Research in Developmental Disabilities 34 (2013) 3654–3659

Contents lists available at ScienceDirect

Research in Developmental Disabilities

A Kinect-based upper limb rehabilitation system to assist

people with cerebral palsy

Yao-Jen Chang a,c,*, Wen-Ying Han b, Yu-Chi Tsai a

a Department of Electronic Engineering, Chung Yuan Christian University, Chung-Li 320, Taiwanb Physical Therapy Services, National Tainan Special School, Tainan 709, Taiwanc Holistic Medical Device Development Center, Chung Yuan Christian University, Chung-Li 320, Taiwan

A R T I C L E I N F O

Article history:

Received 26 June 2013

Received in revised form 14 August 2013

Accepted 14 August 2013

Available online 4 September 2013

Keywords:

Physical rehabilitation

Cerebral palsy

Kinect

Image recognition

A B S T R A C T

This study assessed the possibility of rehabilitating two adolescents with cerebral palsy

(CP) using a Kinect-based system in a public school setting. The system provided 3 degrees

of freedom for prescribing a rehabilitation program to achieve customized treatment. This

study was carried out according to an ABAB reversal replication design in which A

represented the baseline and B represented intervention phases. Data showed that the two

participants significantly increased their motivation for upper limb rehabilitation, thus

improving exercise performance during the intervention phases. Practical and develop-

mental implications of the findings are discussed.

� 2013 Elsevier Ltd. All rights reserved.

1. Introduction

Cerebral palsy (CP) is a term denoting a group of non-progressive, non-contagious motor conditions that cause physicaldisability in human development, mainly in the various areas of body movement (Rosenbaum et al., 2007). People with CPexperience limitations in fine motor control, strength, and range of motion. These deficits can dramatically limit their abilityto perform daily tasks, such as dressing, hair combing, and bathing, independently. In addition, these deficits can reduceparticipation in community and leisure activities, and even negatively impact occupational perspectives (Gabriele & Renate,2009; Lai, Studenski, Duncan, & Perera, 2002; Wagner, Lang, Sahrmann, Edwards, & Dromerick, 2007). Individuals withaccess to intensive multidisciplinary rehabilitation programs demonstrate earlier and faster functional gains on functionalindependence measures. Participating in repetitive exercises can help people with CP overcome the limitations theyexperience. However, one study indicates that only 31% of people with motor disabilities perform the exercises asrecommended (Shaughnessy, Resnick, & Macko, 2006), which can result in negative consequences such as obesity-relatedchronic health conditions. People often cite a lack of motivation as an impediment to them performing the exercisesregularly (Lloyd-Jones et al., 2010). Furthermore, the number of exercises in a therapy session is typically insufficient (Lang,MacDonald, & Gnip, 2007). One solution to this issue is staff intervention; however, it may not be economically viable.

Research suggests that physical therapy can sufficiently stimulate brain to remodel itself and provide better motorcontrol (Kleim, Jones, & Schallert, 2003). Identifying effective methods of encouraging people with CP to perform exercises iscrucial for helping them retain or enhance their motor control and increase their independence. Full reintegration into thecommunity and vocation are the ultimate goals. Strategies incorporating the use of various technologies for the people with

* Corresponding author at: Department of Electronic Engineering, Chung Yuan Christian University, Chung-Li 320, Taiwan. Tel.: +886 3 265 4630;

fax: +886 3 265 4699.

E-mail addresses: yjchang@cycu.edu.tw (Y.-J. Chang), winyn6214@yahoo.com.tw (W.-Y. Han), youji32@gmail.com (Y.-C. Tsai).

0891-4222/$ – see front matter � 2013 Elsevier Ltd. All rights reserved.

http://dx.doi.org/10.1016/j.ridd.2013.08.021

Y.-J. Chang et al. / Research in Developmental Disabilities 34 (2013) 3654–3659 3655

CP have been developed for physical therapy across numerous settings. These innovative technologies provide a selection ofdiffering platforms and contribute to the user sensory experience. In particular, virtual reality (VR) and motion-based gameshave been used for rehabilitation recently. Virtual reality systems demand focus and attention, can motivate the user tomove, and provide the user with a sense of achievement, even if they cannot perform that task in the real world. Burdea,Deshpande, Liu, Langrana, and Gomez (1997) were the first to incorporate VR technology for hand rehabilitation therapy. Byusing sensors installed in gloves, users were provided with VR-based force feedback, and the system provided data recordsregarding the grip forces achieved for every activity, and grip forces and bending angles of fingers. Jack et al. (2000) used a PCbased desktop virtual reality system for rehabilitating hand function in stroke patients. The system used two hand inputdevices, a hand-centric 3D motion capture glove and a force feedback glove, to allow the user to interact with rehabilitationexercises. Both gloves are complicated and may not be affordable to many people with motor disabilities. Virtual realityneeds a special setting such as caves or tents, making it less practical to many people that need it.

Industrial motion sensors (Bach-y-Rita et al., 2002; Foerster, Smeja, & Fahrenberg, 1999; Shih, 2011; Shih, Chang, & Shih,2010a) and, in particular, entertainment oriented ones (Balaam et al., 2011) such as Nintendo Wii Remote (Alankus, Proffitt,Kelleher, & Engsberg, 2010; Shih, Chang, & Shih, 2010b) and Wii Balance Board (Gil-Gomez, Llorens, Alcaniz, & Colomer,2011) show research evidence that they are useful as physical rehabilitation tools. Furthermore, motion-based games thatcombine motion sensor technology and fun with video games can motivate people to engage in exercises that games designpurposefully. Alankus et al. (2010) attached a Wii remote to the arms of a female patient with hemiparesis. The patientcontrolled the motion of a helicopter in a game or caught a baseball by moving her arms. The researchers employed a game-based approach to increase the patient’s motivation to participate in physical rehabilitation and adopted Wii remote featuresto design movements or exercises suitable for rehabilitation. A drawback with motion sensors is that people have to fastenthem on limbs, hold them in the hands or even wear them on the body to detect motions and possibly generate forcefeedback. Wearing sensors can cause inconvenience and discomfort.

A previous study used a Kinect-based system to assist people with motor impairments in rehabilitation (Chang, Chen, &Huang, 2011). By using videos in which physical therapists demonstrated rehabilitation exercises and the action–recognition function in Kinect, the participants were able to independently perform rehabilitation without teacher presence.However, although using Kinect allowed the participants to perform customized movements, the disadvantage of thisapproach is that customized movements can only be identified based on displacement, which limits movements duringrehabilitation activities.

This study aims to enable physical therapists to design diversified movements based on the needs of participants with CP andachieve the goal of developing customized rehabilitation exercises, by designing a system to help motivate people to increasethe number of exercises and improve the motor proficiency and quality of life. We primarily employed the Microsoft Kinectsensory device to establish an upper limb rehabilitation system for people with CP. This system enables physical therapists todevelop series of poses for rehabilitation based on the level of motor impairment for each participant. These poses can betailored for training participants to dress or undress, for training participants to hold objects, and for self-feeding, just to name afew. Moreover, this Kinect-based system detects joint points and subsequently calculates various angles (i.e., shoulder flexion,shoulder extension, shoulder external abduction, shoulder external rotation, shoulder internal rotation, and elbow flexion) forthe requested poses to determine whether the upper limb movements correspond to therapist’s expectations.

Microsoft Kinect is a webcam-style add-on peripheral intended for the Xbox 360 game console. Kinect enables users tocontrol and interact with the game console without the need to touch a game controller, through a natural user interfaceusing gestures. The device comes with an RGB camera and a depth sensor, which in combination provide full-body 3Dmotion capture capabilities and gesture recognition. Going beyond the system’s intended purpose of playing games, weleveraged the human gesture recognition capabilities of Kinect to determine whether a user performed exercise correctly inphysical rehabilitation. Using Kinect means that the users need not be bothered with body sensors and that the system cansave the users from wearing sensors that can be intrusive.

The proposed system measures 3 degrees of freedom (DoF) in upper limb rehabilitation: 1 DoF for elbows and 2 DoF forshoulders. It is a major upgrade of its previous 1 DoF design (Chang et al., 2011). The design with 3 DoF provides therapists withgreater flexibility to prescribe customized rehabilitation programs to address individual needs. With the help of technology, atherapist can specify angles of shoulder flexion, shoulder extension, shoulder external abduction, shoulder external rotation,shoulder internal rotation, and elbow flexion for upper limb rehabilitation. Using this system, therapists can gauge the accuracyof participant upper limb movements by measuring the above-mentioned angles. The system also includes an interactiveinterface with audio and video feedback to enhance motivation, interest, and perseverance to engage in physical rehabilitation.Details of rehabilitation conditions are also automatically recorded in the system, allowing therapists to review rehabilitationprogress quickly. Therefore, this system reduces staff workload, helps assess the quality of physical exercise, and enhances theefficiency of rehabilitation, which increases the ability to perform daily tasks independently.

2. Method

2.1. Participants

After the system was developed, a physical therapist in a special-education school invited two junior-high schoolstudents to test the assistive device. Alice and Belle were identified for inclusion in the study. Alice, age 14, was diagnosed as

Y.-J. Chang et al. / Research in Developmental Disabilities 34 (2013) 3654–36593656

having severe cerebral palsy with inflexibility of upper and lower limb movements and insufficient muscle enduranceafter she was born. She has received substantial rehabilitation since she enrolled in the special education system. Sheuses a wheel chair for transportation. She is an outgoing adolescent schoolgirl and owns good oral communication skills.The function of her right hand far exceeds the function of the other hand. Therefore, her therapist puts more efforts inrehabilitating her left hand to help her perform daily tasks independently. Belle, age 14, was diagnosed as havingacquired muscle atrophy and insufficient muscle endurance. She remains mobile on a wheelchair. She is an introvertand only has very limited communication skills. She did not receive early intervention until she enrolled in the publicspecial education program. The therapist has worked for years to increase her task skills such as self-feeding andreaching for things independently. Neither of the participants had previous experience with Kinect. The names ofparticipants have been changed in this report to protect their privacy. Informed consent was provided at the levels ofthe service organization, individual staff members involved in the study, and the main caregivers on behalf ofparticipants with CP.

2.2. Setting

This study tested the proposed system at a suburban special education school, which receives students with special needsfrom kindergarten to high school. Kinect was connected to an IBM T61 notebook computer on which the rehabilitationsystem software developed in house was installed with Microsoft Windows 7. The computer has an audio module that weused to deliver audio feedback whereas its 14-in. LCD screen was used for visual interaction. The software was coded inMicrosoft C# using Visual Studio as the integrated development platform. The interactive interface with audio and videofeedback can be programmed to reinforce students’ motivation to engage in physical rehabilitation. For optimal performanceof the Kinect sensor, participants were required to perform physical exercise approximately 3 feet in front of the Kinectmodule. See Fig. 1 for a participant performing rehabilitation in front of the system.

2.3. Operational definitions of target behaviors

The rehabilitation movements used in the experiment were as follows: (1) lifting one arm upwards (hand up), (2)lifting one arm to the front (hand forward), and (3) moving a hand to the mouth (hand to mouth). For the movement oflifting one arm upwards, participants who sat on wheelchairs initially put both hands on the wheelchair armrests. Whenexercises began, they lifted one arm upwards and stretched it as far as possible. After maintaining the posture for atleast 3 s, they put down their hand back to the initial position. For the movement of lifting one arm to the front,participants sat on wheelchairs and had their both hands on the wheelchair armrests. When exercise began, they liftedone arm from the initial position until their arm was in parallel with the ground and was stretched as far as possible.After maintaining the target posture for at least 3 s, participants put down their hand to the initial position. For movinga hand to the mouth, participants sat on wheelchairs and had their both hands on the wheelchair armrests. Whenexercise began, they maneuvered the hand that did or was supposed to do the self-feeding from the initial position untiltheir hand reached the mouth. After maintaining the target posture for at least 3 s, participants put down their hand tothe initial position.

2.4. Experimental conditions

The study was carried out according to an ABAB reversal replication design in which A represents the baseline and Brepresents intervention phases (Richards, Taylor, Ramasamy, & Richards, 1999). Based on participant ability, physicaltherapists developed various movements for rehabilitation and established initial and final poses for each movement. Therehabilitation movements used in the experiment were as follows: (1) lifting one arm upwards (hand up), (2) lifting one armto the front (hand forward), and (3) moving a hand to the mouth (hand to mouth). The physical therapist instructed thestudents to perform three types of movements (i.e., hand up, hand forward, and hand to mouth) 15 times each in a session.The participants were requested to rest for 5 min before proceeding with the next type of movement. This process wasperformed until the three types of movements were completed. A movement was considered correct if the participantperformed it to the prescribed standard where the angle of each DoF fell within the expected range. One session per day wasobserved within each study period. Sessions were taped and recorded.

2.4.1. Baseline phases

In the baseline phase, the therapist instructed a participant the content of a rehabilitation program. The participant thenrepeated what she heard. If there was any inaccuracy in movement types or numbers of exercise, the therapist corrected theparticipant until the content of the prescribed program was understood without errors. Then the participant started toperform physical exercise on the wheelchair. During the baseline phases, the participants followed the instructions anddemonstrations of the physical therapist and completed the required movements without assistance. The proposed systemdetected inaccuracy in performance and counted the number of correct movements in each session. However, the system didnot provide any visual cues and any feedback to the participant. The therapist did not interfere with any inaccuracy inperformance that the participant might have.

Fig. 1. A participant performing rehabilitation in front of the system.

Y.-J. Chang et al. / Research in Developmental Disabilities 34 (2013) 3654–3659 3657

2.4.2. Intervention phases

In this phase, the proposed system was used. The participant used the cues on the screen to complete the prescribedprogram. The proposed system detected the inaccurate movements of participants during rehabilitation activities anddetermined whether the movements corresponded to the requirements of therapists. It automatically counted thenumber of correct movements in each session. In addition, picture- and voice-based feedback was used to enrichthe rehabilitation process and increase motivation to engage in rehabilitation. The computer screen displayedfavorite cartoon characters each time a correct posture was achieved and played theme music when a typeof movements was completed. The therapist did not interfere with any inaccuracy in performance that the participantmight have.

2.5. Inter-observer agreement

In addition to machine counting of correct movements throughout the experiment, reliability observers collecteddata on the number of movements performed correctly during at least 40% of the sessions and across all phases of thestudy for each participant (range 43–72%). Two special education graduate students who acted as the observers weretrained by the physical therapist we collaborated within the public special education school to collect data by providingparticipants with the rehabilitation programs, performing participant evaluations, verbally explaining the procedures,and answering all questions regarding the process. The observers responded during participant sessions in threedifferent ways: incorrect, correct and no response. The same two observers were used across both participants; theywere required to have 100% accuracy across 10 consecutive trials during practice before working with participants.Agreement between the trainer and reliability observers on the correctly performed movements was calculated on asession-by-session basis, using the following formula: agreements/(agreements + disagreements) � 100%. The resultingpercentages of agreement were consistently above 97%. Moreover, machine reading and human observation showed anagreement of 98% for all the sessions involving observers. Therefore, the results can be reliably collected from machinereading.

2.6. Results

The first baseline phase lasted for 4 days for both participants. The first intervention phase lasted for 8 days for bothparticipants. The second baseline phase lasted for 3 days for both participants. The second intervention phase lasted for 7days for Alice and 5 days for Belle. Belle dropped two sessions in the second intervention phase because she was ill. Fig. 2shows data for Alice. The baseline and intervention phases observed the number of correct movements of each type. Duringthe first baseline phase (4 sessions), the number of correct movements for each type was 3, 3, and 2 on the average. Duringthe first intervention phase (8 sessions), the number of correct movements for each type increased to approximately 13, 14,and 13, respectively. The number of correct movements decreased to 4, 2, and 4 during the second baseline (3 sessions) andincreased again, reaching nearly 14, 14, and 13 during the second intervention phase (7 sessions). The difference of numbersof correct movements between the baseline and the intervention was significant (p < .05) on the Kolmogorov–Smirnov test(Siegel & Castellan, 1988).

Fig. 3 shows data for Belle. During the first baseline phase (4 sessions), the number of correct movements for each typewas 1, 2, and 3 on the average. During the first intervention phase (8 sessions), the number of correct movements for eachtype increased to approximately 8, 12, and 13, respectively. The number of correct movements decreased to approximately 4,4, and 4 during the second baseline (3 sessions) and increased again, reaching nearly 12, 14, and 13 during the secondintervention phase (5 sessions). The difference of numbers of correct movements between the baseline and the interventionwas significant (p < .05) on the Kolmogorov–Smirnov test (Siegel & Castellan, 1988).

Fig. 2. Number of correct movements for Alice completing the rehabilitation program.

Fig. 3. Number of correct movements for Belle completing the rehabilitation program.

Y.-J. Chang et al. / Research in Developmental Disabilities 34 (2013) 3654–36593658

3. Discussion

This study assessed the effectiveness of the proposed system for motivating physical rehabilitation using a baseline/intervention reversal replication design. Results indicate that for the two individuals with CP, the proposed system inconjunction with operant conditioning strategies may facilitate autonomous physical rehabilitation prescribed by thetherapist.

During the baseline phases, the number of correct movements for each type performed by Alice and Belle were lowbecause the required movements were relatively difficult for them to perform. In addition, the absence of feedback resultedin a lack of enthusiasm for rehabilitation. For example, when performing the hand-up movement, the participants loweredtheir arms before reaching the specified angle; thus, the quality of the movements failed to satisfy the therapist’sexpectations and achieve effective rehabilitation. During the intervention phases, the performance of Alice and Belleincreased substantially compared to those obtained during the baseline phases. Observations during the experimentalprocess showed that after completing the movements, the participants demonstrated considerable interest toward thedisplayed feedback pictures.

A limitation of the study is that the results are based on two cases. Therefore, a general conclusion cannot be extrapolatedregarding the efficacy of the proposed system. Future studies should include experiments involving more participants withmotor impairments and focus on further evaluating the training system. Additionally, more interactive features should beadded to enhance the user experience. This technology can be used in other settings for other types of motor rehabilitation.In addition to people with CP, the technology can be used with other populations, such as people with traumatic brain injury(TBI), stroke, and spinal cord injury (SCI).

People with cerebral palsy (CP) must undergo long-term physical rehabilitation to enhance neural development.Currently, commercial specialized rehabilitation products cannot be easily customized and are expensive; consequently,public special-education schools generally cannot afford these products. One solution is the use of enhanced hardware or

Y.-J. Chang et al. / Research in Developmental Disabilities 34 (2013) 3654–3659 3659

software assistive technology that can repurpose commercial high-technology products, turning them into highperformance assistive devices to match the special needs of persons with disabilities (Chang et al., 2010; Chang & Wang,2010b). This study employed the Kinect technology and image recognition technology to create a rehabilitation systemapplicable for people with CP. The rehabilitation system in this study demonstrated the use of commercial off-the-shelfproducts and leveraged their advantages, such as affordability, availability, after-care service, technical support, and lowconcern about social stigma (Chang & Wang, 2010a; Shih, 2011; Shih et al., 2010a, 2010b). Consequently, the study shows apotential for people to rehabilitate at home because it is affordable and easy to set up, use and maintain, while still providingappropriate and accurate interactions so that the user can practice motor activities independently. Evidence shows that theproposed system with gesture recognition capabilities can be a viable rehabilitation tool that reduces impediments toindividuals performing the exercises. This device reduced staff intervention and enabled the participants to enhance theirmotivation for physical rehabilitation. Their achievement enhances a sense of self-determination, a feeling of independence,and improves the quality of life for individuals with disabilities (Felce & Perry, 1995; Wehmeyer & Schwartz, 1998; Wessels,Dijcks, Soede, Gelderblom, & De Witte, 2003).

References

Alankus, G., Proffitt, R., Kelleher, C., & Engsberg, J. (2010). Stroke therapy through motion-based games: A case study. In Proceedings of the 12th international ACMSIGACCESS conference on computers and accessibility (pp. 219–226).

Bach-y-Rita, P., Wood, S., Leder, R., Paredes, O., Bahr, D., Wicab Bach-y-Rita, E., et al. (2002). Computer-assisted motivating rehabilitation (CAMR) for institutional,home, and educational late stroke programs. Topics in Stroke Rehabilitation, 8(4), 1–10.

Balaam, M., Egglestone, S., Fitzpatrick, G., Rodden, T., Hughes, A.-M., Wilkinson, A., et al. (2011). Motivating mobility: Designing for lived motivation in strokerehabilitation. In Proceedings of the ACM CHI conference on human factors in computing systems (pp. 3073–3082).

Burdea, G., Deshpande, S., Liu, B., Langrana, N., & Gomez, D. (1997). A virtual reality-based system for hand diagnosis and rehabilitation. IEEE Transaction onRehabilitation Engineering, 6(2), 229–240.

Chang, Y. J., Peng, S. M., Chen, Y. R., Chen, H. C., Wang, T. Y., & Chen, S. F. (2010). Autonomous indoor wayfinding for individuals with cognitive impairments. Journalof NeuroEngineering and Rehabilitation, 7, 1–13.

Chang, Y. J., & Wang, T. Y. (2010a). Comparing picture and video prompting in autonomous indoor wayfinding for individuals with cognitive impairments. Personaland Ubiquitous Computing, 14, 737–747.

Chang, Y. J., & Wang, T. Y. (2010b). Indoor wayfinding based on wireless sensor networks for individuals with multiple special needs. Cybernetics and Systems, 41,317–333.

Chang, Y. J., Chen, S. F., & Huang, J. D. (2011). A Kinect-based system for physical rehabilitation: a pilot study for young adults with motor disabilities. Research inDevelopmental Disabilities, 32(6), 2566–2570.

Felce, D., & Perry, J. (1995). Quality of life: Its definition and measurement. Research in Developmental Disabilities, 16, 51–74.Foerster, F., Smeja, M., & Fahrenberg, J. (1999). Detection of posture and motion by accelerometry: A validation study in ambulatory monitoring. Computers in

Human Behavior, 15(5), 571–583.Gabriele, W., & Renate, S. (2009). Work loss following stroke. Disability and Rehabilitation, 31(18), 1487–1493.Gil-Gomez, J. A., Llorens, R., Alcaniz, M., & Colomer, C. (2011). Effectiveness of a Wii balance board-based system (eBaViR) for balance rehabilitation: A pilot

randomized clinical trial in patients with acquired brain injury. Journal of NeuroEngineering and Rehabilitation, 8(30), 1–9.Jack, D., Boian, R., Merians, A., Adamovich, S., Tremaine, M., Recce, M., et al. (2000). A virtual reality-based exercise program for stroke rehabilitation. In Proceedings

of the 4th international ACM conference on assistive technologies (pp. 56–63).Kleim, J., Jones, T., & Schallert, T. (2003). Motor enrichment and the induction of plasticity before or after brain injury. Neurochemical Research, 28(11), 1757–1769.Lai, S., Studenski, S., Duncan, P., & Perera, S. (2002). Persisting consequences of stroke measured by the Stroke Impact Scale. Stroke: A Journal of Cerebral Circulation,

33(7), 1840–1844.Lang, C., MacDonald, J., & Gnip, C. (2007). Counting repetitions: An observational study of outpatient therapy for people with hemiparesis post-stroke. Journal of

Neurologic Physical Therapy, 31(1), 3–10.Lloyd-Jones, D., Adams, R., Brown, T., Carnethon, M., Dai, S., De Simone, G., et al. (2010). Heart disease and stroke statistics – 2010 update: A report from the

American Heart Association. Circulation, 121(7), 46–215.Richards, S. B., Taylor, R. L., Ramasamy, R., & Richards, R. Y. (1999). Single subject research: Applications in educational and clinical settings. New York: Wadsworth.Rosenbaum, P., Paneth, N., Leviton, A., Goldstein, M., Bax, M., Damiano, D., et al. (2007). A report: The definition and classification of cerebral palsy April 2006.

Developmental Medicine & Child Neurology, 49, 8–14.Shaughnessy, M., Resnick, B., & Macko, R. (2006). Testing a model of post-stroke exercise behavior. Rehabilitation Nursing, 31(1), 15–21.Shih, C. H. (2011). Assisting People with Attention Deficit Hyperactivity Disorder by actively reducing limb hyperactive behavior with a gyration air mouse

through a controlled environmental stimulation. Research in Developmental Disabilities, 32, 30–36.Shih, C. H., Chang, M. L., & Shih, C. T. (2010a). A new limb movement detector enabling people with multiple disabilities to control environmental stimulation

through limb swing with a gyration air mouse. Research in Developmental Disabilities, 31, 875–880.Shih, C. H., Chang, M. L., & Shih, C. T. (2010b). A limb action detector enabling people with multiple disabilities to control environmental stimulation through limb

action with a Nintendo Wii Remote Controller. Research in Developmental Disabilities, 31, 1047–1053.Siegel, S., & Castellan, N. J. (1988). Nonparametric statistics for the behavioral sciences. New York: McGraw-HiU Book Company.Wagner, J., Lang, C., Sahrmann, S., Edwards, D., & Dromerick, A. (2007). Sensorimotor impairments and reaching performance in subjects with poststroke

hemiparesis during the first few months of recovery. Physical Therapy, 87(6), 751–765.Wehmeyer, M., & Schwartz, M. (1998). The relationship between self-determination and quality of life for adults with mental retardation. Education and Training

in Mental Retardation and Developmental Disabilities, 33, 3–12.Wessels, R., Dijcks, B., Soede, M., Gelderblom, G. J., & De Witte, L. (2003). Non-use of provided assistive technology devices, a literature overview. Technology and

Disability, 15, 231–238.