Physical activity is an important component of a healthy lifestyle. A myriad of health benefits, both physical and psychological, can be achieved with daily physical activity [1]. Physical activity options, however, are limited for people with mobility impairments. Issues with accessibility and transportation are experienced by many individuals wanting to fulfill their daily physical activity needs. Fitness center parking, inaccessible entrances, limited staff knowledge on inclusive programming, and cost of membership and transportation are some of the limiting factors [2]. Equipment such as a stationary arm ergometer can provide a means to exercise at home but may be seen as boring and difficult to adhere to by many [3]. In an effort to combat the tedium of routine exercise, active video gaming (AVG) has become popular.

Replacing sedentary behaviors during leisure time with AVG has been shown to be an enjoyable option for increasing the amount of physical activity acquired each week [4-12]. The promise of AVGs is particularly appealing for people with disabilities, given the high rate of inaccessible features in the built environment and the difficulty in finding fun and engaging activities [5-8]. Because AVG systems are relatively affordable, they also hold promise as a scalable product for promoting improved health outcomes and higher levels of physical activity and fitness among people with disabilities [13]. Several AVG systems have been in the market for the last 2 decades; however, most are not adapted to suit the needs of individuals with mobility impairments. Using a mixed methods study, the usability of a somatosensory square dance system for older adults was examined [14]. The system was developed to support a popular fitness activity and to eliminate some of the common barriers (risk of injury, noise, and space) experienced by older adults. The system was well received by the participants, providing them with a fun fitness activity that could be safely performed indoors in a private space. The participants included a group of older adults (aged 55-68 years); however, specific adaptation for individuals with mobility impairments was not the focus and therefore was not considered. Another study examined the usability of an AVG platform for older adults with various physical and sensory impairments residing in long-term care homes, with 85% having a mobility impairment [15]. After several rounds of usability testing with residents, staff, and family, the feedback indicated that the system was an enjoyable way to engage in physical activity. Although efforts are being made, there continues to be a pressing need to make AVG controllers accessible to people who are unable to stand for long periods, cannot stand on a small platform because of poor balance or extreme obesity, or those who use a wheelchair for all daily activities.

Our engineering design and development team, with the Rehabilitation Engineering Research Center on Interactive Exercise Technologies and Exercise Physiology for People with Disabilities, previously built a proof-of-concept (PoC) wheelchair-accessible adapted gaming controller for the Wii system [16] and our research team tested its usability rating compared with the commercial off-the-shelf (OTS) balance board controller provided with the product [17]. Participants with mobility impairments found the wheelchair-accessible version to be more usable compared with the Wii OTS board (P<.01). The mean usability scores were 71.7 (SD 18.03) for the adapted controller and 32.1 (SD 36.72) for the Wii OTS board (higher scores reflect greater usability). In addition, a significant negative correlation (r=0.692; P<.001) was found between lower extremity function and system usability scores, indicating that for participants with poorer lower extremity function, the adapted board was perceived as more usable. Furthermore, participants reported activity using the adapted board to be enjoyable, achieving light-to-moderate intensity exercise [18].

As a next step, the aim was to transition from a PoC gaming controller to the performance of highly targeted proof-of-product (PoP) activities for the design of a new wheelchair-accessible AVG controller using the expertise and input of engineers, product designers, potential users, and other stakeholders from the community. The goal was to convert from a Wii-only interface to a universal controller for use with all games available to play on a PC.

Following the design and development process, this study aimed to assess the usability of a wheelchair-accessible GAIMplank video game controller for individuals with mobility impairments. Measures of enjoyment, perceived exertion, and qualitative data on user experience were also collected.

To begin our PoP development effort, approximately 30 semistructured interviews were conducted with end users in various market segments to discover use cases that would be used to drive design features, functions, and capabilities. The physical therapy clinic market segment was found to be lacking in financial resources for investment into technology for nonreimbursable activities such as AVG. The market segment for inpatient rehabilitation and retirement facilities and communities was found to have the financial resources for recreational activities but lacked robust end user interest. The universal video game accessory market was found to be much more robust in terms of interest for nontraditional game controllers to facilitate intuitive player movement in virtual reality environments as well as to encourage AVG. Effective entry into this market requires the game controller to (1) be easily transportable and stowable, (2) have a purchase price comparable with other high-end game controllers (<US $300), and (3) be universally compatible with all major gaming platforms.

The design process began based on the PoC prototype developed previously (Figure 1), which used a custom-built aluminum frame, user interface, and load cell sensors combined with repurposed internal electronics from a deconstructed Wii Fit Board to allow for wireless connection to Wii game consoles. On the basis of the market study results, the preliminary PoP design effort targeted 3 areas for improvement, including reducing the weight of the balance board to <6.8 kg, integrating custom electronics for universal compatibility with all major gaming platforms, and optimizing design for low-cost manufacturing.

A redesign of the PoC balance board platform began with the determination of the critical dimensions for use by standing and seated users while also minimizing the footprint and structural support of the weightbearing structure. Early conceptual designs featured novel solutions to these design challenges through the integration of handrails, footrests, and ramps into the platform design (Figure 2). After evaluating a series of finite element models and simplified prototypes, a resin-infused sandwich composite was the material selected for use to reduce the weight of the balance board while maintaining the stiffness necessary to redistribute the user’s weight to a minimum number of load cells.

The redesign of the PoC balance board electronics began with the selection of the embedded processor, load cell measurement integrated circuits, and wired and wireless communication modalities. The user interface components such as the sensitivity adjustments (overall, front and back, right and left) were upgraded from the PoC’s buttons and knobs to an Android app that can easily provide access to any number of different controller settings and functions to optimize the user’s game play experience. A Raspberry Pi Zero was implemented as the primary processor facilitating Bluetooth connection to an Android tablet as well as allowing remote access to data and programming via Wi-Fi. The USB Human Interface Device periphery connection to PCs and gaming console adapters was facilitated using an ARM Cortex-M4 microcontroller configured as a generic joystick device. Load cell measurements were performed using dedicated low-noise 24-bit analog-to-digital microchips.

The final design of the GAIMplank balance board platform (Figure 3) features rims on 3 sides of the board that serve as a physical barrier at the edge of the board. A downward tapered edge along the back of the board, and a slimline ramp, allowed for easy roll-on access (eg, if the user is in a wheelchair). The GAIMplank platform also features yellow high-visibility graphics that provide the user with a sense of their alignment on top of the board. The weight capacity of the GAIMplank platform is ≤270 kg, which is sufficient to support a wide range of users.

The final design of the GAIMplank Android App features real-time bidirectional information flow and access to a large range of customizable settings for real-time adjustments (Figure 4) to the measured load cells’ bias, dead zones, and sensitivity, as well as to allow for on-demand calibration of the GAIMplank’s output based on a player’s measured weight and range of movement and game play mode (joystick vs directional pad).

The final design of GAIMplank electronics includes a custom printed circuit board (PCB) stack and an LED matrix display (Figure 5). The need for this display was discovered during the early developmental testing of prototypes with 3 volunteers recruited from local video gaming clubs, whereby they required instant feedback to build confidence that the board was indeed responding accurately and quickly to their movements during game play. This display is also used to guide the user through the required movements during calibration (step off, step on, move forward and backward, and move left and right). The PCB stack includes a Raspberry Pi as the central processor for the device, controlling wireless and wired communication, the LED matrix, and the measurement of the position of the user’s center of balance at a rate of approximately 50 Hz. The stack also provides robust fixturing for the 22-pin D-Sub connector, which is used to connect to the load cells embedded in the platform via a flat flexible cable, as well as for the mini USB port for a wired connection to the gaming PC for the game console adapter.

To ensure safety during game play, height-adjustable handrails were located on 3 sides of the GAIMplank. In addition, a Microsoft adaptive controller was integrated into the system to facilitate the use of external buttons and triggers used during game play.

Upon arrival at the laboratory, all study procedures were reviewed with the participant. After being given an opportunity to ask questions, the participants provided informed consent. Before the testing began, basic demographic information, along with information regarding minutes of weekday and weekend physical activity, minutes of weekday and weekend video game play, and video game play experience were collected. All procedures for data collection and video game operation were reviewed with the participants before game play. The participants were then settled at the GAIMplank station. Before playing the first game, a calibration procedure was conducted by following a pattern of movements as indicated on the LED matrix.

The GAIMplank was mapped to operate as a typical joystick controller. Leaning movements of the trunk created corresponding actions in the game (ie, lean forward to move character forward and up, lean right to move character right; Figure 6). The participants played a total of 5 preselected video games. The game Feather, a slow-paced flying game, was played first by all participants to introduce them to the movements required for operating the GAIMplank. The order in which the remaining 4 games were played was randomized. To minimize offensive content, the games had an Entertainment Software Rating Board score of E (Everyone), E+10, or Teen. The description of each game played and movements required is provided in Table 1. Each game was played for approximately 5 minutes.

A total of 21 adults (n=15, 71% men and n=6, 29% women) completed the usability testing. The participant characteristics are provided in Table 2. The mean age of the participants was 48.8 (SD 13.8; range 21-73) years, with 57% (12/21) White, 38% (8/21) Black, and 5% (1/21) Asian. All participants had a mobility impairment resulting from a physical disability, including stroke (9/21, 43%), spinal cord injury (3/21, 14%), amputation (3/21, 14%), cerebral palsy (2/21, 9%), spina bifida (2/21, 9%), or other (2/21, 9%). The primary modes of mobility included walking without an assistive device (7/21, 33%), cane (6/21, 29%), prosthetic leg (1/21, 5%), rollator walker (1/21, 5%), and manual wheelchair (6/21, 29%). The participants played in 1 of the following 3 ways: standing (8/21, 38%), seated in a 4-legged chair (7/21, 33%), or seated in their own manual wheelchair (6/21, 29%).

Participants reported enjoyment of leisure time physical activity, were physically active, and varied in their level of video game play (Table 3). All but 2 participants were aware of AVG. The 2 most preferred devices for playing video games included video game consoles and cell phones.

The usability data for each individual participant are reported in Table 2, with the summary group scores in Table 4. Subscale scores (usability and learning) for the SUS usability tool are also reported. The overall SUS score for all participants was 74.8 (SD 14.5), suggesting above average usability for the GAIMplank. On the basis of the adjective ratings for the SUS scale as provided by Bangor et al [35], the participant scores suggested good usability. When broken down by play style group, the average SUS scores indicated that the usability was best for those who played while seated in a chair, followed by those who played standing, and lowest for individuals who played while seated in their own wheelchairs. A similar pattern of scores was seen with the HITUES scale, with an average score of 4.3 (SD 0.3) for all participants. The HITUES scores can range from 1 to 5, with a higher score indicating greater usability.

RPE and enjoyment scores are reported for each game by play style group in Table 5. RPE scores ranged from light to moderate intensity, with the highest scores for those who played while seated in a chair. Participants rated their experience with playing each game from above average to very enjoyable.

The task self-efficacy questions were used to gauge the participants’ self-efficacy in performing certain video game play tasks using the GAIMplank system. Above average scores were reported by all 3 groups for each task (Table 6). It appears that one exposure to video game play with the GAIMplank did not diminish their task self-efficacy, and the participants felt confident in their ability to complete game video play tasks using the GAIMplank system.

During data collection for several participants (approximately 50%), technical difficulties with the GAIMplank controller occurred. The most frequent were calibration issues. During these sessions, the calibration procedure did not work as intended resulting in an inability of the device to find center, and causing difficulty for players to produce the necessary actions for game play. Recalibration and adjustments to the sensitivity and bias were attempted, but the problem was not corrected. Adjustments to the participant’s positioning on the board were also attempted, but these were not successful. Sometimes it was just one particular game during a session that would not respond to the participant’s movement. A pattern to these difficulties was not detected, and usually during the next testing session the controller worked appropriately. Specific issues based on play style (wheelchair, seated in a 4-legged chair, or standing) were not evident, except for participants who played standing and were not able to distribute their weight evenly. It appeared that the calibration procedure may not have accounted for this shift in body weight. Owing to time constraints, sufficient time was not always available to spend on troubleshooting during participant visits. Although enjoyment and RPE scores were likely affected by technical issues, participants still reported having fun and felt confident in their ability to use the system. Participants also recognized the system as providing an opportunity, typically not available to them, to engage in AVG.

Physical activity options are limited for people with mobility impairments. AVG, also known as exergaming, has the potential to provide a fun and engaging way to increase daily physical activity or reduce sedentary time. This project aimed to develop an AVG controller that was accessible and usable by individuals with mobility impairments. A sample of 21 adults with various mobility impairments was able to successfully access and play a series of standard PC video games using trunk and body movements to produce game play action. Participant usability scores (SUS and HITUES) and qualitative feedback indicated above average usability for the GAIMplank system. A previous study adapted the Wii Fit balance board controller and demonstrated successful use among people with mobility impairments with an average SUS score of 71.7 (SD 18.03) [17]. Physiological testing during game play on the adapted board indicated light-to-moderate intensity exercise for some participants [18]. This previously adapted gaming board controller was limited to use with only Wii Fit games, whereas the new GAIMplank system includes several new options for playing AVGs among people with mobility impairments and allows use with all PC games. The GAIMplank system provides a means for making games that are typically sedentary, using only the hands and fingers, more active by incorporating larger body movements. Furthermore, participants reported game play enjoyment and for some their first opportunity to engage in AVG owing to accessibility issues with OTS systems.

Looking at the average SUS scores for each participant, the lowest score was 40, well below what is considered acceptable usability. Similarly, the participant’s HITUES score (4.18) suggested below acceptable usability. The participant was a regular video game player (120 minutes per week), played seated in his wheelchair, reported moderate level enjoyment across games, game play was rated as a light-intensity exercise, and a large drop in self-efficacy was reported at the end of the session for the questions that pertained to reacting fast enough to choose the next action and moving well enough to maintain successful play. His postplay feedback highlighted issues with responsiveness, which he found frustrating. The next lowest SUS score (52.5) was also reported by a wheelchair user, which can be attributed to the GAIMplank malfunctioning, making the calibration off and the games moving slow. Regardless, this participant who was physically active (960 minutes per week) and a frequent video game player (320 minutes per week) reported above average enjoyment scores, game play as light-intensity exercise, with an improvement in overall self-efficacy after using the GAIMplank system. Of the 4 other participants who reported less than optimal usability (62.5-65), 3 (75%) were standing players who were all physically active, with varying levels of weekly video game play. Two players also had low HITUES scores (4.15) and reported moderate enjoyment, light-intensity exercise, and above average self-efficacy. In both cases, technical issues were encountered with the GAIMplank system during the testing session, resulting in poor responsiveness and making the games very difficult to play. The other standing player typically used a rollator to assist with mobility but was able to play without it. His average enjoyment score was high (86) with light-intensity RPE ratings and maximum self-efficacy scores. Although his SUS score was low (62.5), his HITUES score (4.8) was the highest among all participants. These scores may suggest that although the participant perceived general usability (SUS) issues with the system, he recognized the usability of the GAIMplank as an adapted controller to provide an option for active video game play for individuals with mobility impairments.

At the other end of the spectrum, 7 players reported “excellent” usability (SUS ≥85) for the GAIMplank system. Of those, 2 played in their wheelchair, 2 played standing, and 3 played while seated in the chair. For the standing and seated players, the HITUES scores reflected acceptable usability. All 3 seated players (poststroke) spent little to no time playing video games each week but were physically active, with 2 reporting >400 minutes per week of physical activity. The average enjoyment scores were high, the RPE scores varied, and all were reported to have certainty in their ability to conduct various video game play tasks using the GAIMplank system. Both standing participants were physically active (120 minutes per week); one spent a lot of time playing video games each week (630 minutes), whereas the other played very little, if any. Average enjoyment scores were higher for the person who typically spent a lot of time playing video games each week, whereas RPE was higher for the other participant. Both participants reported high self-efficacy in using the GAIMplank system for video game play. Both participants who played seated in their wheelchairs with high SUS scores were physically active (120 minutes per week); one spent a lot of time playing video games each week (210 minutes), whereas the other one played none. The HITUES scores were lower than those of other participants who reported high SUS scores. Both players enjoyed the games, with RPE scores indicating light-to-moderate intensity exercise. The self-efficacy scores were high for both players at the end of the session.

The results demonstrate the usability of a newly developed adapted gaming board for AVG play, but attention is needed to improve the stability of responsiveness and to address issues experienced by different subgroups of individuals with mobility impairments. Considerations for the engineering team during the next phase of development include different weight distribution patterns during game play by individuals wearing a prosthetic leg, varying movements associated with spasticity among certain players, foot placement outside the board (ie, on stool in front) for some seated players, and overall improvement to better accommodate wheelchair play. Overall, the ability of the adapted controller to provide a fun and accessible option for AVG play was clearly recognized by the players, suggesting the need for additional development and research in this area.

The heterogeneous nature of the sample makes it difficult to determine whether different aspects of the GAIMplank better suited one group over another. The number of participants within each play style group was small and consisted of individuals with various mobility impairments. In addition, participants were recruited from a community-based health and fitness center, so all had experience with physical activity and were currently active with reported weekly physical activity minutes ranging from 60 to 960 (mean 375, SD 257) minutes. Therefore, the results cannot be generalized to more sedentary populations. Previous gaming experience was captured, but its influence on system usability or their game play experience could not be teased out. With regard to the specific games used for the sessions, game selection was limited and may not have appealed to some participants. In addition, sufficient time to learn how best to move for each game was not provided. Initially, qualitative feedback data were not collected from participants. As the need for more in-depth user input was recognized, semistructured interviews were conducted to collect richer feedback data. This change occurred after most of the seated players had completed testing; therefore, limited qualitative data were available from this group.

Continued testing and refinement of the GAIMplank system are underway. The usability testing results were discussed with the engineering and design teams, and a new wave of refinements is planned with a focus on improving the stability of the platform. Specifically, elimination of the Raspberry Pi and addition of a custom-fabricated PCB board are being considered. Following further refinement of the system, the plan is to conduct a feasibility study to examine the use of the GAIMplank for an AVG exercise intervention. The outcomes of interest will include various physiological and psychosocial measures. Other factors to be considered include game play selection and preferences, previous gaming experience, participants’ physical activity level, and single versus multiplayer options. In addition, consideration of factors that affect game play-by-play style (eg, seated and standing) needs to be examined. The potential of integrating other gaming controllers is also of interest.

The GAIMplank, an adapted gaming controller, allowed people with various mobility impairments to engage in AVG. Using the GAIMplank system, standard PC video games can be played while sitting, standing, and in a manual wheelchair. Participants reported above average usability for the GAIMplank and found game play to be enjoyable and provide light-to-moderate intensity exercise. Further design and development work to refine the device is recommended.