Descriptive statistics, including the mean, SD, median, and rating distribution of the measurement items, were calculated.

The interview audiotapes were transcribed verbatim by the researcher (JC). Two researchers (JC and TC) with backgrounds in human factors and health care informatics were involved in the analysis. The interview transcripts were analyzed using a qualitative content analysis approach with 3 steps [29]. Step 1: the interview transcripts were read several times before being independently analyzed by the 2 researchers. During the initial reading, the 2 researchers read through the interview transcripts while listening to the interview audiotapes to verify the accuracy of the transcripts and to gain an overall sense of the interview content. During the subsequent reading, the 2 researchers read through the interview transcripts in detail to gain a thorough understanding of the interview content. Step 2: an inductive approach was used to analyze the data because this approach allows the identification of codes and themes that answer the research question [30]. During the data analysis process, the first interview transcript was independently analyzed by the 2 researchers. Relevant statements from the interview transcript were identified, and codes were generated, which were then refined and developed into themes. The same process was repeated for each interview transcript, with the codes and themes added, revised, or developed each time. In addition, the frequencies of the statements expressed by the participants were recorded. Step 3: after all the interview transcripts were analyzed, the codes and themes were compared between the 2 researchers. In case of discrepancies, the original interview transcripts were checked, and the discrepancies were resolved through discussion between the 2 researchers until a consensus was reached.

NVivo (version 12; Lumivero) was used for content analysis, which was based on a checklist of consolidated criteria for reporting qualitative research [31].

A total of 87.23 minutes of interview audiotapes were recorded. The interviews lasted between 2.43 and 8.30 (mean 4.15; median 3.88) minutes. The results of the content analysis were categorized into positive and negative perceptions of the exercise system. The details of the qualitative data are presented in the following sections.

This study aimed to investigate the perceptions of stroke survivors regarding the use of an immersive VR-based exercise system for poststroke upper limb exercises. Quantitative and qualitative analyses were used to explore these perceptions. The questionnaire results revealed that the participants recognized the exercise system’s usefulness for rehabilitation, found the exercise system easy to use, maintained a positive attitude toward the exercise system, enjoyed using the exercise system for exercises, and expressed their intention to continue using the exercise system. The interviews further uncovered the following 6 themes related to positive perceptions of the exercise system: (1) ease of use, (2) benefits, (3) enjoyment, (4) assistance, (5) accessibility, and (6) game design. Conversely, 3 themes were associated with negative perceptions: (1) difficulty in handling VR devices, (2) uncomfortable experiences when using VR devices, and (3) monotony. The interview findings may help interpret the results obtained from the questionnaire and provide further understanding of the factors influencing the patients’ acceptance and use of VR-based rehabilitation programs. For example, some patients expressed interest in the highly immersive game scenarios, which might have attracted them to use the exercise system to perform exercises. In contrast, some patients expressed undesirable feelings about wearing the headset, which might have led to a negative perception regarding the exercise system and impeded its use. Moreover, our findings serve as a concrete illustration of established technology acceptance models, such as the Technology Acceptance Model and Unified Theory of Acceptance and Use of Technology model. For example, previous research has underscored perceived usefulness as a key determinant of technology acceptance by users [23,24,32,33]. Our study further elucidated that motor function, daily living skills, cognitive ability, and mood were the main aspects of usefulness valued by the participants. In the following sections, we delve into the questionnaire and interview findings in greater detail.

In contrast to previous studies reporting that individuals with limited prior technology experience are less inclined to accept the technology [8,32,34-37], our findings indicated that a lack of experience with VR may not impede patients’ positive feelings about using the exercise system. The interview responses, which are discussed in the following sections, may help explain this. However, the highly positive perceptions might be a result of selection bias, meaning that participants who were satisfied with the exercise system were more likely to adhere to the VR exercises, whereas those who were unsatisfied with it could have withdrawn from the study. Therefore, the patients’ perceptions of the exercise system might have been overestimated.

Ease of use is a critical perceptual aspect affecting technology acceptance [23,24,32,33]. Most patients who have experienced stroke are older adults who often have limited experience with technology and may face declining physical or cognitive functions, as well as technology anxiety [38-41]. Therefore, ensuring that VR-based rehabilitation exercises are easy to use is particularly important for this demographic to promote acceptance and adoption. In this study, almost all participants reported that it was easy to understand how to play the rehabilitation exercise games and use the VR devices. Although 1 participant initially found the games difficult due to unfamiliarity, the participant quickly adapted and found the games easier after practicing several times. This may be because the games were related to daily life (eg, fishing), and there were no complicated tasks or complex operations with the handheld controller. Consequently, participants could easily become familiar with the rules of the games and the use of the VR devices without much training or practice. Another possible explanation may be that most participants were able to perform the exercises independently. Necessary assistance was provided only to ensure their safety and prevent the development of abnormal movements (eg, shrugging the shoulders) by the researcher (JC), with the guidance of physical therapists [11]. These findings suggest several specific implications for VR-based rehabilitation programs. First, the simplicity of game design, mimicking daily life activities, can significantly enhance usability for patients with stroke. Second, minimal training and practice requirements make it feasible for older adults with limited technology experience to quickly adapt to the exercises. Third, providing tailored support to prevent incorrect movements can further improve patient engagement and safety. Consequently, designing technology-based rehabilitation exercises with these considerations can foster greater acceptance, enhance patient adherence, and ultimately improve rehabilitation outcomes for stroke survivors.

Participants perceived improved upper limb motor function and daily living skills after using the exercise system. Although this exercise system was initially designed for motor rehabilitation, participants perceived improvements in reaction speed, thinking, and mood. Such gains have rarely been reported in previous studies. One possible explanation may be that the repetitive practice led to familiarity with the exercises, thereby helping the participants increase their reaction speed. Furthermore, the setup of the games in the exercise system, such as the 1-minute countdown and real-time feedback on task performance (refer to details of the games in Figures 1-5 and Multimedia Appendix 1), might provide participants with a sense of tension and competition. This, in turn, might “facilitate your thinking and make the brain work hard,” as 1 participant expressed. Another reason may be the advantages of immersive VR; that is, wearing the headset isolated the participants from the physical world, so they could fully focus on the tasks in the games. Moreover, being able to complete the tasks in the games provided the participants with a sense of achievement, which might have elicited positive emotions. As 1 participant described, being able to play games made the participants feel “useful,” and thus, “the feeling of pessimism” has been alleviated. Furthermore, the natural scenes and relaxing background music in the games might have helped relieve the participants’ negative emotions due to the stroke. A participant stated as follows:

The results indicate that the positive experiences associated with these games can help reduce the psychological burden of rehabilitation, making the process feel less like a chore and more like a recreational activity. These findings also suggest that the exercise system possesses the potential to influence multiple dimensions of stroke rehabilitation, which should be considered when designing and implementing such VR-based interventions.

Consistent with previous studies that used video games for rehabilitation [42-46], performing exercises by playing the games in the exercise system was described as “exciting” and “fun” in this study. These findings suggest that the exercise system offers an enjoyable possibility for performing upper limb exercises. The enjoyment factor has specific implications for VR-based rehabilitation programs. First, enjoyable exercises can increase patient motivation and adherence and can consequently lead to more consistent participation in the rehabilitation regimen. This is particularly important for stroke survivors, as sustained engagement is crucial for recovery. Second, incorporating enjoyable elements into the exercise design can help alleviate technology anxiety, making patients more willing to try and continue using the system. Motivational and enjoyable elements can include reward systems that provide a sense of accomplishment, features that facilitate social interaction, and real-time positive feedback and encouragement.

Some participants expressed that the virtual objects in the games guide and encourage them to perform movements that they might not be able to perform in the real world. For example, 1 participant said, “That apple-picking [game], together with that bird, could guide my hand to perform the movements.” The rationale behind this effect may be that learning a motor task is more effective when learners’ attention is focused on achieving specific goals (ie, focusing on external cues) rather than focusing on performing specific movements (ie, focusing internally on the movements themselves) [47-49]. For example, the apple-picking game was designed for wrist flexion and extension exercises. It required the participants to extend their wrists to pick a red apple and flex their wrists to drop it. Picking up and dropping the apple (ie, external focus) was relatively easy for the participants to understand because the instructions were specific and the goal was clear. However, if the participants were told to contract and relax muscles in the wrist and forearm (ie, internal focus), it would have been too abstract to understand and execute the movements. These findings suggest that the objects in virtual environments should be carefully designed because they play a vital role in helping patients learn movements and perform therapeutic tasks.

The participants reported that the exercise system provided extra opportunities to perform rehabilitation exercises, reflecting their need for exercises in addition to conventional rehabilitation therapy. A possible explanation may be that the needs for rehabilitation in patients who have experienced stroke, may not always be fulfilled because of constraints on professional and institutional resources in public hospitals [50,51]. For example, the patients in this study usually had to wait several days before getting an appointment with a physical therapist. The patients could only receive 30 minutes of physical therapy with the guidance of therapists every day during hospitalization [11], which was less than the treatment duration suggested in previous research [52,53].

The games in the exercise system were characterized as “novel”; the virtual scenes and auditory components within the games offered the participants a highly immersive experience, described as “being in a wonderland.” These findings suggest that the virtual environment design and the VR device type might significantly impact user experience, which should be carefully considered when designing VR-based interventions. The implications of these findings are multifaceted. First, the novelty and immersive nature of the virtual environment can greatly enhance user engagement and enjoyment, making the rehabilitation process more appealing to patients. This implies that designers should focus on creating visually rich and interactive environments that capture patients' attention and interest. Second, the type of VR device used can affect the level of immersion and comfort. High-quality VR devices that offer clear visuals, realistic audio, and ergonomic design can prevent discomfort and enhance the overall experience. This suggests that investment in advanced VR hardware could be beneficial. Third, the positive descriptions of the experience indicate that carefully crafted virtual environments can provide a significant psychological boost to patients. Feeling as though they are “in a wonderland” can distract patients from the monotony of repetitive exercises, making the rehabilitation process feel less burdensome and more enjoyable. Fourth, these findings suggest that regular updates and new content should be incorporated to maintain novelty and engagement levels. Stale or repetitive content could undermine the benefits of the initial immersive experience, so ongoing development and refreshing content are crucial.

Negative perceptions were mainly related to using the headset and handheld controller. Wearing the headset was heavy, making participants feel sweaty and stuffy, the cable used to connect the headset to the laptop sometimes limited the free movement of the participants’ heads, and the handheld controllers were heavy for some participants. This is due to the technical limitations of current VR technology. A large headset is necessary to support a high-resolution display, and a physical cable is needed to support the transmission of massive amounts of real-time data at a high speed and low latency. A better user experience with an immersive VR device relies on advances in hardware and software. For example, the development of retina displays can reduce the size of a headset and improve its wearable comfort, and advances in wireless transmission technology can facilitate the development of wireless VR devices to promote users’ freedom of movement in virtual environments [54,55].

A few participants experienced mild and short-lived discomfort with the VR exercise system, such as dizziness and eyestrain, as observed by the researcher. These findings indicate that the exercise system overall was well tolerated by the participants. Nevertheless, factors that may be associated with the side effects of immersive VR should be further identified and minimized in the design and development of immersive VR-based interventions for patients. The implications of these side effects are crucial for the safe implementation of VR-based rehabilitation programs. First, identifying the root causes of discomfort, such as prolonged use, visual strain from screen resolution, or motion sickness due to discrepancies in visual and vestibular input, is essential. Addressing these issues could involve optimizing session durations, improving screen technology, and ensuring smooth and realistic motion tracking. Second, providing clear guidelines and training for patients on how to use the VR system correctly can help minimize discomfort. Educating users on taking regular breaks, adjusting headset fit, and recognizing early signs of discomfort can enhance their overall experience. Third, personalizing the VR experience based on individual tolerance levels can further mitigate adverse effects. Implementing customizable settings that allow patients to adjust brightness, contrast, and motion sensitivity can help tailor the experience to their comfort. Fourth, continuous monitoring and feedback mechanisms can be integrated into the VR system to track patient responses in real time. This would allow for immediate adjustments and support, ensuring that the therapy remains both effective and comfortable.

Although the upper limb exercises were designed as games, repetitive practice can become boring and monotonous for some individuals, potentially lowering their motivation to adhere to the exercise system [56]. This finding underscores the need for further exploration of strategies to sustain patients’ interest in VR-based interventions in the long term. Several strategies can be considered. First, incorporating a variety of game scenarios and levels can keep the exercises fresh and engaging, thereby preventing boredom. For instance, rotating through different themes, such as sports, adventure, and daily activities, can maintain patients' interest. Second, designing exercises with progressively challenging tasks can provide a sense of achievement and keep patients motivated. Third, personalizing exercise routines to match the individual's progress and preferences can make the sessions more enjoyable and relevant. Fourth, integrating social elements, such as multiplayer modes or community challenges, can enhance the fun aspect and encourage patients to stay committed to their rehabilitation.