See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/264879360

Designing engaging, playable games for rehabilitation

Article

CITATIONS READS

46 758

6 authors, including:

James Burke Darryl Keith Charles

Ulster University Ulster University
9 PUBLICATIONS   1,143 CITATIONS    125 PUBLICATIONS   3,127 CITATIONS   

SEE PROFILE SEE PROFILE

Philip J. Morrow Suzanne Mcdonough

Ulster University Royal College of Surgeons in Ireland
146 PUBLICATIONS   2,395 CITATIONS    231 PUBLICATIONS   8,997 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Communication Style and Exercise Compliance in Physiotherapy (CONNECT) View project

PhD Motion Tracking in Digital Images View project

All content following this page was uploaded by Darryl Keith Charles on 22 October 2014.

The user has requested enhancement of the downloaded file.

 Designing engaging, playable games for rehabilitation
J W Burke1, M D J McNeill1, D K Charles1, P J Morrow1, J H Crosbie2, S M McDonough2
1
Computing and Information Engineering, University of Ulster,
Cromore Road, Coleraine, BT52 1SA UK
2
Health and Rehabilitation Sciences Research Institute, University of Ulster,
Shore Road, Jordanstown BT37 0QB UK

[email protected], {mdj.mcneill, dk.charles, pj.morrow, jh.crosbie, s.mcdonough}@ulster.ac.uk

1,2
www.ulster.ac.uk

ABSTRACT
Stroke is a leading cause of severe physical disability and can result in of a range of
impairments, including loss of balance, attention and concentration deficiencies, pain,
weakness and paralysis. This paper reports on the design of serious games for upper limb
rehabilitation following stroke. In particular, we focus on identifying principles of video game
design which are important in the context of rehabilitation and show how these principles can
be implemented. We also report on an evaluation of the games for playability, usability and
engagement.

1. INTRODUCTION
Stroke is the leading cause of long-term disability; the World Health Organisation reports that annually some
5 million people are left permanently disabled through stroke. Stroke can often cause severe physical
disability such as attention deficiency, pain, weakness and paralysis, often on one side of the body. Many
hospital-based rehabilitation programmes focus on rehabilitation of the lower limb, in order to promote
mobility. Our interests lie in rehabilitation of the upper limb (hand and arm), which can remain weak in up
to 66% of stroke survivors. It has been shown (Wade et al, 1985) that early and intensive practice of active
functional tasks in an enriched environment show more positive outcomes for upper limb rehabilitation.
Patient motivation has also been recognised as important for therapeutic outcome and many studies have
consistently linked motivation to better therapeutic outcomes (Maclean et al, 2000). Motivation is, however,
a complex subject which has been shown to be linked to both personality traits and the social environment
(Maclean et al, 2002).
Recently there has been a great deal of interest among health professionals in the use of computer games
for rehabilitation. Video games have long been known to be engaging to play, with gamers often playing for
hours at a time, seemingly unaware of the passing of time. If rehabilitation games can be created which
exhibit similar high degrees of engagement it is possible that therapeutic outcome will improve. Recently,
companies such as Sony and Nintendo have developed commercial games combining entertainment with
exercise – for example, the Sony Eyetoy games and Nintendo Wii Fit. Research has indicated, though, that
people with motor function problems can have problems playing commercial games out-of-the-box (Rand et
al, 2004). This raises the prospect of games being designed specifically for rehabilitation.
A number of systems have been reported which use virtual environment technology for various
rehabilitation applications, including phobias, post-traumatic stress disorder and motor function (upper and
lower limbs). Some of these have used virtual reality technology to immerse the user in the virtual
environment; others use standard display technology (computer monitors and TVs). Technology to track the
user’s movements is also common, particularly for motor function rehabilitation. The tracked data is often
used to drive a graphical representation of the user or part of the user in the virtual environment, enabling the
user to achieve a high degree of control and impact onto the virtual world. What the user does in the virtual
world is clearly dependent on the rehabilitation objectives. Activity can be structured around a simulation of
conventional functional-based tasks or be more game-like, where the functional tasks are couched in the

Proc. 8th Intl Conf. Disability, Virtual Reality & Associated Technologies 195
Viña del Mar/Valparaíso, Chile, 31 Aug. – 2 Sept. 2010
2010 ICDVRAT; ISBN 978 07049 15022
context of a game. Several reviews have reported on published work in this area and we refer the reader to
(Sveistrup, 2004) and (Weiss et al, 2004) for further reading.
While a number of bespoke rehabilitation systems have been built which use games to present the activity
to the user, there has been little discussion of the key principles of game design that are important for this
context. The aim of our work is to identify core game design principles for rehabilitation and realise these in
a collection of games for rehabilitation of the upper limb motor function. In this paper we review our ideas
for rehabilitation game design and detail the implementation of one of our games for upper limb motor
function rehabilitation. Further, we report on the evaluation strategy for our games, focusing on playability,
usability and engagement. The paper is organised as follows. In Section 2 we discuss a number of video
game design principles important for rehabilitation. Section 3 details the implementation of one of the games
developed by our group. Section 4 reports on evaluation of the games and conclusions appear in Section 5.

2. GAME DESIGN THEORY, ENGAGEMENT AND REHABILITATION

Video games are renowned for their ability to provoke high levels of engagement amongst game-players and
Rizzo and Kim (Rizzo et al, 2005) suggested that the designers of rehabilitation systems could benefit from
analysing the principles of game design. It has been reported that users of rehabilitation games enjoy game-
like challenges (when they are present) and it would seem that these are perhaps easier to create in computer-
based intervention than in traditional rehabilitation. This enjoyment may translate to improved engagement
and self-motivation which are recognised as important for adherence to a programme of rehabilitation
therapy. Other work in this area includes that of (Goude et al, 2007), who proposed a conceptual model that
supports game idea generation, task design and categorisation of existing games in relation to stroke
rehabilitation, but only a subset of the taxonomy is published. (Flores et al, 2008) criticised existing gaming
scenarios for rehabilitation for being designed solely to test the overall functionality of the interface hardware
and suggested that more attention needs to be given to supporting motivation for active participation.
(Colombo et al, 2007) suggested that the difficulty level of the motor task, the awareness of the performance
obtained and the quantity and quality of feedback presented to the user can influence patient motivation.
In (Burke et al, 2009) we identified three principles of game design which we feel have particular
importance to rehabilitation: meaningful play, handling failure and setting an appropriate level of challenge.
Meaningful play emerges from a game in the relationship between a player’s actions and the system’s
response. Central to creating and maintaining meaningful play is the concept of feedback, the methods by
which the game responds to the changes or choices made by the player. Feedback can be aural (sound
effects, verbal dialogue), visual (text, images, score indicators) and haptic (vibration). It is important that a
user of rehabilitation games is aware of their progress towards goals (short-term and long-term) in order to
achieve effective engagement.
A game responds to player actions in order that the player is aware of the impact of their actions on the
game world. In this way the game encourages the player to learn what actions result in progress through the
game and what actions do not. Feedback has particular importance for rehabilitation, where the concept of
progress through the game may differ from that in traditional video games. Failure, for example, has always
been a prominent aspect of video games, not only present but expected by players – players are often given a
number of ‘lives’ which represent the number of chances they have to complete a task without dying. Failure
to complete a task within this number of lives may result in “game over”, meaning the player has to start the
game (or game level) from the beginning. For motor function rehabilitation games, where the affected limbs
are being mobilised, there is a risk that if the player were to experience failure during their initial play, they
may attribute this failure to poor motor function and this could result in poor game engagement. Clearly this
needs to be tested with users, but we suggest that failure in rehabilitation games should be handled
conservatively – since engagement is a pre-requisite for positive therapeutic outcome, all engagement should
be rewarded, at least initially. By handling failure in a positive way players may be more likely to remain
engaged with the intervention. This does not mean that the game shouldn’t offer challenge and drama, since
these are important reasons for games being engaging in the first place.
The level of challenge a game offers a player is a major influence in how engaging the game is to play.
Previous work has reported that users of rehabilitation games have difficulty with COTS games where the
level of the environments cannot be graded (Rand et al, 2004). The range of motor function among people
with stroke, for example, varies greatly and what might be easy for one person is impossible for another. For

196 Proc. 8th Intl Conf. Disability, Virtual Reality & Associated Technologies
Viña del Mar/Valparaíso, Chile, 31 Aug. – 2 Sept. 2010
2010 ICDVRAT; ISBN 978 07049 15022
optimal engagement, games should present an ideal level of challenge for each individual player; neither too
difficult that it becomes frustrating, nor too easy that it becomes boring. Traditional video games have
different ways of varying challenge for players. Prior to commencement, games often ask players to select a
level of difficulty, for example: easy, medium, hard, insane. This determines the pace of the game, number
of enemies, number of opportunities for the player to acquire health or weapons etc. In motor function
rehabilitation games the initial configuration will affect the pace, size of game elements, distance from user,
etc. While data acquired from standardised assessment tests prior to game-play may help in understanding
the patient’s level of motor function, it is difficult to map values from these tests directly to game
configuration. We have developed a configuration tool as part of our library of games where the user’s range
of movement is assessed using the same interaction technology as used to play the games. Data from this
tool is then used to position game elements so that they are all initially reachable by the user on start-up. As
the game progresses, the game adapts the challenge dynamically by changing the pace and game element
positions and size, as described earlier.
As the game progresses, players will often face levels of increasing difficulty, designed to present more of
a challenge as they become more familiar and skilled with the game. In commercial games, completing a
level is possible only once the player understands enough of the game mechanics and has acquired the
necessary knowledge and skills (including controller skills) to progress through the level to the end. Other
games may not have recognisable levels as such, but the challenge might increase as particular points in the
game are reached, indicating that an appropriate level of understanding and acquisition of skills has been
achieved. We have investigated the use of adaptivity to automatically control the game difficulty, by altering
programmable elements of the game in response to the performance of the player. For players that are
struggling, the pace of the game slows and game elements become larger and easier to reach. On the other
hand, for players that are coping easily with the challenge, the game speeds up, game elements become
smaller and harder to reach.

3. VIDEO CAPTURE GAMES FOR STROKE REHABILITATION

Our group has designed a number of games for upper limb rehabilitation post stroke, using the principles
described above. The games use a low-cost webcam to capture player movements and image analysis
software to track the hand. All software was written in C# in Microsoft’s XNA framework. Four games
were designed in the arcade style (i.e., short, easy to learn). Two were one-handed games, where the affected
limb was used to play; a third game was two handed and a fourth game was also two handed but required the
hand movements to be synchronised. The user, wearing a different coloured glove or mitt on each hand,
faces the computer monitor, where his/her image appears in the background (behind the rendered game
elements). The different coloured gloves allow the software to track each hand’s movements. To play, the
user touches a game element (sprite) with either one or both hands, as appropriate for the particular game.
All the games are adaptive – they get easier or harder depending on the performance of the player. To
illustrate how the game design principles discussed in Section 2 were applied we discuss the design of one of
the games in detail. Details of the other games can be found in (Burke et al, 2009) and videos can be seen at
this link: http://www.vimeo.com/4279126 . Appendix A shows screenshots of the configuration tool and the
games.
In Bubble Trouble, a game designed to improve gross movement of the upper limb, virtual bubbles appear
on the screen at random positions within the player’s range of movement (as determined by the configuration
tool) for a short length of time, and the goal is for the player to burst these bubbles by touching them before
they disappear. The game has two variations, supporting single-handed play and two-handed play. In the
single-handed version the affected limb is used. In the two-handed version, each bubble is colour-coded to
match the colour of its respective glove; an arrow (pointing left or right) inside the bubble also depicts which
hand the bubble should be burst with. The game has a programmable time limit (set on startup), and the
number of bubbles burst within this time limit determines the player’s final score. The player cannot fail in
the game, rather a poor performance would be reflected with a lower score; however, messages at the end of
the game are positive and encouraging regardless of the score. Distinctive and easily identifiable visual and
auditory feedback is used to create meaningful play, such as a popping noise for bursting a bubble, a buzzer
sound for missing bubbles, bold visual messages for hits and misses, colour coding of sprites, particle effects
and score charts. Adaptivity is used to present a level of challenge suitable for the player by dynamically
altering the pace, positions and size of bubbles according to the player’s performance.

Proc. 8th Intl Conf. Disability, Virtual Reality & Associated Technologies 197
Viña del Mar/Valparaíso, Chile, 31 Aug. – 2 Sept. 2010
2010 ICDVRAT; ISBN 978 07049 15022
 4. USER EVALUATION
Although the goal for any motor function rehabilitation system must be improvement in the affected limb(s),
it is necessary to run large-scale user trials over a period of several weeks in order to evaluate the
effectiveness of any intervention on motor function outcome. Unfortunately this was outside the scope of the
project at this time and so we elected to focus on the evaluation of the usability and playability of the games.
Subsequently we identified an additional requirement to evaluate the effectiveness of the engagement quality
of the games. As with the design of the games, the evaluation protocol was developed after discussion
between colleagues from various departments within the University of Ulster: health and rehabilitation
sciences, computer science and psychology. Approval for a 3-phase protocol was sought and granted from
the University’s Research Ethics Committee.
Phase 1 saw the games evaluated by 10 health people, recruited from the students and staff of the School
of Computing. The games were initially demonstrated by members of the research team; the participants
then played each game twice. The main instrument used to collect quantitative data was a usability and
playability questionnaire (Desurvire et al, 2004) which was scored on a visual-analogue scale from 0 to 10.
Results indicated that the majority of participants (70-80%) enjoyed all of the games. Players also felt that
the games were varied, had good replay value, responded consistently and were easy to play due to the
intuitive control mechanisms. More than 80% of players reported noticing that the games adapted to their
performance and all liked this feature, suggesting that the games would be less enjoyable without adaptivity.
One user did express the view that the change in the pace of the games when adaptivity was present was too
aggressive and that the change of pace should be more ‘gentle’. These results helped inform minor
improvements to the games, most notably the addition of improved feedback – audio feedback and larger
graphical cues when game elements were hit and missed.
Phase 2 of the study recruited 3 people with stroke from a local stroke club. One of the participants was
in a wheelchair. Following demonstration by a member of the research team, participants played the games
once, in their home (each game was played for 90 seconds). The same playability questionnaire was used as
in Phase 1 and a further instrument for measuring self-reported exertion (Borg CR10 scale) was used.
Results of this Phase were published elsewhere (Burke et al, 2009) but in summary all participants were
successful in being able to play the games and reported that they found them challenging, fun and varied. All
participants, though, reported that the adaptivity changed the pace of the games too aggressively, in particular
that the games speeded up too quickly when they were performing well.
In Phase 3, 2 people with stroke were recruited to play the games for a period of 3 weeks (3 times per
week). The main focus of the study was to evaluate usability, playability and engagement with the games.
At the end of the intervention participants filled out the same usability and playability questionnaire as
Phases 1 and 2 and also an Intrinsic Motivation Inventory (IMI) (Ryan, 1982). The IMI elicits data about a
player’s self-report of intrinsic motivation (which we interpret as a reliable measure of engagement).
Participants also had a chance to give more general comments on their experiences. We also collected in-
game data which enabled us to study patterns of scoring and gameplay speed (the average speed at which the
game elements e.g., bubbles, moved). Since game elements speed up or slow down through game
adaptivity, gameplay speed gives an additional measure of user performance.
Player A was a 70-year old, right-handed, female with a right hemiplegic stroke 2 years previous to the
study. Her upper limb Motricity Index (Collin et al, 1990) scores prior to the intervention were 63 (right)
and 100 (left). She generally used a wheelchair, although she was mobile with a stick at home. She was not
receiving any therapy at present. The usability and playability questionnaire results indicated that the player
enjoyed playing the games (rating 10 out of 10) and also found them enjoyable to replay (9/10). The player
felt highly involved in the games (9/10), felt that the games responded consistently and found the feedback to
be effective. She felt the controls were intuitive and that the input device was easy to use. The player also
indicated that she preferred adaptive difficulty in comparison to altering the challenge manually (9/10). In-
game scores and gameplay speed graphs are shown for the one-handed Bubble Trouble game in Appendix B
below – results indicate a small rise in scores over the 3 weeks for all 4 games. Similarly, the gameplay
speeds again show a small rise over the 3 weeks. Player A was generally enthusiastic and well motivated,
although her limited range of movement did cause her frustration at times. She commented that the games
were ‘interesting, fun and encouraged me to move my arm’. Results from the Intrinsic Motivation Inventory
are presented in Table 1. The interest/enjoyment subscale is considered the self-report measure of intrinsic
motivation and Player A scored this as 61%.

198 Proc. 8th Intl Conf. Disability, Virtual Reality & Associated Technologies
Viña del Mar/Valparaíso, Chile, 31 Aug. – 2 Sept. 2010
2010 ICDVRAT; ISBN 978 07049 15022
 Player B was a 48-year old, right-handed male with a right hemiplegic stroke less than 1 year previous to
the study. His upper limb MI scores prior to the intervention were 70 (right) and 100 (left). He was mobile
and was not receiving any therapy at present. The usability and playability questionnaire results indicated
that the player found the games enjoyable to play (8/10), replayable (9/10) and consistent (8/10) and effective
(8/10) in response to the player’s actions. The player did indicate that the games could have been more
varied to avoid fatigue (5/10) and that games were sometimes too fast (5/10). He felt that the controls were
intuitive (8/10) and that the input device (gloves) was easy to use (9/10). He also approved of the adaptivity
feature (9/10). In-game scores and gameplay speed graphs are shown for one of the games in Appendix B–
results indicate a small rise in scores over the 3 weeks for all 4 games. Similarly, the gameplay speeds again
show a small rise over the 3 weeks. Player B was generally enthusiastic and well motivated. He commented
that ‘the games help me relax and know my limitations’. Results from the Intrinsic Motivation Inventory are
presented in Table 1. The interest/enjoyment subscale is considered the self-report measure of intrinsic
motivation and Player B scored this as 84%.
Table 1. Intrinsic Motivation Inventory Scores.

Interest/enjoyment Perceived Effort/importance Pressure/tension Value/usefulness

competence
Player A 61% 38% 75% 28% 75%
Player B 84% 69% 75% 25% 88%

Although both users were able to play the games, Player A self-reported perceived competence was much
lower than that of Player B and this is corroborated by the graphs in Appendix B. For Bubble Trouble (one-
handed) Player A’s scores were in the range 77-87 compared to Player B’s range 88-96. To what do we
attribute the improvement in both players scores over the 3 weeks is a difficult question. There may be a
learning effect, where the player gets better through familiarity with the game over time, or it could be due to
an improvement in the player’s motor function. Clearly larger studies, evaluating motor function before and
after intervention, are required to answer this question more fully. Player A’s interest/enjoyment score was
also less than Player B’s, though still a reasonable 61% (versus 84%) and both players rated the games
value/usefulness highly (75% and 88%), their effort/importance highly (75% and 75%) and pressure/tension
lowly (28% and 25%). This data would appear to indicate that both players were intrinsically motivated to
continue playing the games and this was confirmed through observation by the research team.

5. CONCLUSIONS
We have presented ideas for the design of games for upper limb motor function rehabilitation following
stroke. In particular we identified the concepts of meaningful play (feedback) and challenge to be
particularly important for the design of games for rehabilitation. Evaluation of a number of games developed
incorporating these principles show that the games were indeed usable and playable by people with stroke.
Further, the games seemed to stimulate a high level of interest and enjoyment by the participants, which may
indicate the games are engaging to play. We hope to conduct larger studies which would improve the
generalisability of these promising early results.

6. REFERENCES
J W Burke, M D J McNeill, D K Charles, P J Morrow, J H Crosbie and S M McDonough (2009), Optimising
engagement for stroke rehabilitation using serious games, The Visual Computer, 25(12), pp. 1085-1099.
C Collin and D Wade D (1990), Assessing motor impairment after stroke: a pilot reliability study, J Neurol
Neurosurg Psychiatry 53(7), pp. 576-579.
R Colombo, F Pisano, A Mazzone, C Delconte, S Micera, M C Carrozza, P Dario and G Minuco (2007),
Design strategies to improve patient motivation during robot-aided rehabilitation, Journal of
NeuroEngineering and Rehabilitation, 4(3), pp. 33-39.
H Desurvire, M Caplan and J A Toth (2004), Using Heuristics to Evaluate the Playability of Games,
Proceedings ACM SIGCHI 2004, Vienna, Austria.

Proc. 8th Intl Conf. Disability, Virtual Reality & Associated Technologies 199
Viña del Mar/Valparaíso, Chile, 31 Aug. – 2 Sept. 2010
2010 ICDVRAT; ISBN 978 07049 15022
E Flores, G Tobon, E Cavallaro, F I Cavallaro, J C Perry and T Keller (2008), Improving Patient Motivation
in Game Development for Motor Deficit Rehabilitation, ACM Advances in Computer Entertainment
Technology, 352 pp. 381-384.
D Goude, S Björk, and M Rydmark (2007), Game Design in Virtual Reality Systems for Stroke
Rehabilitation, Studies in Health Technology and Informatics, 125, pp.146-148.
N Maclean and P Pound (2000). A critical review of the concept of patient motivation in the literature on
physical rehabilitation, Social Science & Medicine, 50: 495-506.
N Maclean, P Pound, C Wolfe and A Rudd (2002), The Concept of Patient Motivation: A Qualitative
Analysis of Stroke Professionals’ Attitudes, Stroke, 33, pp. 444-448.
D Rand, R Kizony and P L Weiss (2004), Virtual reality rehabilitation for all: Vivid GX versus Sony
PlayStation II EyeToy, Proceedings of the 5th International Conference on Disability, Virtual Reality &
Associated Technology, pp. 87-94.
A Rizzo and G J Kim (2005), A SWOT analysis of the field of virtual reality rehabilitation and therapy,
Presence,14 pp. 119–146.
R M Ryan (1982), Control and information in the intrapersonal sphere: An extension of cognitive evaluation
theory, Journal of Personality and Social Psychology, 43, pp.450-461.
H Sveistrup (2004), Motor Rehabilitation using virtual reality, Journal of NeuroEngineering and
Rehabilitation, pp.1-10.
D T Wade, V A Wood and R L Hewer (1985), Recovery after stroke - the first three months, Journal of
Neurology, Neurosurgery, and Psychiatry, 48, pp. 7-13.
P L Weiss, D Rand, N Katz and R Kizony (2004), Video capture virtual reality as a flexible and effective
rehabilitation tool, Journal of NeuroEngineering and Rehabilitation, pp.1:12

Appendix A: screenshots of the games

Image 1. Range-of-movement configuration. Image 2. “Rabbit Chase”, one handed game.

Image 3. “Bubble Trouble” (Two handed version). Image 4. “Arrow Attack”, two handed game.

200 Proc. 8th Intl Conf. Disability, Virtual Reality & Associated Technologies
Viña del Mar/Valparaíso, Chile, 31 Aug. – 2 Sept. 2010
2010 ICDVRAT; ISBN 978 07049 15022
 Appendix B: Games Scores and Play Speeds

Patient A scores for Bubble Trouble game Patient A game speed for Bubble Trouble game

Patient B scores for Bubble Trouble game Patient B game speed for Bubble Trouble game

Proc. 8th Intl Conf. Disability, Virtual Reality & Associated Technologies 201
Viña del Mar/Valparaíso, Chile, 31 Aug. – 2 Sept. 2010
2010 ICDVRAT; ISBN 978 07049 15022

View publication stats