The methods (including app design, recruitment, data collection, and data analysis) have been comprehensively described in a previous publication [32]. In reporting of methods, the 32-item COREQ (Consolidated Criteria for Reporting Qualitative Research) checklist has been used (Checklist 1) [33].

The AR app was developed to support an individualized exercise program for children and young people undergoing cancer treatment, particularly when face-to-face provision was not possible, the child was at home, or they were immunocompromised. Its design was guided by an interdisciplinary team of software developers, digital health researchers, and pediatric oncology exercise professionals. Key features were implemented with the needs of children with cancer in mind, including professional input during onboarding to record diagnosis and contraindications, child-reported fatigue and activity levels, and a pre-session questionnaire to screen for new symptoms. The app offers upper body, lower body, and core exercises with seated and standing options, and a rate of perceived exertion scale that automatically adjusts exercise volume if intensity is inappropriate.

Participants were recruited through schools and universities. Eligibility criteria are presented in Textbox 1.

Participants were recruited through convenience sampling of age-matched children and young people, representing a mix of rural and urban schools, along with 1 university group. During the recruitment of primary school children, the research team liaised with the school and its teachers, who passed on information about the study to potential participants and legal guardians. Participants were told why the research was being conducted and that the app had been developed for children with cancer in mind. For secondary school pupils, members of the team visited the school to explain the purpose of the study and outline what involvement in the workshops and focus groups would entail. For university students, the team provided details about the study by attending a selection of lectures.

This research has received ethical approval from the Oxford Brookes University Faculty of Health and Life Sciences Research Ethics Committee (David E. Evans, Paul Hough, Robyn Curtis, and Jo Brett; approval 211547). Written informed consent and assent, including parental consent where applicable, were required to participate in the project in accordance with the project approvals granted by Oxford Brookes University Faculty of Health and Life Sciences Research Ethics Committee (registration 211547). All data were fully anonymized before analysis, with any identifying information removed to ensure participant confidentiality. Participation was voluntary, and individuals could withdraw at any time without penalty. Participants did not receive any compensation for their involvement.

Before the workshops, non–user-facing onboarding steps were completed to ensure that participants could access the main functions of the app, even though they did not have a cancer diagnosis. This setup involved selecting the following default options: all body areas were eligible for exercise, activities were not limited to seated movements, participants did not use a prosthetic limb, and they did not experience peripheral neuropathy. At the beginning of each workshop, participants were told why the research was being conducted and that the app had been developed for children with cancer in mind. None of the individuals invited to take part had used or interacted with the AR app.

Participants were given a phone and then progressed through the user-facing onboarding sequence. Once this was completed, participants completed an individualized exercise session, following the AR avatar demonstrations (Figure 1). Each workout consisted of 6 exercises, with the number of repetitions (1-6) and sets (1-3) automatically adjusted by the app’s algorithm based on responses provided during the onboarding process. Following the session, participants completed a postsession questionnaire, which included a rating of perceived exertion measure (Figure 2).

Subsequently, participants took part in focus group discussions to explore the user experience of using the AR app and had the opportunity to share suggested improvements. For the focus groups, a semistructured interview guide was used (Table 1).

Workshops were conducted by KS (PhD, postdoctoral research assistant, female) and HM (MSc [Res], postgraduate research assistant, female), and SW (PhD, postdoctoral researcher and senior lecturer, male, qualitative research and technology expert), and focus groups were conducted by KS and HM. The research team had previous experience in conducting mixed method and qualitative research and specific training as part of their academic qualifications. Focus groups were audio-recorded and transcribed.

Edited transcriptions were printed and reviewed in paper format, with 3 researchers (KS, SW, and HM) collaboratively conducting a coding process. Analysis combined deductive and inductive approaches [34-36]. The deductive component applied a pre-established framework based on key topics—hardware user experience, app interface, app design, AR functionality, avatar design, and exercise prescription. User experiences and suggested improvements were considered separately. Any remaining data from the discussion were inductively coded through intercoder conversations [32].

Features mentioned as suggested improvements were mapped onto the mechanisms and design elements outlined in the taxonomy of motivational affordances (Figure 3) [24]. Applying the taxonomy ensured relevant features were mapped onto evidence-based strategies and a unified terminology was used, which will aid the development and evaluation of future tools.

Moreover, 2 coders (HM and KS) independently mapped the features to the taxonomy to enhance the validity of the coding process. Subsequently, “critical friend” discussions were held to explore the suitability of the respective selections and to encourage reflexivity and enhance the dependability and credibility of data [38]. This approach also encouraged KS and HM to reflect upon their coding and provided room for considering alternative interpretations [38].

In total, 19 primary school pupils, 28 secondary school pupils, and 3 undergraduate (UG) university students took part in the practical workshops, with 39 providing consent to take part in the subsequent gender-mixed focus groups. These were conducted in the respective educational facilities (eg, school classroom). There were 8 focus groups in total—3 with primary school students, 4 with secondary school students, and 1 with university students. Workshops and focus groups were conducted between June and December 2021 at one primary school in Gloucestershire, one secondary school in Oxfordshire, and one university in Oxfordshire (United Kingdom). Participant ages and year groups are presented in Table 2.

Participants were able to test the app for 20‐30 minutes before taking part in semistructured focus groups, which lasted between 16 and 28 minutes.

Throughout the focus groups, participants voiced suggestions for future improvements of the app. Table 3 contains an overview of suggested improvements linked to the following motivational affordance mechanisms: user education, challenges, feedback, cooperation, and comparison.

In addition to these improvements, participants discussed options for incorporating gamification within the app, which was mentioned to hold potential to attract users who usually do not enjoy exercise. One participant added that “[...] rather than playing games [...], [they] would be doing exercise” (Y7). They felt it could be useful to use young people’s interest in computer gaming or other activities and amalgamate this with the exercise-related features of an app. Table 4 contains an overview of gamified features proposed by participants, linked to challenges and rewards as per the motivational affordance mechanisms.

In addition to the suggested improvements shown in Tables 3 and 4, and outside of the scope of features related to motivational affordances, participants highlighted that equipment for using the app should be user-friendly. For example, phone screens should be of sufficient size (>6 inches), and tripods should be sturdy and easy to adjust. Moreover, customization of the app interface (eg, by giving options to change background colors), or of the avatar (eg, by allowing users to change its features, such as age, gender, skin color, hair, body composition, and clothes, or giving nonhuman options, such as, animals or fantasy characters) would be useful and could enhance engagement with the exercise. It was also suggested that allowing users to play their own music or select from a choice of songs could be motivating.

This study explored user experiences of an AR exercise app and mapped users’ suggested improvements against motivational affordance mechanisms and design elements [24]. Most participants reported that they enjoyed using the app. Participants found the demonstrations and varied exercises useful but expressed some concerns regarding data safety and functionality of the AR avatar. Suggestions included improvements, such as additional educational components, incorporating challenges and rewards, as well as a customizable avatar and a social support feature, such as the possibility of exercising with family or friends. Additionally, some suggestions, such as the integration of audio instructions for a greater inclusive design, were made. Finally, users also expressed concerns about personal data safety. When mapped against the motivational affordances taxonomy, the suggested improvements aligned with mechanisms of user education, challenges, feedback, cooperation, and comparison. This study underlines the importance of incorporating a variety of motivational affordances to account for different user needs and preferences.

Within this study, participants’ suggestions frequently mapped against the taxonomy of motivational affordances [24]. It is promising that motivational affordances incorporated within the current app, such as user education on how to perform exercise, seem to be particularly effective in improving physical activity levels among patients with cancer in other research [39]. Participants in our study also suggested new features, such as audio instructions, visual progress bars, and the ability to connect with other users, highlighting the need to further develop design features, such as achievements, credits, and friends [24].

While current studies may assess the general usefulness of mobile apps, a deeper exploration is needed to determine how specific features drive engagement and thus encourage long-term physical activity, especially in populations, such as pediatric oncology, which we know face additional barriers to exercise. Currently, we may only understand the usefulness of motivational affordances, and further research should delve deeper into the mechanisms through which the features work, for example, via belief in capabilities and intentions [40]. Investigating these underlying mechanisms could help optimize app design and ensure that interventions are effectively promoting behavior change, and not only short-term app use. Long-term engagement is particularly important for survivors of children with cancer as physical activity levels are often lower than noncancer controls [41], and observational studies have identified associations between sedentary behavior and adverse effects on cardiometabolic and bone health [42].

For end users to use and positively benefit from an app, they must actively engage with it. Our findings indicated that incorporating challenges and rewards is preferred, which is in line with other app research [43]. However, it has to be considered that when using the app within clinical populations, there may be contraindications (eg, treatment-related side effects) that prohibit users from engaging with the app, or some of its features, for a period of time. When programming the app to provide challenges and rewards, such as continuous usage streaks or increases in exercise prescriptions, it is crucial to also incorporate rewards for the performance of tasks achievable by all users. Additionally, if participation is affected by a reason outside of the users’ control (eg, treatment), they should not be negatively affected (eg, losing a daily streak or points). The approach of personalized gamification has been highlighted by Mahmoudi et al [44], with the addition of customization, where users can select elements they wish to use. Customization of the app interface and avatar was a frequent suggestion within this study. Avatar-based interventions, and specifically customizable avatars, have been shown to promote health-related behavior change, including exercise [45]. Within the motivational affordance taxonomy, there is no clearly defined mechanism dedicated to “customization” [24]. While customization as a reward or as a transactional element involving earned coins may be conceptually aligned with the motivational affordances design features of credits or achievements, there is no explicit mechanism pointing toward users customizing features (eg, interface or avatar) without first fulfilling certain criteria or earning that capability. While customization may be a popular demand, future research should explore whether being able to personalize apps is motivating in itself or whether the motivation is attached to the act of “earning” the reward in order to do so.

The results showed that having a social support feature, such as being able to exercise with family members or friends, may be beneficial for app engagement and links well to friends, teams, and group mechanisms within the taxonomy of motivational affordances [24].

Family apps have been seen to have promising effects on physical activity levels and psychosocial outcomes by promoting child-parent exercise [46]. Including social features in exercise apps for children with cancer may enhance engagement and motivation. One study [47] found that physical and social activities merged when ambassadors (ie, peers or siblings) participated alongside survivors of childhood cancer, supporting enjoyment and adherence. Integrating peer interactions into digital platforms could replicate these benefits, promoting physical activity and well-being. While the current app does not have this feature due to the restrictions of data protection and safety, this concept may be of interest when considering the isolation and reduced opportunities for social interaction often experienced by children with cancer [48].

Interestingly, 1 participant revealed a different perspective and highlighted that the app could be a useful tool for those who do not have family members or friends who exercise. It was mentioned that the AR avatar could serve as a form of “digital friend” who could positively influence exercise participation, which was in line with findings from a study conducted by Thorsteinsson et al [47]. There is some initial research on a sense of belonging and digital support received from apps that can positively influence physical activity levels [49]; however, the findings of this study underline the importance of exploring this concept more comprehensively.

The results highlight a need to enhance accessibility within mHealth applications, particularly through the extension of existing motivational affordance mechanisms, such as assignments, to better accommodate diverse user needs and promote inclusive design. For instance, the inclusion of a text-to-speech function was recommended, which would provide meaningful support for users with reading difficulties or visual impairments. This recommendation aligns with existing research emphasizing the importance of designing inclusive digital experiences [50]. The evidence highlights the importance of accessibility features, such as voice control, voice over, color contrast, and easy access controls, to ensure apps are usable for individuals with disabilities [50]. These features are particularly relevant in physical activity apps, where ease of use and clear instructions are essential for safe engagement. For example, individuals with visual impairments may rely on voice commands and audio feedback to navigate exercise programs, while those with motor disabilities may benefit from customizable controls that allow for seamless interaction. Therefore, app developers should consider the inclusion of such features to ensure inclusivity and enhance user satisfaction across diverse populations.

Results highlight concerns regarding the appropriateness of the app collecting and potentially sharing personal data (eg, age). It is important to recognize that the use of an app that requires any user input should be accompanied by relevant training and information for users. It is expected that, by providing extensive information to users and their parents or guardians, concerns may be mitigated effectively, with users satisfied and subsequently willing to provide accurate information. It has been widely acknowledged that the increased volume of data available through the use of mHealth tools, such as apps, can indeed be used to provide better services if current data protection and sharing regulations are adhered to [22]. Looking ahead, due to the frequent multicomponent treatment approach for children and young people with cancer, it would be crucial to explore how user data within apps (eg, rate of perceived exertion and exercise duration and intensity) can be integrated within digital health records for a more comprehensive treatment approach. As part of such evaluations, specific mHealth reporting and assessment guidelines, such as the mHealth Evidence Reporting and Assessment checklist, should be used [51].

This study provides a rich and in-depth understanding of user experiences when using the app, allowing researchers to study potential unexpected patterns in how users interact with the different app features. Mapping results with motivational affordances allows for a better understanding of certain features that might be more motivating and support app engagement within this population. Our findings can inform the development of effective mHealth tools that support behavior change in the field of exercise and physical activity as well as guide future interventions, including more personalized, user-centered approaches to digital health technology.

Despite our study’s strengths, it is important to acknowledge limitations. First, this study recruited healthy participants. While some perceptions of the app’s functionality are expected to be comparable between patients with cancer and healthy age-matched peers, there may be differences in the usability of the app between these 2 populations. Some pediatric patients with cancer may have a unique motivation with regards to exercise and physical activity, as well as different physical and cognitive needs and preferences; user experiences of some features could differ. Currently, the AR app is at technology readiness level 5. Future studies with the intended clinical population would progress the app to technology readiness level 6.

Second, the study targeted a broad age range (8‐21 y), while the app itself remained largely static across all ages. While differences in cognitive ability, physical capability, accessibility, and design preferences may mean that some features are less engaging for certain age groups, given the lack of existing AR apps for this population. This approach is also pragmatic, considering the high costs associated with technology development, allowing initial feedback to inform improvements before investing in age-specific customizations.

Third, this study focused on short-term app interaction. Given that the app is designed to support exercise prescriptions over several weeks, the brief duration prevented the assessment of features influencing longer-term adherence. Although the controlled setting allowed for the collection of immediate feedback, it did not capture the evolving experiences users may have during extended use. Future studies should incorporate longitudinal testing to better understand long-term engagement and identify which features support ongoing use.

Finally, there was also the potential of researcher bias in the study, as the investigators had a background in sport and exercise science, which may have influenced how data were interpreted or which themes were prioritized.

Despite these limitations, the study provides valuable foundational evidence to guide improvements to the app and inform future research on implementing mHealth tools within pediatric exercise interventions.

Theory-driven evidence for best practice on mHealth apps is often inconsistent or not focused on the overall user experience. This study explored young people’s experiences of an AR exercise app and mapped suggested improvements onto the motivational affordance mechanisms and design elements. Users identified several ways the app could be strengthened, including the addition of educational content, challenges and rewards, a customizable avatar, and social features that enable exercising with family or friends. Suggestions also highlighted the need for greater inclusivity through features, such as audio instructions, and clear safeguards around the security of personal data. Mapping these recommendations onto the motivational affordance taxonomy revealed alignment with mechanisms, such as assignments, achievements, friends, groups, and credits, underscoring the diversity of motivational pathways that can support engagement. Further research directly involving children and young people with a childhood cancer diagnosis is planned to further explore the preferences of those facing unique barriers to physical activity and exercise. Such work will also be crucial for advancing the technology readiness level of the intervention by testing the app with the intended population in real-world settings.