Self-care plays an important role in managing the progression of chronic diseases. A high level of control and effective symptom management can improve both somatic and psychological well-being in patients [1]. Self-care systems use personal goal-setting, social support, self-monitoring, and education to help patients in their daily lives [2]. Among various self-care approaches, technological methods appear to be more effective than others. The global availability and relatively low cost of technologies such as computers and smartphones make them suitable for remote monitoring and self-management [3].

Diabetes mellitus (DM) is a chronic disease that is directly influenced by lifestyle, including habits and behaviors related to glycemic control, dietary adherence, and physical activity. Therefore, encouraging motivation for self-care is a crucial aspect of diabetes treatment [4]. Poor control of diabetic symptoms can lead to mortality, while increasing physical activity and improving glycemic control can help prevent morbidity [5]. According to the “Association of Diabetes Care and Education Specialists 7” framework, any diabetes self-care approach must have a measurable impact on specific factors. These include a healthy diet, physical activity, medication management, and regular blood glucose monitoring [6].

Studies have demonstrated the relative effectiveness of health monitoring applications in enhancing daily self-care among patients diagnosed with diabetes. Numerous programs utilize health tracking processes in the context of diabetes self-care [2]. Most apps focus on blood glucose monitoring, which has been identified as the most important metric for diabetes management [6]. One well-known example is the Glucose Buddy (Azumio Inc) app, which allows users to manually log and track diabetes-related metrics such as blood glucose levels [7]. Mobile self-monitoring apps should include multiple features, such as blood glucose tracking, physical activity monitoring, food intake logging, reminders, note-taking, and social support, to be fully functional [8]. However, it has been shown that relying solely on self-monitoring features without incorporating gamification can limit the program’s effectiveness in improving self-care behaviors [9]. Platforms like SlimKicker (Henley Wing Chiu) are examples of systems designed for tracking health metrics and physical activity without incorporating engaging game design elements [10]. In contrast, MySugr (Roche diabetes care) is a diabetes management app that combines health monitoring with gamification elements to enhance user engagement and adherence [11].

Gamification is the process of using game elements to engage and motivate individuals in serious contexts unrelated to entertainment [10]. Applying gamification in health care is an effective tool for shaping patient behavior through user engagement [12]. Techniques such as self-monitoring, goal setting, reward structures, social incentives, level design, medal/badge systems, and narrative context are among the most well-known and widely used behavior change strategies [13]. Reward and incentive systems have a powerful impact on habit formation and can foster intrinsic motivation for users to repeatedly engage in desired behaviors [14].

It has been shown that gamification is more effective in establishing desirable healthy habits than health monitoring alone [9]. Evidence indicates that gamification in mobile health apps leads to measurable improvements in glycemic control and medication adherence [15]. As an example, the MySugr program allows users to track and monitor their blood glucose levels, medications, and calorie intake while using challenges, achievements, and a monster avatar to encourage regular data logging [14]. Gamification enhances user motivation and supports behavior change by making health-related tasks more enjoyable and goal-oriented [16]. Techniques like point scoring, social incentives, and predictable rewards for desired actions effectively facilitate habit formation and behavioral change by reinforcing target behaviors and shaping lasting routines [17]. One of the most notable examples is exergames, games designed to promote physical activity. Games like Bant (University Health Network) track users’ physical activity using GPS and encourage improvement [10]. It has been shown that incorporating scoring and reward systems in exergames can effectively foster sustained physical activity in users [18].

However, sole reliance on points and rewards can undermine the essence of gamification, making the experience less entertaining and engaging. Scoring elements must be integrated into a compelling and enjoyable context [14]. Studies indicate that using narrative and engaging audiovisual designs, similar to features found in video games, can convey a sense of fun, immersion, and pleasure, helping to reduce the perception of pain or discomfort [6]. A major flaw in traditional gamification designs is the absence of essential game elements such as narration, high-quality graphics, and gameplay, resulting in emotionally unappealing experiences for children and adolescents [19]. Integrating visually engaging elements and fun designs helps maintain user interest and enhances the educational value of diabetes self-care apps [20]. By using compulsion loops, cycles involving anticipation of a reward, performing an action, and receiving the reward, gamified systems can reinforce desired behavioral habits [21]. Some studies describe gamification as an art form, emphasizing that the most effective way to foster behavioral change is through well-designed reward systems. Aesthetically appealing and strategically crafted rewards engage both cognitive and motivational systems to shape lasting habits [22].

Apart from using the mentioned game design elements, creating an immersive atmosphere is necessary to reach an engaging experience. Toward this goal, the use of virtual reality (VR) platforms has been increasingly discussed in recent years [23]. Jo and Park [24] found that VR’s visual aesthetics, including shape, depth, 3D graphics, and visual design, significantly influence perceived enjoyment and adherence to the technology. VR promotes behavior change through technological novelty focused on embodied experience. By incorporating interactive designs, VR creates a sense of flow, presence, and immersion [25]. Studies show VR environments can encourage healthy behaviors, including regular exercise and proper nutrition [26].

The main goal of this research was to design an innovative gamification program for diabetes self-care. To our knowledge, there is limited evidence of gamified VR programs that integrate similar audiovisual and gameplay elements for diabetes self-care. This study helps address this gap by providing an initial implementation and evaluation of such a system. Following the development of the app, the next objective of our research was to assess its effectiveness in improving diabetes symptoms.

To evaluate the efficiency of every self-care method, such as our VR-gamified app, there are some metrics used in previous literature. Blood glucose monitoring is recognized as the most important metric of diabetes self-care [6]. Fasting blood sugar (FBS) is considered a short-term marker of blood glucose levels, while hemoglobin A_1c (HbA_1c) serves as a long-term indicator [27]. Research has identified several key factors that influence blood glucose control, with dietary adherence and physical activity being among the most significant predictors of glycemic regulation [28]. In line with these findings, we used 5 key metrics to evaluate the effectiveness of our designed intervention. These metrics are categorized into two main groups: (1) health symptoms, including FBS, HbA_1c, and BMI, and (2) health-related behaviors, including physical activity and food intake.

We hypothesized that experiencing our VR gamified program would have a greater impact on improving the primary self-care outcomes for diabetes. Based on this expectation, the following 5 hypotheses were formulated:

The study used a parallel-group randomized controlled trial comprising 4 arms, with participants allocated to the groups with respective sizes of 16, 14, 15, and 15 for final analysis. Each group received one kind of self-care intervention for a 6-week period.

Two methodological changes were implemented after trial commencement. First, the physical activity measurement approach was modified from objective pedometer-based tracking to structured self-report assessment to enhance feasibility and adherence in the clinical setting. Second, the Iranian diabetes-monitoring app Idia (Mehr Nutrition Technology and Knowledge Development) was initially planned as a comparator for the VR-gamified group; however, limitations in its expected feature set necessitated its replacement with the MySugr app as one of the comparators.

The target population for this study consisted of children and adolescents aged 7-15 years diagnosed with diabetes, living in Tehran, Iran. Inclusion criteria were (1) a formal diagnosis of diabetes, (2) aged between 7 and 15 years, (3) basic computer/VR literacy, and (4) willingness and ability to participate in a 6-week eHealth trial. Exclusion criteria included any medical conditions that would prevent safe use of VR headsets (eg, cybersickness or nausea). The research team was introduced to potential participants as specialists in health psychology affiliated with the PhD program at the University of Tehran in order to provide a clear professional context for the intervention. A total of 78 participants meeting the inclusion criteria were recruited from The Children’s Medical Center in August 2024 to November 2024, Tehran, Iran, in an offline, closed manner. Following recruitment, participants were provided with face-to-face instructions and a briefing. Written informed consent was subsequently obtained from eligible participants (n=68) who enrolled in the trial. Eight participants lost to follow-up over the 6-week period, resulting in a final sample of 60 participants (mean age 11.8, SD 1.74 years; 33, 55.0% boys and 27, 45.0% girls). All outcomes were self-assessed by online questionnaires throughout the trial timeline.

All participants (or their parents) were instructed to complete a structured data collection form throughout the 6-week study period, which is provided in Multimedia Appendix 2. The form required daily entries, including FBS, food intake, and physical activity, along with weekly entries for body weight. HbA_1c levels were recorded twice, once at baseline (pretest) and once after the final session (posttest). Demographic variables, including age, gender, and type of diabetes, were collected at the outset.

Sample size was calculated using G*Power (Heinrich Heine University Düsseldorf) software based on a repeated-measures ANOVA design with 4 groups. Assuming a medium effect size (f=.25), a significance level of α=.05, and statistical power of 0.80, the minimum required total sample size was estimated to be 52 participants. To account for potential attrition, additional participants were recruited. No interim analyses or stopping guidelines were defined or conducted.

Participants were randomly assigned to one of the 4 groups using computer-generated stratified block randomization based on age and sex. The allocation sequence was generated by the technical assistant who was not involved in the study assessments. Allocation concealment was maintained through a sequentially numbered process, and the care provider responsible for briefing and tutorials remained unaware of group assignments until the moment of allocation.

Due to the nature of the interventions, neither participants nor care providers could be blinded to group assignments. All outcome data were self-reported via online questionnaires, with no external outcome assessor involved. However, the data analyst was provided with anonymized group labels to minimize potential analysis bias. To reduce preconceived biases about the intervention of interest and comparators, the informed consent focused on the shared goal of diabetes improvement and participants’ compensation, ensuring a neutral presentation of all study arms. No similarity of a placebo intervention was applicable here.

Given the repeated measurements of outcomes, a linear mixed model (LMM) was used. To investigate significant differences between groups, one-way ANOVA and paired-samples 2-tailed t tests were used. A correlation analysis was also used as an additional analysis. All statistical analyses were performed using SPSS Statistics (version 27; IBM Corp). The significance level was set at α=.05 for all statistical tests.

Handling of missing data: the variables BMI and HbA_1c contained no missing data, as BMI was derived from weekly weight measurements and HbA_1c was collected only at the pretest and posttest time points. The other 3 outcomes were constructed at the weekly level, as it is an established, accepted practice in mobile health and diabetes care studies. For FBS, physical activity, and food intake, missing values existed only in the daily records. According to the operational definitions of the study, these variables were aggregated to the weekly level using the mean or sum of the available daily entries. Weeks with partial daily logs were still computable, because weekly values can be calculated based on the days that were present rather than requiring a complete daily set. As a result of this aggregation process, the final analytic dataset contained no missing values for any of the weekly variables. No imputation procedure was used.

The study was reported in accordance with the CONSORT (Consolidated Standards of Reporting Trials) 2010 statement. The preparation of the manuscript followed the CONSORT-EHEALTH (Consolidated Standards of Reporting Trials of Electronic and Mobile Health Applications and Online Telehealth) reporting framework (Checklist 1) [34], and the abstract was structured according to the CONSORT guidance for abstracts [35], to ensure transparent and standardized reporting.

Ethical approval was obtained from the ethical committee of Tehran University’s psychology department, under the supervision of the Ministry of Health and Medical Education (MOHME), indexed as IR.UT.PSYEDU.REC.1404.106. All procedures were conducted in accordance with the ethical standards of the institutional research committee and the Declaration of Helsinki. Prior to participation, offline written informed consent was obtained from all participants or their parents. Participants were informed about the purpose of the study, trial procedures, and their right to withdraw at any time without any consequences. Participant privacy and confidentiality were strictly protected throughout the study. All collected data were deidentified and stored using coded identifiers, and no personally identifiable information was included in the analysis or reporting of the results. Access to the data was restricted to the research team and maintained in secure files in accordance with institutional ethical guidelines. No identification of individual participants in any image of the manuscript or supplementary material is possible. Participants received a small compensation to acknowledge the time and effort required to take part in the study. The type and amount of compensation were determined in accordance with the institutional ethical guidelines and were provided equally to all participants regardless of group allocation or study outcomes. No personal or health-related information was collected or transmitted through the VR application. In addition, XR safety procedures were implemented, including supervised use and monitoring for symptoms such as cybersickness, nausea, or any other bodily discomfort.

A total of 78 individuals were recruited and assessed for eligibility. After the initial briefing, 10 were excluded (3 did not fully meet the inclusion criteria, 5 declined to participate, and 2 for other reasons). A total of 68 participants were randomized using stratified block randomization into 4 parallel groups with equal-sized arms (n=17). During the 6-week study period, a total of 8 participants were lost to follow-up. Consequently, 60 participants were included in the final analysis. The detailed participant flow is illustrated in Figure 3.

Participants were recruited and enrolled between August and November 2024. The intervention phase of the trial was conducted from December 2024 to February 2025, during which each participant completed a 6-week follow-up period. The trial concluded as planned without early termination.

Table 1 presents the baseline demographic and clinical characteristics of participants in the 4 study groups prior to the intervention.

Data from 60 participants with complete outcome data were included in the final analysis. The denominator for all analyses corresponds to participants who completed the study and provided postintervention measurements. Analyses were conducted according to the originally assigned groups. The analyzed sample sizes were 16 in the VR gamified group, 14 in the MySugr app group, 15 in the traditional counseling group, and 15 in the control group.

The results of the LMM for all 5 outcomes are summarized in Table 2.

An LMM was fitted with time, condition, and their interaction as fixed effects and a random intercept with autoregressive correlation of order 1 (AR1) covariance for participants. Model fit indices indicated adequate performance (Akaike information criterion [AIC]=2383.29), suggesting that adding time and the interaction improved overall model parsimony relative to the reduced model. The results of the LMM, as reported in Table 2, indicate that the time×condition interaction effect was statistically significant (F_15,280= 2.046; P=.01; partial η²≈.099). However, no significant main effects were observed for time (F_5,280=0.269; P=.93; partial η²≈.005) or condition (F_3,39.91=0.171; P=.91; partial η²=.001), suggesting that overall changes over time and between-group differences were not statistically significant in isolation. Random-effects estimates indicated substantial between-subject variability (random-intercept variance=143.45, SE 29.92). The AR1 correlation parameter was negative (ρ=−.19), indicating weak inverse correlation between adjacent repeated measurements. Residual variance was 29.05 (SE 2.45). The temporal changes in the weekly average of FBS across the 4 experimental groups over 6 repeated measurements are illustrated in Figure 4. The significant interaction effect indicates that the pattern of FBS improvement over time varied by group. Specifically, participants in the VR gamified group showed a distinct trend in reducing FBS compared to other groups, with estimated marginal means decreasing from 123.625 mg/dL, 95% CI 114.147-133.103 at baseline to 118.125 mg/dL, 95% CI 110.387-125.863 at week 6. A paired-samples t test was conducted to examine the change in average FBS levels within the VR gamified group between the first and last sessions. The results revealed a significant decrease in FBS levels over time (t₁₅=−2.612; P=.02; mean difference=−5.500, 95% CI −1.012 to −9.987; Cohen d=0.653, 95% CI 0.102-1.186), indicating improved glycemic control among participants in this group, which is in line with recent studies [36]. This finding supports hypothesis 1.

An LMM was estimated with time, condition, and their interaction as fixed effects and a random intercept with an AR1 covariance structure for participants. Model fit indices (AIC=1089.44) indicated a relatively well-fitting model with adequate parsimony for the repeated-measures structure. According to Table 2, LMM results revealed a significant main effect of condition (F_3,55.96=14.399; P<.001; partial η²≈.435) and a nonsignificant main effect of time (F_5,280.61=1.580; P=.16; partial η²≈.027). The interaction effect of time×condition was significant (F_15,280.61= 2.544; P=.001; partial η²≈.120), indicating that changes in physical activity over time differed significantly between the VR gamified group and the other experimental groups. Random-effects estimates showed modest between-subject variability (random-intercept variance=0.155; SE 0.0678). The AR1 correlation parameter was positive (ρ=0.656), indicating moderate positive correlation between adjacent repeated measurements. Residual variance was 0.979 (SE 0.083). Figure 5 illustrates the temporal changes in weekly physical activity across the 4 experimental groups. While physical activity levels remained relatively stable in the control, traditional, and app-based groups, the VR gamified group exhibited a noticeable upward trend over time. Thus, hypothesis 2 is supported.

The one-way ANOVA analysis for physical activity during the last session revealed a significant difference in average physical activity between the VR gamified group and the other experimental groups (F_3,56=21.234; P<.001; partial η²=.532). A paired-samples t test result showed improvement of physical activity in VR group between the first and last sessions (t₁₅=3.850; P=.002; mean difference=1.625, 95% CI 0.725-2.524; Cohen d=0.963, 95% CI 0.355-1.549). These findings align with recent meta-analytic evidence demonstrating a strong positive effect of exergames on physical activity levels [37].

An LMM was estimated with time, condition, and their interaction as fixed effects and a random intercept with an AR1 covariance structure for participants. Model-fit indices (AIC=4138.80) indicated acceptable fit for the repeated-measures design. According to the LMM results presented in Table 2, the main effect of condition was not significant (F_3,336=0.115; P=.95, partial η²≈.001), while the main effect of time was significant (F_5,336=3.487; P=.004; partial η²≈.059), indicating that the observed downward trend over time was unlikely due to chance. The time×condition interaction effect was significant (F_15,336=2.794; P<.001; partial η²≈.111), suggesting that the pattern of change in food intake over time differed between the VR gamified group and the other groups. Random-effects estimates showed large between-subject variability (random-intercept variance=24,595.20,.20; SE 4685.39). The AR1 correlation parameter was reported as ρ=1.000; however, SPSS notes this value is redundant, typically indicating boundary estimation or insufficient information for precise AR1 correlation. Residual variance was 4737.43 (SE 400.39). Figure 6 illustrates the temporal changes in the weekly average calculated food intake across the 4 experimental groups. While food intake fluctuated irregularly in the control, traditional, and app-based groups, the VR gamified group showed a consistent downward trend. Thus, hypothesis 3 is supported.

Results of the paired-samples t test for the food intake variable in the VR gamified group showed a significant decrease in average food intake from the first to the last session (t₁₅=−3.850; P=.002; mean difference=−147.500, 95% CI −0.725 to –2.524; Cohen d=0.963, 95% CI 0.355-1.549), which is in line with recent studies [38].

An LMM was fitted with time, condition, and their interaction as fixed effects and a random intercept with an AR1 covariance structure for participants. Model-fit indices (AIC=429.61) indicated that the model adequately captured the repeated-measures structure. According to Table 2, the LMM results for BMI showed no significant effects. For time effect (F_5,280=0.602; P=.69; partial η²≈.011), for condition effect (F_3,40.19=0.021; P=.99; partial η²≈.000), and for the time×condition interaction effect (F_15,280=0.660; P=.82; partial η²=.034). Random-effects estimates indicated substantial between-subject variability (random-intercept variance=21.17; SE 4.32). The AR1 correlation parameter was slightly negative (ρ=–0.209), indicating weak inverse correlation between adjacent repeated BMI measurements. Residual variance was 0.0399 (SE 0.00337). Figure 7 illustrates that the weekly calculated BMI remains relatively stable and consistent across all 4 experimental groups throughout the study period. This suggests that the interventions did not have a significant impact on BMI changes over time. Therefore, hypothesis 4 is rejected.

An LMM was fitted with condition, time, and their interaction as fixed effects. A random intercept for participants was included, and the identity covariance structure was used. The model showed acceptable fit (AIC=306.57). The LMM results presented in Table 2 reveal significant effect for the condition effect (F_3,112=2.980; P=.03; partial η²≈.074), the time effect (F_1,112=0.230; P=.63; partial η²≈.002), and the time×condition interaction effect (F_3,112= 0.064; P=.97; partial η²=0.002). Although the main effect of condition reached statistical significance in the mixed model, this effect reflects baseline differences between groups rather than any intervention-related change, as the related effect size is small (partial η²≈.074). Therefore, no meaningful or clinically relevant effect of the intervention on HbA_1c can be inferred. As Figure 8 shows, the pattern of change between the 2 time points did not differ across conditions. Therefore, hypothesis 5 is rejected. Random intercept variance was very small (0.00098), reflecting minimal between-subject variability after accounting for fixed effects. Residual variance was 0.719 (SE 0.096).

Despite the strong correlation between HbA_1c and FBS (r₅₈=0.641; P<.001; 95% CI 0.469-0.770), our results showed no significant change in HbA_1c levels between the pretest and posttest. This finding contrasts with previous studies (eg, Kerfoot et al [39]), which reported significant changes in HbA_1c levels over a 12-month period (P=.048). The discrepancy may be attributed to the relatively short duration of the current intervention.

One exploratory analysis was performed, consisting of a correlation analysis to examine associations between behavioral and metabolic outcomes. As Table 3 shows, significant associations were observed between changes in FBS and physical activity (r₅₈=–0.422, 95% CI –0.610 to –0.188; P<.001) and between changes in physical activity and food intake (r₅₈= –0.292, 95% CI –0.509– to 0.041; P=.02), indicating linked behavioral and metabolic shifts. This pattern aligns with prior findings on the relationship between physical activity and blood glucose levels [40].

All XR sessions were completed without any adverse events or early terminations. No instances of cybersickness, visual discomfort, dizziness, or excessive fatigue were reported, and no participant required a session to be stopped prematurely. Adherence to the intervention was exceptionally high, with all children attending and completing all scheduled sessions. Participants in the VR group (n=16) demonstrated high perceived usability, with a mean System Usability Scale score of 74.2 (SD 7.9). According to established usability benchmarks, this value falls within the Excellent range, indicating that the VR environment was experienced as intuitive, consistent, and easy to navigate. No individual item showed unusually low ratings, suggesting that participants were able to interact with the system with minimal cognitive effort or technical difficulty. TAM responses further supported strong user acceptance of the VR system. Perceived usefulness yielded a mean score of 5.3/7 (SD 0.7), while perceived ease of use was rated 5.6/7 (SD 0.6). These values align with a moderately good acceptance profile, reflecting participants’ perceptions that the VR platform was beneficial, motivational, and easy to learn.

The present study aimed to evaluate the effectiveness of a VR gamified program in improving metabolic indicators and health-related behaviors among children and adolescents diagnosed with diabetes. The intervention was compared with 3 comparators, a gamified mobile app (MySugr), traditional consultation, and standard medical care as a control group, in a longitudinal 6-week parallel-groups RCT design. Overall, the results indicate that the VR gamified intervention produced the most consistent improvements in behavioral outcomes, including physical activity and dietary control, compared to the other 3 parallel groups. Moderate yet promising improvements in FBS were also observed in the VR gamified group relative to the other intervention groups, suggesting that the VR gamified method can potentially yield an improved level of short-term glycemic control compared to other methods. In contrast, the intervention did not produce significant improvements in long-term metabolic indicators, including BMI or HbA_1c, in the VR gamified group compared to other interventions.

The behavioral outcomes including physical activity and dietary control are consistent with recent studies, indicating that interactive gamified environments can enhance the development of sustained health behaviors [41]. The significant improvement in short-term glycemic control (FBS) alongside a lack of significant improvements in long-term metabolic control markers, including HbA_1c and BMI, are in line with recent studies, as long-term metabolic indicators typically require sustained behavioral modification over longer periods to demonstrate measurable changes [42]. The overall improvement trends observed in the VR-gamified group highlight the importance of emotional engagement in self-care interventions. This finding is consistent with previous studies emphasizing that interactive gameplay and immersive audio-visual experiences are critical factors in the success of health gamification systems [43]. In contrast to conventional health-tracking programs that rely primarily on point/badge reward structures, the VR gamified intervention used in this study integrated immersive 3D environments with action-based gameplay that directly represented users’ health symptoms and behaviors. Mentioned features may explain the stronger behavioral engagement observed in the VR gamified group and may also clarify why traditional point/badge gamification approaches may produce weaker effects, as reported in some studies [44].

Real-world implementation of this VR intervention could align with existing pediatric diabetes education pathways in which families and educators work together to support self-care. In practice, parents can help supervise home-based sessions, while school health staff may provide complementary reinforcement during the day. Follow-up by diabetes educators or community nurses could be incorporated into routine clinical visits or remote check-ins, enabling the VR experience to serve as an adjunct to ongoing education rather than an isolated tool [45]. However, equitable access to VR remains a challenge. Variability in the availability of VR devices and limited technological resources in some families or schools may restrict adoption. To mitigate such barriers, future implementations may require shared devices in clinical settings or mobile-based alternatives to ensure broader accessibility. Due to the experimental design of this study, the external validity and generalizability of the findings should be interpreted with caution.

Several limitations should be considered when interpreting the findings of this study. First, the relatively small sample size (n=60) may limit the generalizability of the results. Second, the intervention period was relatively short (6 weeks), which may not have been sufficient to observe meaningful changes in long-term metabolic indicators, including BMI and HbA_1c. Third, outcomes were measured through self-reported data, which may have caused reporting bias. Future research should therefore incorporate larger samples, longer intervention durations, and objective behavioral tracking tools to provide more robust evidence regarding the effectiveness of VR gamified self-care interventions.

The VR gamified experience developed in this study appears to offer a promising complement or alternative to existing self-care methods, with preliminary indications of potentially enhanced engagement and outcomes. However, interpretations of the behavioral findings should be calibrated to the limitations of measurements, including reporting bias, limited sample size, short duration of the intervention, variations in adherence and fidelity, and the novelty effect of the VR game. Results should be understood as preliminary behavioral signals rather than evidence of clinical improvements. While the program effectively fostered behavioral engagement and adherence trends, no clinically causal inference can be drawn. Future studies with new game mechanics, more engaging designs, larger samples, longer follow-ups, and objective tracking tools can scale the usability of the present intervention and extend related findings.

This study combined gamification elements with immersive VR features to develop a novel self-care system for children and adolescents diagnosed with diabetes. The intervention was grounded in the behavioral economics framework, specifically the theoretical sequence of reward, motivation, behavior, and habit formation, which is central to modern gamification approaches. By offering engaging gameplay within an immersive VR environment, the game established a direct connection between real-life diabetes and in-game mechanics. The goal was to foster intrinsic motivation through in-game rewards, thereby encouraging users to sustain desirable self-care behaviors. Findings support the potential effectiveness of VR gamified intervention, highlighting its power to serve as a complementary or alternative digital tool for enhancing diabetes self-management. However, clinical implications should be interpreted cautiously.