As the first step in this study, we researched existing knowledge on XR-based rehabilitation technologies and frameworks gained through an explorative literature review (Adams et al., 2007) and market analysis. This allowed us to understand and frame the problem using different viewpoints.

A mini-literature review was undertaken, accessing the databases of Google Scholar, PubMed, and Elsevier were searched with the search query in February 2021:

Further, only peer-reviewed English articles published after 2004 were included. For articles related to general rehabilitation, no time threshold was set. Relevant additional literature found during extraction was added, too. After scanning the abstracts, evaluating the full article, removing duplicates, scoping, and systematic reviews, the final selection amounted to 25 publications. We identified four reoccurring themes across the literature—efficacy, motivation, ownership, and technology—and used those to structure the brief review presented below.

As part of the study to discover how XR-based exergames can play a part in home rehabilitation for lower limb function and how this can be incorporated into the rehabilitation ecosystem, we interviewed Subject Matter Experts, therapists, and patients. This way, we could identify gaps and differences in the parties’ perceived realities. The questions for patients aimed to capture descriptive experience instead of normative recitals. This approach came with a preconceived motive to obtain emotional responses to challenge perceptions from the therapist interviews. For example, therapists made claims that sometimes contradicted the patients’ experiences. At the same time, the questions sought to uncover experiences with logistical and systemic factors, such as specific interactions with healthcare institutions.

The interviews and subsequent data analysis were carried out by two authors, who were in their final year of graduate studies in industrial design engineering at the Norwegian University of Science and Technology. They received supervision from the other two authors, one of whom was pursuing a Ph.D. in medical technology with a specialization in neurorehabilitation, while the other was an associate professor in design with a focus on human-computer interaction and industrial design.

We separately interviewed three Subject Matter Experts (SME) about the challenges of ABI rehabilitation. One of the interviewees was a researcher in motor rehabilitation post-stroke and a practicing physiotherapist, and two were researchers in exergames for motor rehabilitation and former physiotherapists. All three SMEs had an average research experience in motor rehabilitation and exergaming of 8 years (SD: 3.5 years).

The physiotherapist was interviewed about 1) how to interact with patients, 2) how to measure rehabilitation progress, 3) how home rehabilitation works, and 4) their thoughts about XR in rehabilitation.

The two experts on exergames were interviewed about different topics related to exergames, such as 1) exergames used in rehabilitation, 2) XR used in rehabilitation, 3) how to test exergame prototypes with patients, 4) how to measure progression, and 5) how to design for motivation.

The themes we identified during the literature analysis and interviews with the SMEs formed the basis for creating interview guides for both patients and therapists. These guides helped us explore these themes in depth and gather insights from those involved in rehabilitation.

Three female and two male therapists, who were recruited via rehabilitation centers, were interviewed. Two were occupational therapists, and three were physiotherapists. They had an average experience of 18 years (SD: 8.4 years) working in their fields.

The purpose of the interviews was to get a foundational understanding of how therapists work with patients with ABIs. Form and frequency of feedback and follow-up, goal setting, and communication were the focus of our probes. We were also curious about the challenges therapists encounter in handling logistical and technological aspects of their work. Lastly, the therapists spoke of their experiences and thoughts on XR and exergames and if they were familiar with the related concepts and practices.

The themes from the therapist interviews informed our subsequent patient interviews. Our goal during patient interviews was to gain insights into the individuals’ personal experiences and pathways.

We interviewed three male ABI patients. The average time passed since injury was 14 months (SD: 3.9 months), and they were all in their 50 s or 60 s. Two of them were from the same facilities as the therapists, recruited with the help of their care personnel. One ABI patient reached out through an open post published on Facebook. All participants taking part in the study were living in Norway. They had all received an information letter about the details of the study before recruitment. A summary of interview themes and subjects was sent to the patients beforehand for preparation, expanding on information from the initial information letter.

As ABI patients’ cognitive capacity and capability can be limited, the interviews had to be adapted to avoid unnecessary strain. Therefore, an upper time limit of 30 min was set following the advice of therapists.

The patients were interviewed about 1) their injury, 2) rehabilitation details, 3) rehabilitation motivation and challenges, and 4) their thoughts about using technology and XR in rehabilitation.

The study was approved by the Sikt—Norwegian Agency for Shared Services in Education and Research (reference number: 250396) and adhered to the declaration of Helsinki. The participants were informed they could refrain from the study without any consequences. Due to the ongoing COVID-19 pandemic, all the interviews were conducted and recorded in Microsoft Teams, then transcribed and summarized in short-form notes.

The pandemic also constraint the number of included study participants, due to the high evaluation cost, considering the targeted interview groups. Although their number might seem too low to obtain relevant statistical power, based on the mathematical model of Nielsen and Landauer (1993), their number suffices for heuristic evaluation at the given costs.

After conducting the interviews, the following steps were to transfer the recordings to a secure server (NTNU NICE-1) and transcribe them into text for further processing using NVivo. Here, we applied affinity mapping. We performed a comprehensive thematic analysis following the six-step data analysis process model by (Braun and Clarke, 2012): familiarization, coding, generating themes, reviewing themes, defining and naming themes, and writing up. We familiarized ourselves with the interview data through individual readings. We systematically coded key insights from these interviews in NVivo, marking each interview cue. The initial identification and subsequent reviewing of themes was a collaborative effort within the research team, with the aid of digital tools like Miro, as well as physical whiteboards and sticky notes to facilitate the process. We iterated on the clusters to arrive at meaningful, bounded themes, finally naming them at the point where they appeared mutually exclusive. Final theme names were then used to structure our design recommendations.

It is important to note that our approach to analysis differed between the literature review and SME interviews on the one hand and therapist and patient interviews on the other. The first two informed and structured our research, while the latter constituted the primary data collection and thematic analysis phase.

At the time of the data collection (June 2021), the commercial XR-based rehabilitation tools listed in Table 1 were available. This shows that the market has reacted to the availability of XR technologies. Still, as two of the five products have already ceased, the difficulty of providing a functioning and desirable product is underlined. We further gave an overview on their respectively targeted patient group (stroke: 5 out 5, TBI:1 out of 5), whether the XR hardware used was commercially available (2 out of 5) or specially built (2 out of 5) for the application if any supervision is needed and the product thus more applicable for clinical applications (2 out of 5), how the delivery of training is designed, what the feasibility, how the feasibility can be assessed and its viability for home use. That information was based on the manufacturer’s publications.

During the thematic analysis of the stakeholder interviews, several themes related to XR-based exergames for ABI rehabilitation emerged. These are presented below. Quotes from interviewees were translated by the authors from Norwegian to English.

Concerns about the safety of new technologies were clearly mentioned in the interviews and related literature and are essential for the acceptance of new rehabilitation tools by all involved stakeholders (Jordan and King, 2011). However, only a few studies are concerned with the safety recommendations of XR provided by head-mounted displays (Jordan and King, 2011; Maggio et al., 2019). Unsafe training procedures can lead to adverse effects like pain, fatigue, dizziness, and falls (Morone et al., 2014; Rüth et al., 2023).

Our results indicated that especially the patient’s disabilities and their lack of ability to judge their capabilities lead to increased risks of injury. As this is already challenging for the patients themself, understanding their limitations and identifying potential risks is even more difficult for non-matter experts, such as game developers. The individuality of the patient’s disabilities and their change over time increase the difficulty for the developers even further when creating safe exergames. Therefore, the possibility of initial individualization and ongoing adaptation to the functional capabilities should be a central part of the game mechanics. Such customizations, together with good onboarding, are thought to increase accessibility via safety (Jordan and King, 2011). This can be done by finding the delicate balance between being challenging enough to promote functional improvement and easy enough to avoid injury and demotivation.

Besides the patient’s current stage, the environment for which the exergames are intended is also an important safety factor. On one hand, do XR technologies allow for the training of potentially dangerous real-life situations, like participation in traffic, in the safety of one’s home (Pietrzak et al., 2014). On the other hand, this could also create other risks. Home training and related exergames are often unsupervised and occur in suboptimal environments. The use of physical safety equipment, such as safety harnesses, handrails, or chairs, could be incorporated to promote safety if they do not disrupt other central design recommendations. Besides using safety equipment, Maggio et al. (2019) indicated that familiarity with the hardware and its control is a crucial element that can convey safety without the need for supervision and increase acceptability. Using such additional safety equipment, immersive VR exergames seem to be considered safe, even for home use (Broeren et al., 2008; Darekar et al., 2015; Tuena et al., 2020). So could the little-discussed application of AR add to safety during exergaming by allowing the patient to keep the relation to the physical space around him, which can avoid collision accidents or be used as additional support.

The rehabilitation process for individuals with ABI is highly individual due to the varying nature of the resulting functional deficits and varying personal preferences. The rehabilitation process must be adapted correctly to ensure better safety, motivation, and functional outcomes (Krishnan et al., 2023). As the XR technologies allow for the simulation of a wide variety of real-life environments, the modification of sensory presentations and feedback, and the adaption of task complexity, they can seemingly fulfill those requirements (Pietrzak et al., 2014). To make such an adaptation viable in clinical practice, creating an extensive library of exercises, easily individualizable, modular games, or applying machine learning algorithms automatically adapting to the patient’s capabilities should be considered (Pirovano et al., 2013; Muñoz et al., 2019). The first two suggestions, however, need external intervention to choose the correct exercises, which can be impractical for home training. Machine learning-based approaches could assess the movement during training and adapt specific game parameters and feedback to adjust it to the patient’s changing needs automatically, as suggested by SME and literature (Osgouei et al., 2020; Tharatipyakul and Pongnumkul, 2023). Although this technology has its first applications in fitness training, its feasibility in such a complex field as rehabilitation is yet to be systematically assessed.

For a clinical intervention, its efficacy is central for its application to be considered. As mentioned in the literature review, XR exergames showed us to be able to improve motor function and are currently considered a supplementary tool for therapy.

Clear and continuously updated therapy road maps that align with personal goals are paramount for a good progression of the rehabilitation process. Constant qualitative and quantitative measurements of the patient’s performance aid both the therapist and the patient in understanding and communicating deviations from a general baseline and intervening promptly if necessary. Exergames could provide this by allowing for the personalization of individual goals and adaptable level design that considers continuous performance measurements via integrated sensor systems and reports and automatically adapts to the exercise difficulty (Osgouei et al., 2020). Giving comprehensible performance reports alongside the game can increase the patient’s feeling of ownership over the rehabilitation process. By constantly pushing the current boundaries of the user’s abilities to a reasonable degree, functional improvement is facilitated, but overexertion should be avoided. Especially as patients tend to overestimate their own ability, detecting eventual overtraining and promoting rest periods can prevent injuries and stagnation of the recovery (Cho et al., 2023).

Further, the game should prevent maladaptive movement, which can cause insufficient rehabilitation outcomes or even injuries. A certain degree of freedom in executing the movement is acceptable, as it provides a more enjoyable gaming experience; however, for good progress, compensatory movement should be discouraged if possible. This can be facilitated by intelligent game design or performance measurements, like movement trajectory deviations (Alankus and Kelleher, 2012).

Limitations in the technology of current systems also fail to address the needs of therapists and patients. For clinicians, introducing technology in general into rehabilitation routines is met with hesitancy. While patients are interested in incorporating novel technologies into their training, therapists are more conservative in adopting them in their established routines. The main hurdles are the limited cognitive ability of some patients to use complex systems and therapists’ limited resources to get familiar with the systems and set them up. However, these experiences were based on games not tailored for rehabilitation. In contrast, like other at-home rehabilitation technologies, XR has the potential to reduce the resource requirements of motor rehabilitation programs. This can be accounted for by reducing time-consuming transportation, especially for remote patients, and facilitating labor-intensive traditional rehabilitation practices (Pietrzak et al., 2014).

Several aspects must be addressed to scale the use of XR systems in rehabilitation, especially in the home setting. The product design should focus on flexibility and ease of use rather than vast functional capabilities. Accessibility should be the priority, which also applies to the user interfaces, as the end-user might have restrictions or may not even consider trying a new technology. Therefore, changes to commercial XR hardware and setup instructions must be considered, as patients might struggle with the correct attachment of the headset, controllers, and other accessories, as done by some currently available solutions (Table 1). Meanwhile, the underlying technology offers new opportunities that could enhance the enjoyment and efficacy of the gameplay itself. Advances in hardware make personalization mechanisms feasible on smaller systems, reduce costs, and increase gameplay fidelity. Improved connectivity creates opportunities for more seamless social gameplay and remote rehabilitation interventions. No- or low-code platforms could also lower the threshold so game development can become more rapid and widespread, as Baldassarre et al. (2021) demonstrated. Lowering this threshold moves us closer to applying co-creation in the game design process and, thus, applicable applications.

The use of exergames in rehabilitation has been extensively studied due to their inherent positive motivational aspect (Hung et al., 2016; Pacheco et al., 2020). Motivation is directly related to increased training adherence and, therefore, functional improvement (Maclean et al., 2002; Kil and Son, 2020). Due to the immersive nature of XR applications, they are seen as inherently motivational and, therefore, are investigated for their use in rehabilitation (Maggio et al., 2019).

In relation to this, our results indicate that by considering the patient’s individual needs, preferences, skills, and goals, their motivation can be optimized during rehabilitation. The primary concern of therapists is keeping patients motivated throughout their rehabilitation and leveraging their hobbies and interests as key drivers. Frequent goal setting and adjustment with the patient is a critical component and serves as a reference point that can be used to monitor and display progress, which is a motivating factor. Therefore, personalized virtual environments and tasks that are constantly adjusted to the patient’s current state, as well as feedback on the patient’s progress and goals, should be included in the exergames.

After transitioning to the home environment, patients can often become passive and preoccupied with activities of daily living, as indicated by our interviews and the literature review. The lack of further goal setting and display of progress, a frequent activity in the inpatient phase, can lead to decreased motivation. To counteract this, patients expressed the desire for gaming platforms to promote initiation from the user, while therapists emphasized the importance of games promoting independent activity and exercise. Small nudges that keep the patient active and motivated can be deciding in maintaining progress. Incorporating the key drivers—i.e., fishing, biking, and other recreational activities—used in the inpatient phase could reinforce this.

Social belonging and interaction are crucial for the not seldomly isolated patient group, and the importance of incorporating social play was discussed previously (Alankus et al., 2010). However, in the early stages of rehabilitation, simple communication can be challenging due to aphasia, making it difficult for patients to engage socially. As patients progress and become more independent, other complex social components become increasingly important.

Once patients transition to outpatient care, they may experience social isolation. Patients often perceive their social circle at home as a significant source of motivation and support, with family and friends acting as supplementary caregivers in many situations. In cases where ambulant supervision is infrequent or unavailable, the involvement of family and friends becomes even more vital in the rehabilitation process. Providing them with tools to assist the patient’s progress can be beneficial.

Integrating social factors and mechanics into games can be an effective approach. Cooperative gameplay and competitive multiplayer are motivational mechanics in general and specifically in ABI rehabilitation (Sweetser and Wyeth, 2005). Asymmetric gameplay, shared virtual spaces, and in-game communication tools are examples of mechanics that facilitate social interaction in games. Social game mechanics need to align with general social factors to enhance motivation and increase the likelihood of user engagement and participation.

Our research shows that the rehabilitation journey in the aftermath of an ABI involves feeling a loss of agency. This loss is perceived but also literal, as the patient often suffers significant motor and cognitive impairments. The paths of the interviewees, from the hospital to rehabilitation facilities and eventually back home, revealed a narrative of disorganized logistical practices. There was an apparent disconnect between the desires, recommendations, and actual decisions made, causing confusion and disorientation. While the experiences at the rehabilitation facilities were generally positive and productive, the lack of predictability in the process left the patients feeling a diminished sense of control and agency.

Interestingly, the most remarkable accomplishments in terms of logistics were a result of the patients’ independent decision-making. This suggests they could navigate the rehabilitation journey more effectively when given the chance to make independent choices.

Ownership of the activities in the rehabilitation process enhances the sense of agency and engagement, which can significantly contribute to its efficacy (Maier et al., 2019). When patients feel a sense of ownership, they are more likely to be motivated, actively participate, and take responsibility for their recovery.

In a gaming context, fostering ownership is key to patient rehabilitation. It is crucial to give patients a sense of control over their experience and insights into their rehabilitation progress. Thus, by providing in-game customization options, such as adjusting the general game theme and, to some extent, exercises based on their interests and goals, as well as giving understandable metrics on their performance and progression, agency in decision-making can be promoted. Tailoring the game experience to their preferences can increase their engagement and motivation to participate regularly. It is important to note that transparency is paramount in implementing changes like level difficulty and performance feedback to ensure patients experience ownership and avoid confusion and frustration. When done correctly, this feedback loop can reinforce efficacy and progress (Schmid et al., 2016).

We propose several design recommendations for XR-based rehabilitation games based on the emerging themes, the patient journey, and the discussion above. The design recommendations are presented in Table 4, and their source and rationale are indicated for reference. These design recommendations can guide designers and developers of XR-based rehabilitation systems. While certain themes are universally applicable for all exergames in motor rehabilitation, we argue that certain themes carry greater significance when applied to XR-based rehabilitation games compared to other exergame applications. Specifically, we want to highlight five key themes: safety, usability, efficacy, technology, motivation, and ownership, which diverge in some respects from non-XR-based exergames.

Foremost among the considerations of the involved stakeholders is the safety aspect of XR-based systems, as they most notably can impede spatial orientation and increase the risk of injuries, which might be reduced by using AR environments instead of fully immersive VR environments. For all stakeholders involved, safety was of the utmost importance and was pivotal in trying any new solutions. The significance of safety is further emphasized, that besides proving clinical efficacy, new medical applications must prove their safety by complying with national and international regulations (e.g., Directive 2007/47/EC of the European Parliament 2007).

Although the usability of XR technologies has increased significantly over the last few years, they still require a more elaborate and unfamiliar setup than other exergame systems, such as screen-based counterparts. Thus, assistance in setup, calibration, automatic error handling, and general onboarding is essential to appeal to patients and therapists.

Moreover, the XR system offers additional sensors, including position sensors and even hand-tracking capabilities, that can be leveraged to provide realistic virtual interactions and fine-grained performance assessments, increasing clinical relevance.

The realism and customization capabilities inherent in XR technologies enable an elevated degree of personalization within environments tailored to individual users, which the end users specifically emphasized. This, in turn, cultivates heightened adherence and motivation.

Other frameworks and design guidelines have been proposed previously and highlight similar themes crucial for the successful development of exergames. While the majority of these frameworks predominantly target 2D clinical exergames, specific themes are transferable to XR at-home exergames. Further, most frameworks concentrate on adaptation and individualization strategies tailored to the diverse end-user group. This is either achieved through prior intelligent game design that incorporates the user’s capabilities into account or through automatic adaptation algorithms (Hardy et al., 2015; Pirovano et al., 2016; Tadayon et al., 2020). In contrast, the framework by García-Martínez (2015) prioritizes the creation of ownership by providing appropriate feedback on performance and results. Additional recurrent themes were motivation, appropriate use of sensor technologies, the importance of social interactions, the efficacy of the digital intervention, and the sustainability of a solution (García-Martínez et al., 2015; Hardy et al., 2015; Pirovano et al., 2016; Li et al., 2020; Tadayon et al., 2020). Notably, only the framework proposed by Li et al. (2020) focuses explicitly on using head-mounted VR, guiding the design, development, and evaluation of exergames for health in a broader context.

Compared to the previously proposed guidelines, our design recommendations are an important tool for any game developer who wants to understand the central stakeholders beyond mere clinical requirements. Our recommendations extend the scope by prioritizing direct stakeholder interaction to uncover emotional needs. Moreover, by delivering the design recommendations in a brief manner with relevant contextual information, we ensure their accessibility to non-subject matter experts, including game designers or other stakeholders without a clinical background. It is important to note that while our approach facilitates the initial understanding and design phases, it does not replace the necessity for iterative co-design workshops with the central stakeholder. It enhances and accelerates the planning and execution, leading to a more comprehensive, efficient, and sustainable development journey and desirable games for the patients.

Several limitations of this study must be acknowledged. Firstly, due to the restrictions encountered during the COVID-19 pandemic, we could only recruit a limited number of interviewees, all of whom were reached exclusively online. While this approach ensured adherence to health guidelines and timely completion of the interviews, it inevitably narrowed the diversity of perspectives included in our research. Furthermore, it is essential to note that specific stakeholders, like family and friends of patients, game developers, and municipalities, who could have contributed different perspectives, were not included in our study. The absence of these perspectives may result in an incomplete picture of the subject matter.

Secondly, our research is geographically limited to Norway. Although this localized focus seems disadvantageous as it limits the generalizability of our findings to a global audience, this can also be seen as the opposite. The potential to highlight the impact of remoteness of the end-user can aid in understanding the challenges of in-home rehabilitation.

Lastly, it is essential to emphasize that further validation and verification of these findings is advised to ensure the robustness and reliability of our research outcomes and aid in a potential framework for designing in-home XR-based rehabilitation games.