensive exploration of the relationships between users and information systems (36, 38). While the ECM effectively analyzes the impact of factors such as information quality and service quality on users’ perceptions and their relationship with usage intentions, it lacks an analysis of users’ perceived value and emotional shifts following the usage of information system products/services, as well as the subsequent impact on sustained usage intentions (36, 38). However, the C-A-C framework can effectively predict users’ perceived value and emotional shifts following product usage. Thus, in this study, we adopted the C-A-C theory as a support framework for the ECM, enhancing the construction of a relationship model framework between users and health science popularization videos.

The value of media information can be classified into two categories: utilitarian and hedonic. Utilitarian value refers to the practical utility of the information content, whereas hedonic value represents the impact of the information content on users’ mental and emotional states (39, 40). During the process of using media information, individuals’ expectation confirmation is often influenced by the content and format of the information, as well as the perceived value post-usage. These influences subsequently contribute to the generation of subsequent sustained behaviors (41, 42). Similarly, health science popularization videos, as a type of media information, possess the same value attributes as media information. For instance, when individuals have health-related needs, they actively seek out and engage in health science popularization videos to obtain health information, accompanied by their expectations. In this process, the ability of the information content and format to evoke perceived value and whether this perceived value aligns with their expectations before searching not only impact individuals’ emotional shifts, satisfaction, and trustworthiness but also holds paramount significance in determining whether they develop a sustained intention to use.

Building on the C-A-C model and incorporating the ECM, this study constructs a novel theoretical framework that comprehensively investigates how the content and format of health science popularization videos influence individuals’ perceived value, expectation confirmation, emotional shifts, satisfaction, trustworthiness, and intention for sustained usage. Simultaneously, it unveils the interplay among these factors. This endeavor not only provides theoretical and practical guidance for the enhancement and development of existing health science popularization videos but also exerts a subtle yet influential role in elevating individuals’ health literacy levels in their everyday lives.

Health science popularization videos constitute a comprehensive means of conveying health information, encompassing factors such as health information content, professionalism, information presentation, and interactive communication (43). These factors are manifested through the professional articulation of health bloggers, presentation of information content, and interactive communication within health science popularization videos. During this process, individuals acquire health knowledge by consuming video content, which influences their perceived value and intention for sustained usage. Therefore, in this study, the image, professionalism, information quality, and interactive quality of health bloggers were consolidated as critical factors that impact users’ perceived value.

Based on the C-A-C and ECM models, a new model was built to explore the impact of health science popularization videos on users’ emotional states and willingness for continued use, as shown in Figure 1.

Based on the aforementioned hypotheses, this study designed a questionnaire and invited 10 participants with experience using health science popularization videos to complete a pilot-test. Ambiguous or easily misunderstood sections of the questionnaire were then revised to ensure the participants’ comprehensive understanding of the content and enhance the quality of the questionnaire. After completing the initial draft revisions, a pilot-test was conducted in a residential community in Taiyuan, China, resulting in the collection of 67 valid questionnaires. Following the reliability coefficient standards proposed by Nunnally and Bernstein (77), the Cronbach’s alpha value from this test exceeded 0.7, indicating acceptable internal consistency and stability of the questionnaire.

In terms of content validity, five experts with backgrounds in medicine and communication were invited to validate this questionnaire, which yielded an Index of Objective Consistency (IOC) of approximately 1, indicating that the content validity of the questionnaire was appropriately confirmed (78).

This study was approved by the Ethics Committee of Shanxi Bethune Hospital (YXLL-2022-021). The survey was conducted with the consent of each individual, and all resplendent data are kept confidential. The questionnaire was designed and produced on the Wenjuanxing (China) platform and distributed and collected using mobile devices. Online surveys offer several advantages (79): (1) Sampling is not limited to a solitary geographic area; (2) expenses are kept at a comparably low level; and (3) the survey boasts a swift response rate. To enhance participants’ engagement and trust, the instructions section of the questionnaire informed participants of the relevant requirements and explained that sincere completion of the questionnaire would lead to a random reward of no less than 10 RMB.

This study collected 560 surveys sourced exclusively from second- and third-tier Chinese cities. The collected data represent a reliable representation of urban dwellers in China (80). Table 1 outlines a detailed description of the questionnaire that was analyzed. A total of 57 responses were excluded during the data screening process due to incompleteness or a response duration of less than 5 min. Following thorough examination, 503 valid questionnaires were retained, resulting in a response rate of 89.8%.

This study employed established scales from previous studies with appropriate modifications based on the research context. Regarding blogger images, reference was made to Ohanian’s study (81). The professionalism of bloggers was influenced by the study by Clyde et al. (82) Video quality and video interaction quality were drawn from the works of Lin and Wang (83), as well as Liu (84). The concept of perceived value was informed by the study Kim et al. (85) and Gupta and Kim (86) Expectation confirmation and satisfaction dimensions were based on the studies by Bhattacherjee (21), Bhattacherjee et al. (36), and Oliver (87). The measurement of pleasure was guided by the study by Watson et al. (88). Trust and intention to continue use were informed by the studies by Bhattacherjee (21), Bhattacherjee et al. (36), Davis et al. (89), and Bart et al. (90). The measurement elements were formulated utilizing a 5-point Likert scale, encompassing five gradations: “strongly disagree,” “disagree,” “undecided,” “agree,” and “strongly agree,” which were, respectively, assigned values of 1, 2, 3, 4, and 5. The questionnaire items are listed in Table 2.

To uphold translation quality, we employed a procedure involving translation and subsequent back-translation, given that the data were gathered in China. As the primary step, we engaged two linguistic professors to evaluate the significance and comprehensibility of each item. Collaboratively, we proceeded to render the English questionnaire in Chinese with expert guidance. Subsequently, a doctoral student undertook the task of translating the Chinese questionnaire back to English. Ultimately, we meticulously scrutinized the translated elements against the original English rendition, refining the translation and rectifying any divergences to establish congruity between the two English iterations.

In this study, the research hypothesis model was validated using the Smart-PLS. First, a comprehensive analysis of the reliability and validity of the model was conducted by estimating and validating the path coefficients and explanatory capacity of the structural model. These sequential processes aimed to substantiate the dependability and credibility of the model’s constructs, ultimately affirming their interconnectedness among these constructs (91). This method allowed us to investigate the causal relationships between variables and concurrently handle model construction and measurement items (92). Smart-PLS is not only lenient with respect to the normality and randomness of the sample but also suitable for addressing relationships among variables in datasets with non-normal distributions, and it possesses advantages in predicting complex predictive models (93). Majchrzak et al. (94) advised that a prudent guideline is to maintain a sample size that is at least 5–10 times the maximum number of model paths. In the context of this study, the sample size was 503, and the maximum count of path coefficients reached 13. This aligns with the conditions that render PLS a suitable choice for this analysis. For this study, the SmartPLS software (Version 3.3.7), developed by Ringle et al. (95), was employed.

By analyzing the reliability of the variable items and the internal consistency, convergent validity, and discriminant validity of each construct, this study ensured that each item factor loading met the requirement of a threshold of 0.6 (96). The composite reliability (CR) for each construct exceeded the threshold of 0.7 (97), indicating a high degree of internal consistency within the constructs. Regarding convergent validity, the average variance extracted (AVE) index for each construct exceeded 0.5, suggesting satisfactory convergent validity for each construct (98), as shown in Table 3. As shown in Table 4, the square root of AVE for all variables was greater than that of the other variables, indicating good discriminant validity of the questionnaire.

Within the framework of this study, the assessment of the proposed model’s comprehensive quality was conducted using the goodness of fit (GOF) measure, as put forth by Tenenhaus et al. (99). The computation of the GOF is as follows:

G O F = A V E ¯ × R 2 ¯ = 0.767 × 0.587 =0.670.

Based on the aforementioned outcomes, the GOF value was 0.670, surpassing the established threshold of 0.36 (100), proving the validity of the model. Furthermore, to validate the model according to the fit criteria proposed by Henseler et al. (101), an SRMA < 0.08 and NFI > 0.90 are considered indicative of good fit. Our analysis yielded an SRMR of 0.052 and an NFI of 0.833. Although the NFI is <0.90, it remains within an acceptable range. Additionally, Hair et al. (102), have emphasized the importance of the R² value in evaluating the explanatory strength of the endogenous latent variables, which serves as a crucial reference for model fit. The higher R² value indicates stronger predictive and explanatory power (97). In this study, The R² values for the latent variables ranged from 0.315 to 0.724, indicating that the model possesses good fit. As shown in Figure 2. Q² is utilized to assess the predictive ability of the model and serves as a metric for model fit as well (102). A Q² value greater than 0 indicates predictive power, with higher values signifying stronger predictive capability. In this study, the Q² values ranged form 0.216 to 0.558, thus supporting the conclusion that our model demonstrates good fit. In summary, the model in this study exhibits good fit, as evidenced by the GoF, SRMR, NFI, R², and Q² values.

In this study, PLS was employed to examine the path coefficients of the hypothesis model, reflecting the causal relationships and their strengths among the dependent variables. The R square values indicate the percentage of variance explained by the dependent variables and represent the predictive ability of the model. Bootstrapping was utilized to estimate the significance of each path coefficient. By resampling the data, the estimated values provide more accurate results than conventional asymptotic approximations (103). Thus, this approach was employed in this study to assess the significance of the relationships between variables.

In testing the hypotheses, it was found that image appearance, professionalism, video quality, and interaction quality of health science communication video bloggers positively influenced users’ perceived value. Specifically, the path coefficients and associated statistical values were as follows: (BA → PV, β = 0.303, t-value = 4.832; BP → PV, β = 0.184, t-value = 4.524; VQ → PV, β = 0.182, t-value = 3.324; VIQ → PV, β = 0.138, t-value = 3.134), supporting research hypotheses 1, 2, 3, and 4. In the context of examining the relationships among perceived value, satisfaction, and pleasure, it was demonstrated that perceived value positively influenced both satisfaction and pleasure. The following path coefficients and associated statistical values were observed: (PV → S, β = 0.298, t-value = 7.106; PV → P, β = 0.562, t-value = 15.228; P → S, β = 0.520, t-value = 11.844), supporting research hypotheses 5, 6 and 7. Regarding the hypotheses related to expectation confirmation, it was revealed that expectation confirmation positively influenced both perceived value and satisfaction. The path coefficients and associated statistical values were as follows: (EC → PV, β = 0.207, t-value = 4.668; EC → S, β = 0.124, t-value = 2.434), supporting research hypotheses 8 and 9. Examining the relationships among pleasure, satisfaction, and trust, it was found that pleasure and satisfaction positively influenced trust. The path coefficients and associated statistical values were as follows: (P → T, β = 0.340, t-value = 5.836; S → T, β = 0.447, t-value = 7.496), supporting research hypotheses 10 and 11. Finally, research hypotheses 12 and 13 were also supported, as it was observed that satisfaction positively influenced continuous usage intention (S → CUI, β = 0.538, t-value = 13.023). Additionally, trust positively influenced continuous usage intention (T → CUI, β = 0.339, t-value = 7.608). Thus, all the research hypotheses of this study were supported. The model’s path coefficients and significance levels are presented in Table 5 and Figure 2.

The Sobel test was used to determine the statistical significance of the proposed mediation model in this study. This involved obtaining Z-values that approximated the p-values. Z-value greater than 1.96 indicates the presence of a significant indirect effect between the independent and dependent variables (104). To further estimate the effects of the mediating variables, this study employed the bootstrap method with bias-corrected confidence estimates (105). Specifically, 95% confidence intervals for individual mediating effects were derived using 5,000 bootstrapped resampling iterations. If the value of 0 falls outside the 95% confidence interval, it indicates a statistically significant mediating effect, as shown in Table 6.

Perceived value, a result of users’ comprehensive evaluation of products (services), plays a key role in influencing users’ satisfaction, loyalty, and continuous use intention (56, 59). Typically, users are first influenced by factors such as the content and quality of products (services) during the usage process, which create the perceived value of the product. The level of this perceived value affects users’ emotions and subsequent behavior (62). The results of this study confirmed this process. The appearance, professionalism, information quality, and interaction quality of health science popularization video bloggers have a significant positive impact on users’ perceived value. This shows that, to improve people’s perceived value of health science popularization, it is necessary to comprehensively improve the content and format of health science popularization videos and satisfy users’ needs in terms of their visual, auditory, content, and experiential aspects.

Expectation confirmation refers to the evaluation of user expectations after using an information system. During the process of continued information behavior, users will have expectations of the information system before using it. The perceptions generated during the use of the information system will be compared with the expected expectations. If the expectations are confirmed, it will lead to satisfaction and a willingness to continue to use it (36). This study confirmed that the confirmation of users’ expectations from health science popularization videos had a positive impact on perceived value and satisfaction. This is consistent with previous research results (65–67). This indicates that users’ expectation confirmation and perceived value are interdependent. To increase users’ satisfaction with health science popularization videos, it is necessary to ensure that users’ expectation confirmation and perceived value are consistent so that they can be satisfied and continue to use them.

Pleasure, satisfaction, and trust reflect users’ psychological well-being, encompassing not only their emotional states during the utilization of products (or services) but also their influence on the intent for continued usage (69, 73, 74). This study substantiated that users’ pleasure and satisfaction were influenced by perceived value and, in turn, these feelings impacted users’ trust and intent for the continuous usage of health science communication videos. These findings are consistent with previously relevant research outcomes. Emotional changes constitute a multifaceted process influenced by numerous factors. In the context of users engaging in health science popularization videos, when their emotions lean toward pleasure, it leads to an elevation in satisfaction and trust, thereby deepening the genesis of their intention for continued usage. Thus, increasing people’s intention to continue using health science popularization videos involves cultivating positive emotional experiences throughout their utilization, thus reinforcing user satisfaction and trust in these videos.