Developing an intelligent activity-based client-centred training system with a user-centred approach

Abstract

BACKGROUND:

In neurorehabilitation, clinicians and managers are searching for new client-centred task-oriented applications which can be administered without extra costs and effort of therapists, and increase the client’s motivation.

OBJECTIVE:

To develop and evaluate a prototype of an intelligent activity-based client-centred training (i-ACT) system based on Microsoft Kinect ${}^{\@setsize{\tiny}{7pt}{\vipt}{\@vipt}\textregistered}$ .

METHODS:

Within an iterative user centred process, the i-ACT prototype was developed and necessary features were established for use in neurological settings. After the test trial with a high fidelity prototype, the value, usefulness, and credibility were evaluated.

RESULTS:

Seven therapists participated in focus groups and 54 persons with neurological problems participated in test trials. A prototype was established based on the user’s experience. Results show that clients and therapists acknowledge the value and usefulness (clients 5.71/7; therapists 4.86/7), and credibility (clients 21.00/27; therapists 14.50/27) of i-ACT.

CONCLUSIONS:

Therapists want to be able to record an endless range of movements and activities which enables individualised exercise programs for persons with disabilities. For therapists it is important that the system provides feedback about the quality of movement and not only results. In future work, clinical trials will be performed towards feasibility and effectiveness of i-ACT in neurorehabilitation and other rehabilitation domains.

Keywords

Client-centred Kinect®rehabilitation task-oriented technology

1. Background

Persons with central nervous system diseases (PwCNS), such as multiple sclerosis (MS), stroke and spinal cord injury (SCI), experience loss of motor function which influences their performances of activities of daily life (ADL) [1, 2, 3, 4, 5]. To optimise rehabilitation outcome, rehabilitation should be client-centred (CC) and task-oriented (TO) [6, 7, 8, 9, 10, 11, 12, 13]. Although this approach has been proven beneficial and effective in neurological rehabilitation, implementation is difficult as it may be more time consuming for therapists, and costly for rehabilitation centres [6, 7, 8, 10, 14, 15, 16]. Therefore, clinicians and managers of rehabilitation centres are always searching for new client-centred task-oriented (CCTO) applications which can be administered without extra costs and effort of therapists, and increase the patient’s motivation [14, 16, 17, 18, 19, 20].

Technology-based systems, especially robotic systems, have already been used effectively in rehabilitation [5, 12, 13, 16, 21, 22, 23, 24]. However, they have some restrictions regarding incorporation into practice, especially in the home situation because of high costs [12, 13, 21], difficulty to use and/or install [12, 13, 21, 22, 23, 24], and the lack of possibilities to incorporate a full CC approach by offering individualised exercises. Only a few systems incorporate aspects of CC approach [7].

Other than robotics, there are off-the-shelf systems that are easy to use and low-cost, such as the Nintendo Wii and Microsoft Kinect ${}^{\text{(R)}}$ (MK). These systems are, in combination with virtual reality, promising as patients are motivated to use them and treatment adherence is great [14, 17, 18, 19, 20, 25]. However, up until now, they do not incorporate the CC approach [18, 26, 27, 28, 29, 30] and the commercially available devices and (exer) games, are not designed to meet rehabilitation goals such as feedback on compensation strategies, coordination patters, etc. [15, 27, 28, 31, 32].

In order to develop low-cost, off-the-shelf technologies suitable for CC rehabilitation use, the requirements and expectations of therapists and patients towards these technologies such as MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ were investigated [33]. The aim of this research was to develop a high fidelity CCTO rehabilitation system by taking the expectations and requirements of therapists and patients into account, named i-ACT (intelligent Activity-based Client-centred Training).

2. Method

2.1 User-centred process

A user-centred design approach was applied within an iterative process using focus groups and user tests to develop the i-ACT. The flow of the process is presented in Fig. 1. The in- and exclusion criteria are presented in Table 1. Ethical approval was obtained from the Ethical Comité of Jessa Ziekenhuis, Hasselt, Belgium (Reference number B243201420129). Written informed consent was obtained from all subjects before participation.

In phase one, focus groups with therapists (physio- and occupational therapists) were held at four rehabilitation centres in Flanders, Belgium (Jessa Ziekenhuis campus St-Ursula, Rehabilitation and MS centre Overpelt, Ziekenhuis Oost-Limburg campus St-Barbara, and MS Centre Melsbroek). During these focus groups, specific features were presented to therapists, such as criteria examples, feedback examples, visualisation of an exercise, visualisation of avatar etc. Therapists could give feedback and suggest adaptations to the presented features.

Figure 1.

Flowchart of the user-centred process.

Table 1

In- and exclusion criteria

	Patients	Therapists
Inclusion	Minimum 18 years of age	Therapist in rehabilitation centre
	Diagnosed with MS, stroke or SCI	Minimum 2 years of experience
	Upper limb dysfunction	Understand and speak Dutch
	Actively involved in rehabilitation program
Exclusion	Not able to understand and answer questions in Dutch
	Visual disabilities
	Attention disorder

Based on the results of phase one, engineers developed a basic version of i-ACT with adapted or new features (see results section).

Phase two consisted of user tests, organised with the basic i-ACT in two rehabilitation centres (Jessa Ziekenhuis campus St-Ursula and Rehabilitation and MS centre Overpelt), and completion of three questionnaires. The test trials were performed with PwCNS and therapists. The system was placed on a standard table of 74–75 cm high. The participants had to sit or stand at a distance of 1–3 m. Therapists were asked to record one new movement before performing 3 $\times$ 4 repetitions of their own movement. Patients were asked to perform 3 $\times$ 4 repetitions of an exercise that was provided by the therapists. Extra materials such as a ball, cones, chair, table, etc., were provided. Both patients and therapists were asked to give feedback and make suggestions regarding the developed features of the basic version. After the test trials, the participants were asked to complete three questionnaires to gather relevant clinical data for the development of the prototype: 1) a self-prepared questionnaire (further referred to as i-ACT questionnaire) containing various questions on features of the system, such as colours, representation of environment, feedback, etc., 2) the subscale value/usefulness (range of scores 0–7 with a total of 49) of the Intrinsic Motivation Inventory (IMI) [34, 35, 36] to estimate the value/usefulness for the participants, 3) the subscale credibility (range of scores 1–9 with a total of 27) of the Credibility/Expectancy Questionnaire (CEQ) [37] to measure the rationale credibility of i-ACT.

Based on the feedback, suggestions, and the clinical outcomes gathered during phase 2, the basic version was adapted and more features were developed towards a prototype of the i-ACT (see results section).

2.2 Data analysis

Focus groups were analysed using thematical analysis. Descriptive analysis of questionnaire results was executed with Excel 2007 and IMB SPSS Statistics (version 22.0; Armonk, New York, USA).

3. Results

3.1 Participants

Feedback and suggestions of all 61 participants in phase 2 will be described. Seven therapists participated in the focus groups. Fifty-four persons were involved in the tests. Characteristics are presented in Table 2.

Table 2
Participant characteristics

	Focus groups ( $n=$ 7)		Test trials ( $n=$ 54)
	Clients	Therapists	Clients	Therapists
Gender (m/f)	0	1/6	17/11	3/23
Age (y)	NA	39.60 $\pm$ 10.09	59.00 $\pm$ 12.50	29.00 $\pm$ 7.03

Age values are means $\pm$ standard deviation. NA: not applicable.

3.2 Focus groups

Out of the focus groups with the therapists, 4 main topics where revealed: users’ requirements, graphic requirements, apparatus, and feedback.

Additionally 2 topics where added in the results section, “creating activities and training” and “quality of movement”, which are involved in the 4 main topics and are relevant for understanding the concept and choices made during the development of the i-ACT prototype.

3.3 Users’ requirements

For the therapists it was clear that the application should allow to 1) record and store a movement for therapists, 2) to view a movement in a 3D environment, 3) to view a real time movement of the patient, and 4) to compare the real time movement of the patient with the recorded movement of the therapist (The system or technology should give feedback to the patient. I think that sometimes patients don’t really know whether they did the exercise correct. For example, did I use enough force, did I move correctly, etc.; occupational therapist). An overview of information regarding use and implementation of specific features, setup and content can be found in Table 3.

The information gathered from the therapists during the focus group in phase 1, led to the development of a basic i-ACT version. This version consisted of a recording feature and a real time feature, but no comparison could be established. The environment was a black background with a white stickman.

With the basic i-ACT, test trials were performed with patients and therapists (phase two). From the answers of the patients in the i-ACT questionnaire, as seen in Table 3, it is clear that a lot of requirements relate to the requirements of the therapists. Furthermore, patients stated their preference to be able to practise activities ( $n=$ 23) compared to playing games ( $n=$ 11). Twenty patients wanted a combination of activities and games (Games are great, you forget that you are exercising. But I know I need my exercises and pay attention to the correct movement, so activities are also important. So if I can choose, I would like to have both; Patient with MS).

The information of both phases was implemented in the development of the prototype, which is described in the next sections.

Table 3
Therapist’s and patients’ requirements

Therapist’s requirements
Environment	Setup and virtual environment as simple as possible but attractive
Client-centred	Ability to assign movements/exercises/trainings to specific patient
	Ability to adapt training parameters to specific patient
	Adaptability of tolerated range of motion
	Adaptability of size and position of target
Feedback	Compensatory movements should be detected and appropriate feedback given
	Feedback during movement/exercises
	Auditory as well as visual feedback
	Feedback regarding performance of movement/exercises
	Feedback for patients regarding results in graphs
Patient’s requirements
Environment	Application needs to be simple, clear and intuitive
	Plain virtual environment with clear distinction between background and avatar
Client-centred	Specific activities/exercises which patients also perform during guided therapy
	Specific exercise activities or exergames instead of only games
Feedback	Colour coding: red is wrong, green is good
	If possible besides colour coding also sound when movement is wrong

3.4 Graphic requirements

Concerning virtual setup and environment, all therapists were unanimous that the environment should be as simple as possible for patients, especially dealing with PwCNS. The environment should be simple but attractive in order to motivate persons to exercise. From the patient’s perspective, the application needed to be simple, clear and intuitive (I don’t like it when there is too much going on…I can’t focus when there is a lot to see on the screen; patient with stroke). For the prototype, a plain room scene was chosen, as seen in Fig. 2 instead of a fully equipped room, e.g. a kitchen or garden.

Figure 2.

3D scene in therapist mode.

During the focus groups and test trials, patients expressed that there had to be a clear distinction between the background and the avatar. The latter was suggested above a stick man (which was used during the focus groups and tests), or a real image of themselves (I don’t like looking at myself. If you just put a normal looking man or puppet there, that’s fine; Patient with MS). For the prototype a simple avatar was developed (as seen in Fig. 2).

3.5 Apparatus

For therapists it is important to have an apparatus which is easy to set up, does not take up too much space, is easy to use (It’s ok if you spend a bit more time the first time you use it for a patient, but the next session, it should just work, so that you don’t lose time afterwards …. It should store the patients data; occupational therapist), and is not too expensive (Financially there are some possibilities for buying technology. The problem is, that is changes constantly and you have to make smart decisions so you can use the technology for longer than just one year. And most of the time, the additional licenses are the most expensive; occupational therapist). If possible, therapists prefer a system where the patient is not strapped in or sensors have to be placed on the body. Therefore, developers chose to use the MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ for Windows because it is possible to record different streams without body sensors: body frame, body index, depth view, IR view, audio, compressed colour, gesture recognition and face detection. The MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ for Windows has a depth sensor (512 $\times$ 424; 30 Hz) and a colour camera (1080 p, 30 Hz). The size of the MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ for Windows is small (24.9 cm $\times$ 6.6 cm $\times$ 6.7 cm) and the weight is considered light (1.5 kg). The MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ software development kit (SDK) 2.0 was chosen as it creates the opportunity to quickly develop applications that use the MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ stream [25]. The SDK supports any .Net language [38].

For the technical development of i-ACT, the cross-platform Unity3D was used. The MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ SDK v2.0 has an integrated Unity plugin, which makes it easier to develop environments using the MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ [38]. Another reason to use Unity3D, is the graphical aspect of the project. Implementing game aspects like sound, game objects, backgrounds, lights, and other graphical items are easier with Unity3D compared to other developing tools. Also, implementation of a game loop is easier with Unity3D. The patient flow of i-ACT is similar as in a game loop. Furthermore, the project requires implementation of a 3D avatar. The avatar must be drawn and positioned using the data of the MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ stream.

The application is built using an administrative and client model. The administrative model can be seen as the therapist, the client model as the patient.

3.6 Creating activities and training

Therapists reported that individualised training where the patient is the central component, is very important in neurorehabilitation. As mentioned before, a CCTO training is key to optimising rehabilitation outcome. To install a CCTO training programme based on the needs of the patient, therapists should be able to record and set up any activity (and as a consequence certain analytical and/or functional movements) that is important to the patient. The steps therapists need to take for recording movements and setting up exercises should be intuitive, logical, and understandable.

To go from an activity towards a training, therapists suggested to work with the terms “movement”, “exercise” and “training”, which are considered basic terms in physical exercise.

Consequently, the following jargon is implemented in i-ACT. Movement is the posture or movement recorded by the MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ . A movement has no settings regarding the quality of movement. The only info is the data that is necessary to draw the avatar on the screen. Changeable options are the ‘selected body parts’ and the ‘preposition’. Deselected body parts will not be recorded, these body parts will be drawn in a default position. A preposition will force the avatar in a standing or sitting position. A movement can be analytical or functional. To compare the real time movement of a patient with a recorded movement, parameters are required. An exercise consists of a movement with parameters concerning repetition, stability and mobility. A training consists of two or more exercises.

The reported requirements of the therapists towards training parameters that can be used in exercises and trainings are adaptable speed of recorded movement, adaptable number of repetitions, sets and amount of rest between sets and/or repetitions, and recording of compensatory movements.

The user management is designed as simple as possible. A therapist can log in to add, delete or change movements, exercises and/or trainings. Therapist can also assign a movement, exercise and/or training to a specific client. When patients perform a training, the movements and/or exercises are provided one by one.

3.7 Quality of movement

During the focus groups, therapists suggested that it would be important to be able to change the tolerated amount of compensatory movement (range of motion (ROM)) allowed within a movement and the size of the target the patients need to reach. To address these two suggestions, point(s) of stability (i.e. body regions that are marked as areas where no movement is allowed beyond a certain bandwidth, ie to avoid compensatory movement) and point(s) of mobility (i.e. target areas where the patients have to pass during the movement trajectory in order to guide movement) were implemented.

Visually, the point of stability is a sphere around a joint. The therapist can decide the amount of spheres, the size of the sphere(s) and the joint(s). If in the real time frame the joint moves out of the sphere, it is decided that the movement at that point in time, for the given joint, is out of range. Then, the sphere turns from green to red and a textual remark is given on the screen (as seen in Fig. 3).

Figure 3.

Point of stability in (left) and out of margin (right).

The comparison between the movement registered from the therapist and performed by the patient, allows to provide feedback on the performance of the movement. The difference between the real time movement of the patient and the recorded movement of the therapist can be considered as the quality of the movement. In theory it can be assumed that if the recorded video stream is identical to the real time video stream, the movements are equal; If the movements are equal, the quality of the real time movement is ‘perfect’ and vice versa. The difference in ROM and timing tolerated between the two movements can be seen as the margin of quality. The tolerated difference in ROM can be set by the therapist. To calculate the tolerated difference, two frames have to be compared with each other. The difficult part is to determine which frame of the recorded movement must be compared with which real time frame. In a perfect situation it would be easy: compare the first frames, than the second frames, etc. This is not applicable since it is impossible that the real time movement is moving with the exact same speed as the recorded movement. There is always a slight delay. This would result in a sort of timing game instead of monitoring the quality. Figure 4 is a simple representation of this problem. Another way to calculate the tolerated difference, is only to take the next recorded frame if the previous frame is within margin. In theory this can work, practically however, the result is a frustrated patient, trying to complete a simple exercise; The data from the MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ is never exactly the same when comparing two movements, recorded with different settings and/or different environment and/or different persons with different clothes. This is also shown in Fig. 4. The comparison would only go further if the real time movement would be identical to the recorded movement. Programming applications by using a compare function as described above, would result in games that would frustrate the patient as the movement of the avatar would not be fluent. Therefore, the compare method in i-ACT is done by comparing the real time frame with the first frame of the recorded movement.

Figure 4.

Comparing movements. In the situation on the bottom right, the quality of the real time movement (right) is not good since the frame is different from the recorded frame (left).

Since the stability only compares the real time view with the first recorded frame, there has to be a way of forcing the user to do the movement as shown by the recorded movement. Waypoints will have this effect, and were named mobility points or targets.

Given that the avatar on the screen is drawn notwithstanding to the body dimensions of the person in front of the MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ , the avatar always has the same height and ratio. This is possible since the avatar is drawn using rotations. Waypoints are calculated using the avatar’s position, i.e. time label and body joint. With those parameters, the position of where the body joint would be rendered, is calculated. This is the position of the waypoint, which will be checked for collisions during the real time movement. An example of waypoints can be seen in Fig. 5.

Figure 5.

Representation of waypoint in user (client) mode.

3.8 Feedback

Feedback regarding quality of movement during the exercise is explained above. In the i-ACT questionnaire, participants were asked about their preference regarding feedback (i.e. timing of feedback, style, and type of feedback). The responses of all participants are presented in Table 4.

Table 4
User’s responses in terms of feedback

Timing of feedback
During	42.60%	( $n=$ 23)
After	24.07%	( $n=$ 13)
Both	33.33%	( $n=$ 18)
Style of feedback
Auditory	20.37%	( $n=$ 11)
Visual	35.19%	( $n=$ 19)
Textual	3.70%	( $n=$ 2)
Auditory and visual	38.89%	( $n=$ 21)
No preference	1.85%	( $n=$ 1)
Type of feedback
Feedback on performance	68.52%	( $n=$ 37)
Feedback on result	14.81%	( $n=$ 8)
Both	16.67%	( $n=$ 9)

During the exercise, visual cues on the right side of the screen to see how many movements patients have performed and still have to perform, and on the bottom of the screen a visual presentation of the training is implemented. See Fig. 6 for some of these features.

Figure 6.

Features in user (client) mode. 1: Repetitions completed and uncompleted; 2: Visual representation of training progression; 3: Written representation of training progression; 4: Symbolic representation of training progression.

After finishing a training, results are shown in numbers and stars (up to 5 stars) representing the amount of good movements.

3.9 Usefulness and credibility of i-ACT

Table 5
Results of IMI and CEQ

	IMI ${}^{*}$	CEQ ${}^{\phi}$
Patients	5.71 (5.07–6.50)	21.00 (19.00–25.00)
Therapists	4.86 (3.39–5.32)	14.50 (9.00–21.00)

IMI: Intrinsic motivation inventory; CEQ: Credibility expectancy questionnaire. Scores are presented as median (Q1–Q3). ${}^{*}$ : score on 7. ${}^{\phi}$ : score on 27.

Median values of IMI and CEQ are presented in Table 5. The data from the IMI subscale value/usefulness shows that even in the early development participants value the basic i-ACT. And although both patients and therapists find i-ACT useful in neurorehabilitation, patients score higher than therapists. Data from the CEQ credibility subscale are similar: therapists score lower than patients on CEQ, respectively median of 14.50 and 21.00 on a total of 27.

4. Discussion

The aim of this study was to develop and evaluate a prototype of an intelligent activity-based client-centred training system using an iterative user-centred process. A prototype was developed based on experiences from patients and therapists. The developed system, intelligent Activity-based Client-centred Training (i-ACT), was found valuable and useful, and patients as well as therapists believe in the i-ACT as a training tool.

For the technical development of i-ACT, the Unity platform was chosen. Unity is a typical developing tool for 2D and 3D games. The advantage of using Unity is the capability of using a game loop, objects, scripts, etc. which gives developers a fast way of designing interactive games. Although i-ACT is not a game, it requires an advanced administration interface with menus, drop down boxes, graphics, etc. These latter features are much more difficult to develop with Unity, but the choice was made in favour of a good user experience.

An important feature of i-ACT is the possibility of a wide range of exercises in order to be CC. For therapists, all the exercises must be quick and easy to perform, if possible without any interaction of the therapist. The variety of movements/exercises asks for a lot of parameters. Developing a programme with a wide range of exercises, with as less as possible parameters to adjust is a challenge.

The feedback given to the patient concerning quality of movement can be given in real time or after an exercise. Feedback after the exercise is naturally seen from a developer’s point of view. After an exercise is completed, i-ACT can analyse the movements and give the correct feedback. From a therapist’s and patient’s perspective, the feedback must be real-time. This makes the feedback-feature difficult since it is impossible to compare the complete movement/exercise. To counteract these issues, the points of stability and waypoints were created, as described above. Also, MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ is based on sensors that register the movement of the player itself, so it replicates the real movement of the patient, with incorrect or impaired movements, rather than replacing the movement with visually normal movement [25]. So compensational movements are registered in real time. Providing feedback in real-time, is important. Baran et al. [39] demonstrated that providing real-time feedback is an effective motor training tool. They used the Adaptive Mixed Reality Rehabilitation (AMRR) system for reach and grasp tasks [39].

Additionally to the feedback on quality of movement, therapists would like to be able to have video footage of the patient as well, even knowing that ethical considerations need to be made. This latter could not be addressed during the development process as ethical issues would interfere.

Also, in the discussions with therapists, it was suggested to implement auditory cues when a target is reached. This will be implemented in a future i-ACT version.

Furthermore, therapists suggested the implementation of multiplayer feature. This issue is considered but not implemented in i-ACT as this is in conflict with the CC approach.

During the test trials, patients expressed that there had to be a clear visual distinction between the background and the avatar. A human-like avatar was suggested above a stickman, or a real image of themselves.

The research group acknowledges that the interface and content should be made more attractive and motivational, thus a different 3D scene with another, more human-like avatar will be implemented in the future prototype. Also, the feedback for patients after exercise and/or training will be made more attractive.

The weaknesses of the i-ACT are that the system might not be as accurate as sensor-based systems in training. While the validity and accuracy of the MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ sensor is strong regarding postural control and standing balance [40, 41], and reproducibility in planar motion analysis is similar to traditional marker-based stereogrammetry [42]. To our knowledge, no evidence is provided concerning accuracy of full body movements measured with MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ compared to traditional marker-based stereogrammetry. Furthermore, i-ACT does not register any real-object manipulation. At the moment, only persons can be detected by the i-ACT and movements of these persons are registered. Additionally, MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ is not further supported by Microsoft, so the developers are looking into other devices to run the i-ACT software.

The i-ACT is not another exergame system. To the authors’ knowledge, there is no other technology-based system available that applies a full CCTO approach, provides feedback on the quality of movement, and have possibilities in rehabilitation as well as home environment [43]. Other technology-based systems that have been developed and implemented (aspects of) CC, such as TagTrainer [44], do not provide feedback regarding the quality of movement, only on the result. The systems that do provide feedback on the quality of movement, often use body worn sensors [7, 24]. Other studies based on MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ sensor are mainly focussing on reaching and grasping, while feedback is given whether a target is reached or not [43]. The major strength of i-ACT is that it uses a full client-centred approach. Also, patients can exercise their activities and receive feedback concerning their quality of movement. Furthermore, the i-ACT is affordable as it is a system with the MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ , an off-the-shelf product, and the software is developed by engineers and researchers within a research project.

In the focus groups, therapists were positive about the features developed for i-ACT. The main concern was the virtual environment which was not yet fully developed. Most quoted statements by participants are implemented in the most recent prototype, such as possibility to change the amount of compensatory movements allowed and the size of the presented target.

The results from the clinical tests indicated that both patients and therapists acknowledge the value, usefulness and credibility of the prototype presented in the trials. The high median of 21.00 in patients, suggests that patients think that i-ACT could be a successful tool to support them in their rehabilitation. These findings are in accordance to the research by Park et al. who stated that additional VR training with Xbox Kinect ${}^{\text{(R)}}$ can be an effective approach to improve motor function during stroke rehabilitation [25].

Some methodological considerations have to be made as not all participants completed the IMI; These participants answered ‘not applicable’ to the questions of the subscale value/usefulness. Their reason was that the trial was only a one time test and based on that they could not paint a clear picture of the value/usefulness of i-ACT and/or developed features. Also, focus groups and trials were held at four rehabilitation centres in Flanders (Belgium) with a selected group of PwCNS and therapists working in neurorehabilitation. Therefore no general conclusions can be drawn for the total population of neurological disorders. Furthermore, the target group should be expanded when testing the next iteration with patients and therapists.

Further development should incorporate integrating an avatar, feedback during but also after exercises with a focus on auditory and visual feedback about quality of performance. Participants suggested a combination of activities with games. The solution could be to develop an exergame where exercise and gaming are combined by performing functional movements in a game-like environment. The researchers acknowledge the fact that virtual reality increases the motivation and adherence [14, 19] but developers need to be cautious as the goal of the project is to develop a CCTO training with MK ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ and not a new game.

In future work, research studies are planned to evaluate the feasibility, motivation, usability, credibility/expectancy and effectivity of i-ACT in different rehabilitation centres with a variation of disorders.

5. Conclusion

The prototype of i-ACT is a client-centred task-oriented training system based on Microsoft Kinect ${}^{\@setsize{\scriptsize}{9.5pt}{\viiipt}{\@viiipt}\textregistered}$ for Windows that allows therapists to record a wide range of movements and activities as exercise instructions, while providing real time feedback about the quality of movement. Individual training programs and training parameters, combined with point(s) of stability and mobility, can be set by the therapist. The i-ACT is already found useful by patients and therapists, and patients also strongly believe that i-ACT can successfully support them in their rehabilitation.

The feedback and results from this study will be used to adjust the i-ACT further. In future work, clinical trials will be performed towards feasibility and effectiveness of i-ACT in neurorehabilitation and other rehabilitation domains.

Footnotes

Acknowledgments

The authors want to thank all participants of this study. They also thank Frederik Smolders, Lisa Pass, Stephen Beuls, Celine De Pauw, Lieze Vandevenne and Peter Feys for their contribution. This study was part of the Technology Transfer (Tetra) project “Open intelligent rehabilitation framework for client-centred functional therapy with motion detection systems” (project number 140324) funded by Flemish Government Vlaio.

Conflict of interest

None to report.

References

Van Peppen

Kwakkel

Wood-Dauphinee

Hendriks

Van der Wees

Dekker

. The impact of physical therapy on functional outcomes after stroke: what’s the evidence? Clin Rehabil. 2004; 18: 833-62.

Rietberg

Brooks

Uitdehaag

Kwakkel

. Exercise therapy for multiple sclerosis. The Cochrane Library. 2005.

Smidt

de Vet

Bouter

Dekker

Arendzen

de Bie

, et al. Effectiveness of exercise therapy: a best-evidence summary of systematic reviews. Aust J Physiother. 2005; 51(2): 71-85.

Kwakkel

van Peppen

Wagenaar

Wood Dauphinee

Richards

Ashburn

, et al. Effects of augmented exercise therapy time after stroke: a meta-analysis. Stroke. 2004; 35(11): 2529-39.

Gijbels

Lamers

Kerkhofs

Alders

Knippenberg

Feys

. The Armeo Spring as training tool to improve upper limb functionality in multiple sclerosis: a pilot study. J Neuroeng Rehabil. 2011; 8: 5.

Spooren

Janssen-Potten

Kerckhofs

Bongers

Seelen

. ToCUEST: a task-oriented client-centered training module to improve upper extremity skilled performance in cervical spinal cord-injured persons. Spinal Cord. 2011; 49(10): 1042-8.

Timmermans

AAA

Geers

RPJ

Franck

Dobbelsteijn

Spooren

AIF

Kingma

, et al., eds. T-TOAT: a method of task-oriented arm training for stroke patients suitable for implementation of exercises in rehabilitation technology. Rehabilitation Robotics, 2009 ICORR 2009 IEEE International Conference on; 23–26 June 2009.

Wressle

Eeg-Olofsson

Marcusson

Henriksson

. Improved client participation in the rehabilitation process using a client-centred goal formulation structure. Journal of Rehabilitation Medicine. 2002; 34(1): 5-11.

Donnelly

Eng

Hall

Alford

Giachino

Norton

, et al. Client-centred assessment and the identification of meaningful treatment goals for individuals with a spinal cord injury. Spinal Cord. 2004; 42(5): 302-7.

10.

Spooren

Janssen-Potten

Kerckhofs

Bongers

Seelen

. Evaluation of a task-oriented client-centered upper extremity skilled performance training module in persons with tetraplegia. Spinal Cord. 2011; 49(10): 1049-54.

11.

Timmermans

Seelen

Geers

Saini

Winter

te Vrugt

, et al. Sensor-based arm skill training in chronic stroke patients: results on treatment outcome, patient motivation, and system usability. IEEE Transactions on Neural Systems and Rehabilitation Engineering: A Publication of the IEEE Engineering in Medicine and Biology Society. 2010; 18(3): 284-92.

12.

Timmermans

Lemmens

Monfrance

Geers

Bakx

Smeets

, et al. Effects of task-oriented robot training on arm function, activity, and quality of life in chronic stroke patients: a randomized controlled trial. J Neuroeng Rehabil. 2014; 11: 45.

13.

Vanmulken

Spooren

Bongers

Seelen

. Robot-assisted task-oriented upper extremity skill training in cervical spinal cord injury: a feasibility study. Spinal Cord. 2015; 53(7): 547-51.

14.

Paquin

Crawley

Harris

Horton

. Survivors of chronic stroke – participant evaluations of commercial gaming for rehabilitation. Disability and Rehabilitation. 2016; 38(21): 2144-52.

15.

Saposnik

Levin

. Virtual reality in stroke rehabilitation a meta-analysis and implications for clinicians. Stroke. 2011; 42(5): 1380-6.

16.

Prange

Jannink

Groothuis-Oudshoorn

Hermens

Ijzerman

. Systematic review of the effect of robot-aided therapy on recovery of the hemiparetic arm after stroke. Journal of Rehabilitation Research and Development. 2006; 43(2): 171-84.

17.

Gamito

Oliveira

Coelho

Morais

Lopes

Pacheco

, et al. Cognitive training on stroke patients via virtual reality-based serious games. Disability and Rehabilitation. 2017; 39(4): 385-8.

18.

Webster

Celik

. Systematic review of Kinect applications in elderly care and stroke rehabilitation. J Neuroeng Rehabil. 2014; 11: 108.

19.

Lozano-Quilis

Gil-Gomez

Albiol-Perez

Palacios-Navarro

Fardoun

, et al. Virtual rehabilitation for multiple sclerosis using a kinect-based system: randomized controlled trial. JMIR Serious Games. 2014; 2(2): e12.

20.

Lloréns

Noé

Colomer

Alcañiz

. Effectiveness, usability, and cost-benefit of a virtual reality-based telerehabilitation program for balance recovery after stroke: a randomized controlled trial. Archives of Physical Medicine and Rehabilitation. 2015; 96(3): 418-25. e2.

21.

Feys

Coninx

Kerkhofs

De Weyer

Truyens

Maris

, et al. Robot-supported upper limb training in a virtual learning environment: a pilot randomized controlled trial in persons with MS. J Neuroeng Rehabil. 2015; 12: 60.

22.

Fasoli

Krebs

Hogan

. Robotic technology and stroke rehabilitation: translating research into practice. Top Stroke Rehabil. 2004; 11: 11-9.

23.

Lemmens

Timmermans

Janssen-Potten

Pulles

Geers

Bakx

, et al. Accelerometry measuring the outcome of robot-supported upper limb training in chronic stroke: a randomized controlled trial. PloS One. 2014; 9(5): e96414.

24.

Duff

Chen

Cheng

Liu

Blake

Wolf

, et al. Adaptive mixed reality rehabilitation improves quality of reaching movements more than traditional reaching therapy following stroke. Neurorehabil Neural Repair. 2013; 27(4): 306-15.

25.

Park

Lee

. Effects of virtual reality training using xbox kinect on motor function in stroke survivors: a preliminary study. Journal of Stroke and Cerebrovascular Diseases: The Official Journal of National Stroke Association. 2017; 26(10): 2313-9.

26.

Knippenberg

Verbrugghe

Lamers

Palmaers

Timmermans

Spooren

. Markerless motion detection systems as training tool in neurological rehabilitation: a systematic review. Submitted.

27.

Brokaw

Eckel

Brewer

. Usability evaluation of a kinematics focused Kinect therapy program for individuals with stroke. Technology and Health Care: Official Journal of the European Society for Engineering and Medicine. 2014.

28.

Sin

Lee

. Additional virtual reality training using Xbox Kinect in stroke survivors with hemiplegia. American Journal of Physical Medicine & Rehabilitation [Internet]. 2013; 92(10): 871-80. Available from: http://onlinelibrary.wiley.com/o/cochrane/clcentral/articles/711/CN-00960711/frame.html.

29.

Levac

Pradhan

Espy

Fox

Deutsch

. Usability of the ‘Kinect-ing’ with clinicians website: a knowledge translation resource supporting decisions about active videogame use in rehabilitation. Games for Health Journal. 2018; 7(6): 362-8.

30.

Valdés

Glegg

SMN

Lambert-Shirzad

Schneider

Marr

Bernard

, et al. Application of commercial games for home-based rehabilitation for people with hemiparesis: challenges and lessons learned. Games for Health Journal. 2018; 7(3): 197-207.

31.

Pastor

Hayes

Bamberg

. A feasibility study of an upper limb rehabilitation system using Kinect and computer games. Conf Proc IEEE Eng Med Biol Soc. 2012; 2012: 1286-9.

32.

Bao

Mao

Lin

Qiu

Chen

, et al. Mechanism of Kinect-based virtual reality training for motor functional recovery of upper limbs after subacute stroke. Neural Regeneration Research. 2013; 8(31): 2904-13.

33.

Knippenberg

Spooren

. Opportunities of markerless motion detection systems for use in neurological rehabilitation: a qualitative study on patient and therapist perspective. Austin Journal of Robotics and Automation. 2016; 3(1).

34.

Deci

Eghrari

Patrick

Leone

. Facilitating internalization: the self-determination theory perspective. Journal of Personality. 1994; 62(1): 119-42.

35.

Ryan

. Control and information in the intrapersonal sphere: an extension of cognitive evaluation theory. Journal of Personality and Social Psychology. 1982; 43(3): 450.

36.

Ryan

Mims

Koestner

. Relation of reward contingency and interpersonal context to intrinsic motivation: a review and test using cognitive evaluation theory. Journal of Personality and Social Psychology. 1983; 45(4): 736.

37.

Devilly

Borkovec

. Psychometric properties of the credibility/expectancy questionnaire. Journal of Behavior Therapy and Experimental Psychiatry. 2000; 31: 73-86.

38.

Microsoft. Develop for Windows 2019 [updated 2019]. Available from: https://developer.microsoft.com.

39.

Baran

Lehrer

Siwiak

Chen

Duff

Ingalls

, et al. Design of a home-based adaptive mixed reality rehabilitation system for stroke survivors. Conf Proc IEEE Eng Med Biol Soc. 2011; 2011: 7602-5.

40.

Clark

Pua

Fortin

Ritchie

Webster

Denehy

, et al. Validity of the Microsoft Kinect for assessment of postural control. Gait & Posture. 2012; 36(3): 372-7.

41.

Clark

Pua

Oliveira

Bower

Thilarajah

McGaw

, et al. Reliability and concurrent validity of the Microsoft Xbox One Kinect for assessment of standing balance and postural control. Gait & Posture. 2015; 42(2): 210-3.

42.

Bonnechere

Jansen

Salvia

Bouzahouene

Omelina

Moiseev

, et al. Validity and reliability of the Kinect within functional assessment activities: comparison with standard stereophotogrammetry. Gait & Posture. 2014; 39(1): 593-8.

43.

Knippenberg

Verbrugghe

Lamers

Palmaers

Timmermans

Spooren

. Markerless motion capture systems as training device in neurological rehabilitation: a systematic review of their use, application, target population and efficacy. J Neuroeng Rehabil. 2017; 14(1): 61.

44.

Tetteroo

Timmermans

Seelen

Markopoulos

. TagTrainer: supporting exercise variability and tailoring in technology supported upper limb training. Journal of Neuroengineering and Rehabilitation. 2014; 11(1): 140.

Developing an intelligent activity-based client-centred training system with a user-centred approach

Abstract

BACKGROUND:

OBJECTIVE:

METHODS:

RESULTS:

CONCLUSIONS:

Keywords

1. Background

2. Method

2.1 User-centred process

3. Results

3.1 Participants

Table 2 Participant characteristics

3.3 Users’ requirements

Table 3 Therapist’s and patients’ requirements

3.6 Creating activities and training

3.7 Quality of movement

Table 4 User’s responses in terms of feedback

Table 5 Results of IMI and CEQ

5. Conclusion

Footnotes

Acknowledgments

Conflict of interest

References

Table 2
Participant characteristics

Table 3
Therapist’s and patients’ requirements

Table 4
User’s responses in terms of feedback

Table 5
Results of IMI and CEQ