Serious Games for Assessment and Training in Post-stroke Robotic

Upper-limb Telerehabilitation

Cristina Rodríguez-de-Pablo

, Joel C. Perry

, Sivakumar Balasubramanian

, Aitor Belloso

Andrej Savić

2,3

, Tijana Dimkić Tomić

4,5

and Thierry Keller

Rehabilitation Technologies, Health Division, TECNALIA, San Sebastian, Spain

Tecnalia Serbia Ltd, Belgrade, Serbia

Signals and Systems Department, University of Belgrade - School of Electrical Engineering, Belgrade, Serbia

Clinic for rehabilitation Dr Miroslav Zotović, Belgrade, Serbia

University of Belgrade, School of Medicine, Belgrade, Serbia

Keywords: Serious Games, Upper-limb Rehabilitation, Robotic Rehabilitation, Quantitative Assessment, Reach

Training, Motivation, Stroke, Home-based Telerehabilitation.

Abstract: Research shows that better results in post-stroke rehabilitation are obtained when patients receive more

intensive therapy. However, the increasing affected population and the limited healthcare resources prevent

the provision of intense rehabilitation care. Thus, there is a need for a more autonomous and scalable care

provision methods that can be transferred out of the clinic and into home environments. Serious games in

combination with robotic rehabilitation can provide an affordable, engaging, and effective way to intensify

treatment, both at the clinic and at home. Furthermore, they can offer quantitative assessment of motor

performance, allowing individualized treatments and to keep the patient and their therapists informed about

therapy progress. Towards this end, a set of games for assessment and training of upper-limb motor

impairment after stroke with the ArmAssist is presented. A special effort has been made to design the

assessment games in order to be able, not only to measure the effectiveness of the training, but also to

compare the assessment results with the standard assessment scales used in the clinic. Feedback from

usability testing of previous versions of the system has also been crucial for the final design. Preliminary

results of an ongoing clinical testing are presented.

1 INTRODUCTION

Research confirms that better results in terms of

rehabilitation outcome are obtained in specialized

care centres where patients receive more therapy per

day and with more quality for extended periods of

time (Sluijs et al., 1993; Kwakkel, et al., 1997).

However, the amount of professionally-

supported rehabilitation training provided to the

average patient is still far from the ideal. A Dutch

report published in 2008 (Peerenboom et al., 2008)

reported that the average treatment time for stroke

patients in skilled nursing facilities was about 4.5

hours per week, and only about 2 of them was spent

in physical therapy.

Healthcare resources are suffering from constant

cutbacks due to economic constraints and the

prevalence of stroke continues to increase

dramatically together with the aging of the

population. As a matter of fact, European statistics

as a whole report that nearly 1 million people

experience a first or recurrent stroke each year

(Hesse et al., 2005). In that context, the amount of

intensive therapy required cannot be provided in the

regular way that rehabilitation care has been

provided up to now. There is a need for more

autonomous and scalable care that can be transferred

out of the clinic. Robot-assisted rehabilitation based

on serious games may offer that care.

Robots offer precision and repeatability of

movements that can be used to provide safe and

intensive rehabilitation exercises. Serious games can

provide an enjoyable and effective way to motivate

patients to increase both the quality and quantity of

therapy by decreasing the monotony of performing

hundreds of repeated motions and by providing

126

Rodriguez-de-Pablo C., Perry J., Balasubramanian S., Belloso A., Savi

c A., Dimki

c Tomi

c T. and Keller T..

Serious Games for Assessment and Training in Post-stroke Robotic Upper-limb Telerehabilitation.

DOI: 10.5220/0005168601260134

In Proceedings of the 2nd International Congress on Neurotechnology, Electronics and Informatics (VirtRehab-2014), pages 126-134

ISBN: 978-989-758-056-7

 2014 SCITEPRESS (Science and Technology Publications, Lda.)

challenging performance feedback. In addition, these

technologies can also be used to remotely assess the

motor impairment of patients without the need for

the therapist to be present. Therefore, a large

increase in robotic rehabilitation based on serious

games has been seen over the last years (Robertson

et al., 2010). Recent commercial players in the field

include Telefonica, MediTouch, HomeTelemed,

Tyromotion and Hocoma.

However, even with the players involved,

commercial success is limited. The solution may lie

in the development of a system that can be easily

integrated with current practice and which can verify

its effectiveness correlating the results of both

assessment and training tools with the assessment

scales and physical therapy performed at the clinic.

Towards this end, this paper presents a new

version of 5 games for assessment and 5 games for

training of post-stroke arm used with the ArmAssist

together with the TeleReha web platform for at-

home telerehabilitation. ArmAssist allows the planar

movement of the arm by supporting the arm while

measuring the movement parameters (2D position,

orientation and arm support/lifting force) that are

used to control the games. The TeleReha platform

allows the patient to perform game-based

rehabilitation both at the clinic and at home, while

the doctor is able to monitor the progress, update the

therapy correspondingly, and communicate with the

patient when necessary. The system components and

functionality have been previously described in

publications (Zabaleta et al., 2011; Perry et al.,

2012; Arcas Ruiz-Ruano et al., 2012).

2 GAMES

2.1 Design Process

A first set of requirements for the games was defined

after a comprehensive review of the key elements

that promote the rehabilitation process and the role

of gaming in robot-aided rehabilitation therapy was

done (Perry et al., 2011). A more detailed list of

requirements and games to be developed was drawn

out integrating this preliminary information with a

series of interviews and focus groups with clinicians.

Then, an iterative approach was used for the

games' design and implementation, meaning that the

games were tested by patients and therapists all

along the implementation process.

Throughout the game design and implementation

process, game design principles and ergonomic and

user interface design standards were closely

observed. Design criteria considered included

aspects such as clarity of instructions and feedback,

optimization of the level of challenge in order to

keep patient motivation, consistency between games,

suitability toward visual or cognitive impairments,

and robustness, i.e., tolerance to user error. A

detailed description of the methods used for the

design was already presented in (Rodriguez-de-

Pablo et al., 2012).

A first version of the games was implemented

and tested in a 12-week clinical pilot test with 9

patients. This usability testing and its results were

further described in (Perry et al., 2013). Then, a

second version of the games was developed in order

to first, include the usability feedback obtained in

previous testing and second, improve the assessment

methodology in order to be able to properly evaluate

the effectiveness of the system and compare the

results of that assessment with the standard scales

used at the clinic. Finally, a shorter usability testing

of the new version of the software was carried out in

the Rehabilitation Service at ‘Hospital Universitario

de Álava- Sede Txagorritxu’ (Spain). The feedback

extracted from that testing was used for the design of

the final version of the software, which is currently

being used in an ongoing clinical study.

2.2 Requirements

In this section, the requirements for the redesign of

the games are presented. A detailed list of the initial

requirements of the games was already presented in

(Rodriguez-de-Pablo et al., 2012).

The main goal of the redesign was to improve

the games in order to be able to properly evaluate

the effectiveness of the system and compare the

assessment results with the standard scales used at

the clinic. For that, first of all, levels in the

assessment games needed to be removed in order to

facilitate intra- and inter-patient comparison. They

had to be kept, however, for the training games in

order to adapt the challenge, motivate the patient and

avoid frustration. Also, it was necessary to make

some of the games simpler and faster; for example

different versions of the Control of Force

assessment game were developed and later

determined to be too time consuming and

cognitively taxing for a rapid assessment. Others

needed a redesign in order to avoid undesired

movements; therefore, the new version of the

Control of Movement game asks only for reach

movements, and the new Range of Movement game

measures only when the user stops in order to avoid

counting ballistic movements. It was observed to be

SeriousGamesforAssessmentandTraininginPost-strokeRoboticUpper-limbTelerehabilitation

127

of major importance the fact that the score was

perfectly aligned with the metric evaluated, as that is

what the patients focused on when trying to

improve. In that way, the reaching movement

requirements have been removed from the new

Control of Force in order to focus on the control of

the lifting force and its smoothness evaluation. And

last, some improvements needed to be done in the

data collection to address some issues detected in the

post-processing work of data from the previous

games.

In addition, most of the feedback obtained in the

usability testing was taken into account for the

redesign. Some examples of the changes included

are:

 The improvement of the selection/de-selection

strategies, with an optimization of the timing in

the waiting strategy and a refinement of the

feedback in the lifting force strategy;

 The improvement of feedback, e.g., in the

Point to Point game when a piece is inserted;

this was not always properly understood by

patients and they kept trying to insert it;

 The clarification of some instructions, e.g.,

providing the hint in the Words game or

allocating a concrete space where the Jigsaw

Puzzle has to be assembled;

 The improvement of the visualization, e.g.,

making the back image of the cards in the

Memory game unmistakable with the content

itself;

 The empirical adjustment of the time-outs and

level parameters.

2.3 Games’ Description

2.3.1 General Features

There exist two kinds of games. The assessment

games are short tasks (1-2 min.) designed to provide

the therapist with an objective evaluation of the

different parameters that define the arm movement

and to adjust correspondingly the training games.

The training games are longer, more complex and

entertaining tasks, which can also integrate cognitive

components and whose objective is to motivate the

patient to train harder and longer. In both of them,

reach extension movements are encouraged, which

require simultaneous abduction/ adduction of the

shoulder and extension of the elbow.

In all the games (Fig. 1) the level is composed by

two aspects: the Movement level (ROM) and the

Force Level (ROF). The ROM, the range of

movement that each game requires, is determined by

the performance of the patient in the first assessment

game, Range of Movement. For accessibility reasons,

the visualization of the games is always the same,

that is, the objects keep their sizes and their

positions; however, the area of the mat where the

ArmAssist has to be moved varies according to the

range of motion of the patient. This avoids some

visualization problems when the range is too small

and has proven to be well understood by patients.

The ROF is the range of vertical force required by

each game, i.e. the percentage of the arm weight that

the patient must lift. This is determined by the

performance of the patient at the second assessment

game, Range of Force.

Figure 1: Influence of the result obtained at ROM and

ROF in the other games.

Furthermore, in the training games a task level

component is included. This is avoided in the

assessment games in order to facilitate the intra- and

inter-patient comparison of the evaluation results.

Each game presents five levels and is scored based

on a combination of evaluated features. The game

levels are automatically adapted by the platform

based on performance. The adaptation method

adopted is the following: a game score of 100

percent or two consecutive scores of at least 80

percent prompt a level increase. Still, if considered

inadequate, the level can be modified by the

therapist from within the platform at any time.

Each time that a game starts, the user will be

asked to move the device slowly over the mat in

order to calibrate the device position. This will not

take more than two seconds. Also, before playing

any game, the user is asked to calibrate the weight of

his arm in a relaxed state in order to calibrate the

vertical force component of the device. This second

calibration needs to be done only once; however, it

can be repeated when desired for changes in the

position of the table or chair that may affect the

weight of the arm.

There are two strategies for selection/ de-

selection of objects in the training games.

ArmAssist’s natural strategy is the use of the vertical

force, that is, lift the arm to pick up a piece, move

and relax the arm to leave it. However, for some

NEUROTECHNIX2014-InternationalCongressonNeurotechnology,ElectronicsandInformatics

128

patients this strategy has proven to be more difficult

to understand or to control in the early stages of

rehabilitation. Therefore, another strategy is

provided which does not require the use of the

vertical force, requiring instead a brief pause over

the piece to select it. The piece will be deselected

when the patient pauses after having moved it.

Moreover, in order to avoid unwanted selections, to

reactivate the selection of a part, the user must bring

the cursor out of the piece and place it back over

briefly again. Proper timing is key for this strategy

to work properly. When the therapist deems

appropriate, he will enable the use of force in the

patient’s profile. In both cases, the strategy that

should be used to pick or drop an object is indicated

by a message and icon at the top of the screen (Fig.

2). In order to facilitate the comprehension of the

vertical force selection strategy, the feedback force

is also displayed, indicated in green when the

required threshold lifting force is being performed,

and in red when not.

Figure 2: Indications for the two selection strategies.

Strict overall times and intermediate countdowns

in the case of inactivity are employed in all the

assessment games to ensure that assessments are

carried out efficiently. The intermediate countdowns

ensure, for example, that the cursor is placed at the

required area or that the achieved value is improved

in a determined amount of time. In the training

games, the total time can be assigned by the

therapist. In the games in which the time is

associated with the task level, the game will be

launched successively as many times as necessary to

complete the total time assigned. A ‘direct exit’ has

been enabled in case the user or therapist wants to

finish a game before the assigned training time; this

can be done pressing the center bar of the device

with both hands. However, when exited in this way,

no data is stored, nor the level updated, etc. thus, it

should not be used as a normal course of action.

General performance indicators are stored for all

the games in each session. During the assessment

games, full force and trajectory information is also

stored in order to allow a detailed post-processing

analysis. The data collection has been improved in

order to solve some problems found in previous data

analysis. The sampling frequency has been increased

from 10 to 100Hz and descriptive data of every state

of the game is collected. In addition, data is split in

blocks, each block corresponding to analog groups

of movements, e.g., extension and flexion; for each

of them the target position and force, events, times

and performance are stored.

2.3.2 Assessment Games

The Force Calibration game (Fig. 3(a)) serves to

weigh the patient's arm in a relaxed position and

thereby calibrate the vertical force characteristic of

the device. Therefore, it should be performed before

any other game. Moreover, without any limitation of

time, it helps the patient to have an initial contact

with the vertical force characteristic of the device,

which generally causes some confusion for the

patients at first. In this way the patient can try as

much as needed to use this feature and see the

feedback result of his actions on the screen. When

the patient understands this concept and is prepared,

correct posture of the torso and the relaxation of the

arm must be checked and then the calibration can be

started. This arm weight calibration needs to be done

only once; however, it can be repeated at any time.

The Range of Movement game (Fig. 3(b))

evaluates the range of extension movements from a

central position towards the different directions of

the transverse plane. Being the one that defines the

ROM required by the rest of the games, this should

always be the first game to be played; or just after

the Force Calibration game in the case that this later

was never played before. In the game, different

sectors have to be deleted with a reach extension

movement of the arm from the central ball to the

furthest point reached in that direction as indicated

by the white arrow. The patient needs to stop at the

furthest point reached in order for the result for that

sector to be saved, thereby avoiding the acceptance

of ballistic movements. Then he must return to the

centre ball and start again with another sector. In

order to improve the user feedback, there are two

ways in which sectors are visually erased (Fig. 3(b)):

instantaneously as the cursor passes (in light green),

which gives real-time feedback to the user; and once

they stop (in white), when the score is really saved,

and that part of the sector deleted permanently. The

patient must ensure that he remains upright during

the game. The game detects if the patient leans over

the table in order to achieve a greater range of

movement; in that case the game is interrupted until

he corrects the posture.

The Range of Force game (Fig. 3(c)) assesses the

SeriousGamesforAssessmentandTraininginPost-strokeRoboticUpper-limbTelerehabilitation

129

arm support/lifting capacity in different positions of

the plane by placing the cursor over a circular target

and lifting the arm. As the arm is unloaded from the

device, the size of the target is increased in

proportion to the lifting force in order to reach the

diameter of a peripheral ring which indicates the

target unloading level. Being the one that defines the

ROF required by the rest of the games, this should

always be the second game to be played, after the

Range of Movement game.

The Control of Movement game (Fig. 3(d))

evaluates the patient's motion control when he has to

follow a path from one point to another and stop in a

very restricted time. For that, the user must first,

move the cursor to the centre ball; second, to the

target that is blinking and stop in the centre as

quickly as possible. Then, go back to the centre and

start the task over moving to the new ball that blinks.

As in the Range of Movement game, the patient

must ensure that he remains upright during the

game; otherwise, the game will be interrupted until

he corrects the posture.

The Control of Force game (Fig. 3 (e)) evaluates

the patient's ability to control the level of vertical

force, that is, the ability to lift the arm. Keeping the

arm in a centred position on the mat, the patient

must lift the arm slightly trying to keep the needle in

the middle of the dial, always in the green area. This

area will become narrower, and therefore will

require a greater control of the arm as the time

passes. The patient should keep this position until

the progress bar completes. The goal value is based

on the ROF evaluated in the Range of Force game.

For consistency with other games, in the case that

the user prefers the feedback on the right side of the

screen, the user must try to keep the green circle

within the two blue circles.

2.3.3 Training Games

These games aim to train in an enjoyable way the

reach extension movements of the arm, generally

having to pick objects and move them to different

positions of the plane. To accomplish this, the user

has to first, place the cursor over the corresponding

object and stop briefly, or lift the arm in the case that

the vertical force component is being used, which

will be indicated; second, move it to the

corresponding position; and third, stop briefly again

or rest the arm in the previously mentioned case.

Each game presents five levels and is scored based

on a combination of evaluated features. Time

available is also adjusted according to the level in

each case.

In the Words game (Fig. 4(a)) the missing letters

in the words have to be completed with the spare

letters at the bottom of the screen. The letter will go

back to its initial position if it is introduced into the

wrong space or is released, that is, if we stop for 1

second after moving the piece or rest the arm. As the

level increases, the number of missing letters to be

filled in each word, the difficulty of the word and the

(a) (b)

Figure 3: Assessment games: (a) Force Calibration (b) Range of Movement (c) Range of Force (d) Control of Movement (e)

Control of Force.

NEUROTECHNIX2014-InternationalCongressonNeurotechnology,ElectronicsandInformatics

130

(a) (b)

(c)

(d)

(e)

Figure 4: Training games: (a) Words (b) Jigsaw Puzzle (c) Memory (d) Point to Point (e) Solitaire.

difficulty of the task also increase. Task difficulty,

for example, increases by having to fill in vowels or

consonants and the accuracy needed for a letter to be

considered selected or introduced. In the Point to

Point game (Fig. 4(d)) each object has to be placed

in its box. The levels are defined by the number of

elements to be introduced and the accuracy needed

for an object to be considered selected or introduced.

For the Memory game (Fig. 4(c)) the user has to

discover pairs of matching cards, remembering the

ones previously discovered by the user or the

opponent (the PC). The number of pairs and the

intelligence of the PC define the levels. The users

can also train by putting together a Jigsaw Puzzle

(Fig. 4 (b)) whose number and size of pieces

increase with the levels. As a fifth training game, the

user can play the typical Solitaire game (Fig. 4(e))

with different numbers of cards dealt and times to

complete the task.

3 TESTING

The games are currently being used informally at the

Rehabilitation Service at ‘Hospital Universitario de

Álava- Sede Txagorritxu’ (Spain) in order to get

further usability feedback.

Furthermore, a clinical study is being conducted in

the Clinic for Rehabilitation ‘Dr Miroslav Zotović’

in Belgrade, Serbia. The main purpose of this study

is to determine the correlation between ArmAssist

assessment results and standard clinical assessment

scales. Patients included in this study are 40 to 70-

year-old stroke survivors in the sub-acute phase with

unilateral paresis, able to understand simple

instructions, and who have some voluntary

movements in elbow and shoulder joints. The

program lasts four weeks during which patients have

at least one hour of conventional therapy and at least

thirty minutes of additional therapy per day, five

days per week. The additional therapy differs

depending on whether the patient was randomly

assigned to the ArmAssist group, where they

exercise within the TeleReha platform, or the

Control group where they receive work therapy,

designed to match the training received by the

experimental group. Each group comprises 15

patients.

All patients will be assessed at the beginning and

the end of the program. The Fugl-Meyer Assessment

(Sanford et al., 1993), the Action Research Arm Test

(Van der Lee et al., 2001), the Barthel index (Collin

et al., 1988), the Modified Rankin scale (De Haan et

al., 1995), the Beck depression scale (Beck et al.,

1961), the Wolf Motor Function test (Wolf et al.,

2001) and the Modified Drawing Test (Kostić et al.,

2013) are among the measurements used. In

addition, an experimental psychologist interviews

patients to evaluate comfort, pain, fatigue,

enjoyment, benefits, desire to continue, and

motivation to exercise.

SeriousGamesforAssessmentandTraininginPost-strokeRoboticUpper-limbTelerehabilitation

131

4 RESULTS

Initial informal feedback from patients and

clinicians at the usability testing has been highly

positive. It shows that the system is easy to use and

integrates well in the clinical setting. The majority of

the issues and feedback items have been addressed

in the redesign and this was very much appreciated

by the users involved.

Patients felt motivated, trained for longer and

their perception of pain and fatigue was lower. The

knowledge of performance clearly motivated them

to continue the training and to improve in every

session. For that reason, it was observed that, in

order to avoid frustration, the difference between

assessment and training games must be clear for the

patients; the significant difference being that the

maximum values for the first ones are defined by

healthy patients and are not adapted to their capacity

as in the case of the second ones.

Therapists considered the system useful and

appreciated the fact that it allows them to treat

several patients at the same time.

(a)

(b)

Figure 5: Full position data visualization for (a) Range of Movement game (b) Control of Movement game.

Figure 6: Data block analysis for the Control of Movement game.

NEUROTECHNIX2014-InternationalCongressonNeurotechnology,ElectronicsandInformatics

132

Still, the number of training games is too low to

keep the user motivated after the first weeks. For

that reason, a special application is currently being

developed in order to be able to use the ArmAssist

as a normal mouse and, in that way, be able to surf

the Internet and play any other games available.

However, standard games in the Internet usually

need an advanced use of the cursor; therefore,

precision of the cursor positioning must be improved

for the application to work properly.

Preliminary analysis of the first data obtained

from the ongoing clinical study shows that the

quality of the data has been improved, which will

facilitate post-processing and conclusions extraction.

For example, a one-shot analysis of the range and

control of motion of the patient in the different

directions of the plane can be performed, shown in

(Fig. 5). Also, detailed analysis of each trial, e.g.

extension plus flexion movement, can be carried out,

providing very visual and useful information of the

movement. As an example, the detailed analysis of

a block of data of the Control of Movement game is

presented in (Fig. 6), were smoothness of the

movement can be clearly observed.

However, further data is needed in order to extract

any trends and conclusions regarding the validity of

the assessment and the effectiveness of the training

games. This data will be further obtained in the

ongoing testing.

5 CONCLUSIONS

A new version of assessment and training games has

been developed for at-home post-stroke arm

rehabilitation. The aim was to include feedback

gathered in previous usability testing and to improve

the assessment methodology in order to be able in

the future, not only to measure the effectiveness of

the system, but also to compare the assessment

results with the standard assessment scales used in

the clinic up to now. Initial feedback from patients

and clinicians of the new version of the games has

been highly positive. Preliminary analysis of data

from the currently ongoing testing shows that the

quality of the data has been improved; this will

facilitate post-processing and conclusions extraction

regarding effectiveness and assessment validity.

Future work comprises the analysis of data obtained

from the ongoing clinical study, the corresponding

redesign of games according to the results obtained,

if necessary, and the development of new training

games.

ACKNOWLEDGEMENTS

The authors would like to thank the Rehabilitation

Service at the ‘Hospital Universitario de Álava-

Sede Txagorritxu’ for their contributions to the

design of the games and J. Arcas, H. Zabaleta and D.

Valencia for their contributions to the whole system.

REFERENCES

Sluijs, E.M., Kok, G.J., Zee, J., 1993. Correlates of

exercise compliance in physical therapy, In Physical

Therapy, 73 (11), pp. 771–782.

Kwakkel, G.,Wagenaar, R.C., Koelman, T.W., Lankhorst

G.J., Koetsier, J.C., 1997. Effects of intensity of

rehabilitation after stroke, a research synthesis, In

Stroke 28.8 (1997): 1550-1556

Peerenboom, P. B. G., Spek, J., Zekveld, I., Cools, H. J.

M., van Balen, R., & Hoogenboom, M. J., 2008.

Revalidatie in de AWBZ, omvang, aard en intensiteit.

Etc Tangram. In LUMC/PHEG Verpleeghuisgenees-

kunde, 31. [Online]: http://www.vumc.nl/afdelingen-

themas/69713/27797/tangram_RapportRevalidatieD1.

pdf, accessed April 2013.

Hesse, S., Werner, C., Pohl, M., Rueckriem, S., Mehrholz,

J., Lingnau, M.L., 2005. Computerized Arm Training

Improves the Motor Control of the Severely Affected

Arm After Stroke A Single-Blinded Randomized Trial

in Two Centers. In Stroke, 36:1960-1966.

Robertson, J. V. G., Roby-Brami, A., 2010. Augmented

feedback, virtual reality and robotics for designing

new rehabilitation methods. In Rethinking Physical

and RehabilitationMedicine, J.-P.Didier and E.

Bigand, Eds., pp. 223–245. Springer.

Zabaleta, H., Valencia, D., Perry, J., Veneman, J., Keller,

T., 2011. Absolute position calculation for a desktop

mobile rehabilitation robot based on three optical

mouse sensors. In Engineering in Medicine and

Biology Society, EMBC, 2011 Annual International

Conference of the IEEE (pp. 2069-2072). IEEE.

Perry, J. C., Zabaleta, H., Belloso, A., Rodriguez-de-

Pablo, C., Cavallaro, F. I., Keller, T., 2012.

ArmAssist: Development of a functional prototype for

at-home telerehabilitation of post-stroke arm

impairment. In Biomedical Robotics and Biomecha-

tronics (BioRob), 2012 4th IEEE RAS & EMBS

International Conference on (pp. 1561-1566). IEEE.

Arcas Ruiz-Ruano, J., Perry, J.C., Rodriguez-de-Pablo, C.,

Keller, T., 2012. TeleREHA: online/offline web

platform for telerehabilitation of post-stroke arm

impairment. In Intl. Conf. on Neurorehabilitation,

ICNR 2012, November 14-16, Toledo, Spain.

Perry, J. C., Andureu, J., Cavallaro, F. I., Veneman, J.,

Carmien, S., Keller, T., 2011. Effective game use in

neurorehabilitation: user-centered perspectives. In

Handbook of Research on Improving Learning and

SeriousGamesforAssessmentandTraininginPost-strokeRoboticUpper-limbTelerehabilitation

133

Motivation through Educational Games, IGI Global,

Ch. 32, pp:683-725.

Rodríguez-de-Pablo, C., Perry, J. C., Cavallaro, F. I.,

Zabaleta, H., Keller, T., 2012. Development of

computer games for assessment and training in post-

stroke arm telerehabilitation. In Engineering in

Medicine and Biology Society (EMBC), 2012 Annual

International Conference of the IEEE (pp. 4571-

4574). IEEE.

Perry, J.C., Rodriguez-de-Pablo, C., Cavallaro, F. I.,

Belloso, A., Keller, T., 2013. Assessment and

training in home-based telerehabilitation of arm

mobility impairment. In Journal of Accessibility and

Design for All (CC) JACCES, 2013, Vol 3, No 2: 117-

135. ISSN: 2013-7087

Sanford, J., Moreland, J., Swanson, L. R., Stratford, P. W.,

and Gowland, C., 1993. Reliability of the Fugl-Meyer

assessment for testing motor performance in patients

following stroke. In Physical therapy, vol. 73, no. 7,

pp. 447-454

Van der Lee, J. H., De Groot, V., Beckerman, H.,

Wagenaar, R. C., Lankhorst, G. J., and Bouter, L. M.,

2001. The intra-and interrater reliability of the action

research arm test: a practical test of upper extremity

function in patients with stroke. In Archives of

physical medicine and rehabilitation, vol. 82, no. 1,

pp. 14-19

Collin, C., Wade, D. T., et al., 1988. The Barthel ADL

Index: a reliability study. In International Disability

Studies 10(2): 61-63

De Haan, R., Limburg, M., et al., 1995. The clinical

meaning of Rankin 'handicap' grades after stroke. In

Stroke 26: 2027-2030

Beck, A.T., Ward, C. H., Mendelson, M., Mock, J.,

Erbaugh, J., 1961. An inventory for measuring

depression. In Archives of General Psychiatry, 4, 561-

571.

Wolf, S. L., Catlin, P. A., et al., 2001. Assessing Wolf

motor function test as outcome measure for research in

patients after stroke. In Stroke 32: 1635-1639.

Kostić, M. D., and Popović, M. D., 2013. The modified

drawing test for assessment of arm movement quality.

In Journal of Automatic Control, 21(1), 49-53.

NEUROTECHNIX2014-InternationalCongressonNeurotechnology,ElectronicsandInformatics

134