This 3-arm randomized controlled trial used a 1:1:1 allocation ratio to evaluate the effectiveness of an exergame designed using monocular camera technology for individuals at high risk of developing T2D. A total of 45 participants meeting specific inclusion criteria were recruited and randomly assigned to 1 of 3 groups: a control group performing traditional video-guided exercises, a Kinect-based exergame group, and a monocular camera–based exergame group. The sample size was calculated based on a power analysis with an expected medium size at 0.48, where previous research reported medium effect sizes (0.40‐0.50) in evaluating usability and adherence in Kinect-based exergames for individuals with prediabetes [18], aiming for 80% power and a significance level of .05 to detect meaningful differences between the groups. The intervention was conducted in a controlled setting, ensuring consistency across groups while collecting physiological and experiential data.

A total of 45 individuals at high risk for T2D were recruited through offline outreach facilitated by community managers affiliated with local residential service centers in Minhang District, Shanghai, China. They supported recruitment by identifying eligible participants through existing health records and outreach channels. The inclusion criteria were as follows: (1) 35 years of age and older; (2) a score of 25 or higher on the Chinese Diabetes Risk Score; (3) community-dwelling individuals, not residing in assisted living or long-term care facilities; (4) physically capable of engaging in light to moderate exercise, as determined by self-report and physician clearance; (5) normal cognitive function that enables the participant to complete the experiment independently or with minimal assistance; and (6) written informed consent provided by participants or their families. Exclusion criteria included individuals with (1) diagnosis of T1D or T2D, (2) current participation in another exercise intervention study, (3) severe cognitive impairment with Mini-Mental State Examination score of <24, and (4) major mobility limitations such as severe osteoarthritis and recent orthopedic surgery.

Kinect group features the same software functionality and design as the application based on bullet-screen cameras, but the hardware is built using a Kinect sensor. The gamified platform is primarily developed and presented using Unity 3D. The Kinect 3D motion-sensing camera incorporates real-time dynamic capture and image recognition capabilities, offering new possibilities for interactive approaches to motion therapy.

The control group followed a set of 6 instructor-led exercise videos, which included crossover steps, high knees, lateral raises, punching movements, downward leg punches from a standing position, and elbow-to-chest expansions. Each session in the control group was designed to reflect the same aerobic and resistance exercise level as the Kinect group and the experimental group, ensuring comparability across groups in terms of physical exertion and activity content.

Participant recruitment was conducted from March to April 2024 in the Hongqiao Community. Individuals meeting the inclusion criteria were provided with comprehensive information about the study and signed written informed consent. Baseline demographic data were collected prior to random assignment using block randomization by researcher into one of three groups: (1) control group (traditional exercise control), (2) Kinect group (Kinect-based exergame), or (3) monocular group (monocular camera–based exergame).

Three days prior to the intervention, all participants received instructional videos demonstrating the 6 targeted movements to be performed during the exercise sessions. Participants were instructed to review and practice these movements to ensure familiarity with the protocol.

On the day prior to the intervention, trained researchers and caregivers delivered and assisted in setting up the required equipment—either a Kinect V2 sensor or an Aoni C33 monocular camera (manufactured by Shenzhen Aoni Electronic Industry Co, Ltd, resolution: 1920×1080 pixels, frame rate: 30 fps, field of view: 90°)—in participants’ homes. Setup was scheduled in the afternoon, approximately 1‐2 hours post meal, to ensure participants’ comfort and safety during physical activity.

On the first intervention day, baseline data, including age, gender, T2D risk, resting heart rate, and perceived fatigue, were measured prior to the session. Each intervention session lasted 10 minutes and consisted of 2 components: 5 minutes of aerobic exercise followed by 5 minutes of resistance training. Participants in the control group followed a prerecorded instructional video demonstrating the 6 aerobic movements, followed by guided resistance exercises. The monocular group used a custom-developed exergame based on monocular camera technology. Real-time visual instructions were displayed on a television screen, guiding participants through the same movement protocol. The Kinect group used an exergame designed with Kinect-based skeletal tracking, offering a similar visual and instructional format as the monocular system.

During the aerobic component, participants performed 6 distinct movements, each lasting 50 seconds. For the resistance component, participants selected their preferred resistance level (ranging from 0.25 to 1.5 kg), and resistance bands were applied to the arms and legs under caregiver supervision. Caregivers and researchers refrained from intervening during the session unless directly requested by participants to minimize external influence.

Immediately following the first intervention, participants’ postexercise heart rate, perceived fatigue (via Borg Rating of Perceived Exertion Scale), intrinsic motivation (via Intrinsic Motivation Inventory [IMI]), and user experience (including perceived enjoyment, challenge, and usability) were assessed using standardized instruments.

Participants were then instructed to continue the 10-minute intervention daily for 7 consecutive days independently without additional notification from the researchers. They were asked to document their participation by recording short videos as proof of session completion. At the end of the 1-week period, researchers returned to collect the devices and retrieve adherence data, including the number of completed sessions and any user feedback.

Data collected from the physiological measurements (resting and exercise heart rates) and the questionnaires were analyzed to compare the differences between groups and evaluate the effectiveness of the interventions. ANOVA analysis was used to understand the differences of the data collected from 3 experiment groups [22], and 2-tailed t test (for normally distributed result) or nonparametric test (for non–normally distributed result) was used to understand the significance of differences between 2 groups. For pairwise comparisons between 2 groups, independent-samples 2-tailed t tests were applied for normally distributed data. For data not meeting normality assumptions, nonparametric tests were used: the Kruskal-Wallis test for comparisons among 3 groups and the Mann-Whitney U test for pairwise group comparisons. Statistical analyses were conducted using SPSS (version 26.0; IBM Corp), with a significance level set at P<.05. Cohen f value is used to understand the effect size, and the critical points for small, medium, and large effect sizes are 0.1, 0.40, and 0.80, respectively.

This study was approved by the Shanghai Jiao Tong University’s institutional review board (H20230160I), and all procedures were conducted in accordance with the ethical standards outlined in the Declaration of Helsinki. Due to the initial pilot-phase nature of the project, the trial was retrospectively registered at ClinicalTrials.gov (NCT06950528), prior to data analysis. All participants provided written informed consent prior to participation. The study ensured strict confidentiality of participant data, and no personally identifiable information was collected or disclosed. Participants received a US $7 shopping voucher after completing the study. No identifiable images or recordings were used in the manuscript or supplementary materials.

This study recruited 45 participants identified as being at high risk for T2D. The CONSORT (Consolidated Standards of Reporting Trials) flowchart is shown in Figure 2. Participants aged between 40 and 54 years, with a mean age of 47.12 (SD 6.92) years. Participants were classified as high-risk based on the Chinese Diabetes Risk Score, with an average mean score of 33.38 (SD 5.04). As shown in Table 1, an ANOVA analysis of baseline characteristics among the 3 participant groups revealed no statistically significant differences. No negative incidents were reported.

The amount of exercise was assessed through heart rate and perceived fatigue. As shown in Table 2, the average heart rate for the control group, Kinect group, and monocular group was 86.34 (SD 7.48), 90.81 (SD 7.71), and 89.29 (SD 6.52), respectively. The between-group comparison revealed no significant difference in heart rate (P=.76; Cohen f=0.117), indicating a small effect size.

Exercise performance was evaluated using 2 indicators: heart rate as an objective physiological measure and perceived fatigue as a subjective self-report metric. As shown in Table 2, the average heart rates for the control, Kinect, and monocular groups were 86.34 (SD 7.48) bpm, 90.81 (SD 7.71) bpm, and 89.29 (SD 6.52) bpm, respectively. A one-way ANOVA revealed no statistically significant differences among the groups (F_2,42=0.28; P=.76), with a small effect size (Cohen f=0.117). These results suggest that the type of system used did not substantially impact cardiovascular exertion during the exercise session.

In terms of perceived fatigue, all participants reported increased fatigue after the intervention (P<.05). As shown in Figure 3, the control group reported the highest average fatigue level (mean 11.60, SD 1.12). However, between-group comparisons did not reach statistical significance (P=.25; Cohen f=0.265), indicating a moderate effect size. Post hoc comparisons revealed that the monocular group experienced significantly lower fatigue than the control group (P=.04), while the Kinect group’s difference was not significant (P=.15). These findings suggest a potential advantage of the monocular system in reducing perceived exertion during gameplay.

User experience following the first intervention session was assessed through self-reported measures of intrinsic motivation and game experience. Intrinsic motivation scores, as measured by the Interest/Enjoyment subscale of the IMI, were as follows: control group (mean 26.06, SD 1.87), Kinect group (mean 35.13, SD 3.20), and monocular group (mean 34.00, SD 4.41). A one-way ANOVA revealed a significant between-group difference in intrinsic motivation (P<.001), with a large effect size (Cohen f=1.204). Post hoc comparisons indicated that both the Kinect-based and monocular camera–based exergames significantly enhanced intrinsic motivation compared with those of the control group. However, there was no statistically significant difference between the Kinect and monocular groups (P=.44), suggesting comparable motivational effects across the 2 exergaming modalities.

Game experience, as measured by the GEQ and shown in Table 3, revealed significant differences in perceived challenge. Participants in the monocular group reported a higher challenge score (mean 3.45, SD 0.39) than those in the Kinect group (mean 2.96, SD 0.51), with the difference reaching statistical significance (t₂₈=−2.71; P=.01). One possible explanation is the Kinect system’s superior motion-tracking capabilities, which may have facilitated a stronger sense of virtual body ownership and smoother gameplay, thus reducing the perceived difficulty.

For all other dimensions of game experience—including sensory and imaginative immersion, flow, perceived competence, tension or annoyance, and emotional responses (positive and negative)—no statistically significant differences were observed between the 2 exergame groups (P>.05). While not statistically significant, participants in the Kinect group reported slightly higher tension levels, potentially due to unfamiliarity with the hardware. Conversely, the monocular group reported marginally higher negative affect, which may reflect minor limitations in motion-tracking responsiveness or accuracy.

User engagement over the 1-week intervention period was measured by the number of completed exercise sessions. On average, participants in the control group completed 2.33 sessions (SD 1.29), the Kinect group completed 2.93 sessions (SD 1.79), and the monocular group completed 3.60 sessions (SD 1.59). Although the monocular group exhibited the highest level of usage, between-group comparisons did not reach statistical significance (P=.49). Similarly, no significant difference was observed between the Kinect and monocular groups (P=.29).

A thematic analysis of the qualitative user feedback was conducted to contextualize participants’ experiences across groups. Feedback was first coded and then categorized into three major themes: usability, motivation, and future intention to use. (1) Usability: Most participants across all groups found the systems easy to operate. However, specific issues were reported. Four participants in the Kinect group mentioned challenges related to hardware setup (eg, spatial calibration and sensor alignment), while 1 participant in the monocular group cited difficulties with interface responsiveness. (2) Motivation: Lack of engagement was a recurring theme in the control group, with 6 participants indicating low motivation due to the static and repetitive nature of the game. (3) Future intention to use: Despite these challenges, participants in the exergame groups showed strong willingness to continue. In the Kinect group, 66.67% (10/15) expressed interest in future use, while 80% (12/15) of the monocular group reported a clear intention to continue engaging with the exergames.

In this study, we evaluated the impact of exercise interventions using 3 modalities—traditional offline exergames, Kinect-based exergames, and monocular camera–based exergames—on exercise performance and user experience in a high-risk T2D population. The findings revealed several noteworthy insights into the potential of monocular camera–based exergames as an affordable and convenient alternative to Kinect-based systems in developing exergame for high-risk T2D population.

From an exercise performance perspective, all 3 modalities—traditional, Kinect-based, and monocular camera–based exergames—elicited comparable elevations in participants’ heart rates, indicating similar levels of cardiovascular engagement. This finding aligns with prior research demonstrating that exergames can effectively promote physical activity across diverse technological platforms [23]. While the heart rate differences did not reach statistical significance, the effect size suggests a small practical effect that may still be meaningful in public health applications, especially for preventive care in at-risk populations.

Notably, participants reported lower perceived fatigue when engaging with digital exergames, particularly the monocular camera–based system, and although these differences did not reach statistical significance, the moderate effect size and significant pairwise difference between the monocular and control groups suggest a clinically relevant trend. This trend suggests that digital interventions may offer a more comfortable exercise experience while maintaining effectiveness, corroborating previous studies that have highlighted the potential of exergames to enhance exercise enjoyment and adherence [24]. The observed lower perceived fatigue in the monocular camera–based exergame group may be attributed to the system’s ability to provide real-time feedback and adjust to users’ movements, thereby enhancing engagement and reducing the subjective experience of exertion. Such features are consistent with findings from studies emphasizing the importance of interactive and adaptive elements in exergame design to optimize user experience and outcomes [25].

In terms of user experience, digital exergames were found to foster higher levels of inner motivation than traditional offline exergames, highlighting their potential to engage users more effectively [26,27]. The difference in inner motivation showed a large effect size, indicating the robust motivational potential of digital exercise systems for behavior change. Although there was no significant difference in inner motivation between Kinect-based and monocular camera–based exergames, these 2 systems exhibited distinct strengths in the user experience. Monocular camera–based exergames excelled in providing sensory and imaginative immersion, flow, competence, and positive affect, offering an engaging and enjoyable exercise environment [28]. In contrast, Kinect-based exergames performed better in the dimensions of tension, negative affect, and challenge, with the challenge dimension showing a statistically significant advantage, suggesting Kinect’s ability to reduce negative emotional responses during gameplay [29]. The higher challenge scores in the monocular group may reflect user frustration stemming from occasional inaccuracies in motion tracking or delayed feedback, which can reduce the sense of control and increase cognitive effort.

To enhance the long-term appeal and sustain user engagement, current research indicating incorporating adaptive difficulty levels, personalization features, and reinforcement mechanisms is crucial [30]. Adaptive difficulty can dynamically adjust the game’s challenge to match the user’s skill level, preventing boredom and frustration. Reinforcement mechanisms, including real-time feedback and rewards, can further encourage consistent participation [31]. Future design may enhance these design elements to improve adherence and motivation in exergame interventions.

Nevertheless, participants using the monocular camera system reported fewer setup difficulties and a higher willingness to continue using the system postintervention. Compared with Kinect, the monocular system’s low hardware burden, environmental adaptability, and plug-and-play configuration make it highly promising for home-based and scalable interventions. These findings highlight the practical potential of monocular camera–based exergames as an accessible solution for promoting physical activity in resource-limited settings, where traditional fitness infrastructure or expensive hardware such as Kinect may not be feasible.

Despite encouraging results, this study has several limitations. First, the small sample size and short follow-up duration limit generalizability. Future studies should include larger, more demographically diverse populations. Second, the controlled indoor setup does not fully reflect real-world conditions; testing in-home or community settings is needed to assess ecological validity. Third, the study did not examine long-term adherence or health outcomes (eg, HbA_1c, weight, and insulin sensitivity), which are critical for evaluating the sustained impact of digital interventions. Fourth, although Kinect-based exergames scored higher in the challenge dimension, participant feedback suggested that the monocular system’s challenge scores may have resulted from tracking errors or interface latency, rather than intentional game difficulty—indicating a need to distinguish between technical and design-related difficulty. Fifth, the study did not conduct subgroup analyses based on gender, which could influence user experience and outcomes due to differences in physical ability, digital familiarity, or motivational preferences. Sixth, the study was originally planned to be conducted at a different location; due to logistical challenges, it was relocated to another community site, which may affect comparability with the original protocol.

Future research should evaluate long-term health outcomes such as HbA_1c, insulin sensitivity, and weight control associated with these interventions; conduct trials in real-world environments such as community centers or homes to assess ecological validity; investigate personalized game adaptation strategies that adjust challenge and feedback based on individual skill level and fatigue patterns; and develop cross-platform, low-barrier versions of monocular camera–based systems for scalable deployment in resource-limited settings.

The findings suggest that monocular camera–based exergame offers a viable alternative to Kinect-based exergame for at-home use by T2D high-risk populations. It provides similar exercise outcomes and user experiences while offering the advantages of simplicity and greater adaptability. However, further improvements are needed to enhance user experience, particularly in balancing the challenge dimension.