Error Augmented Robotic Rehabilitation of the Upper Limb

A Review

Aris C. Alexoulis-Chrysovergis

, Andrew Weightman

Emma Hodson-Tole

and Frerik J. A. Deconinck

School of Engineering, Manchester Metropolitan University, Chester Street, Manchester, U.K.

School of Healthcare Science, Manchester Metropolitan University, Chester Street, Manchester, U.K.

Keywords: Robotic, Rehabilitation, Upper, Limb, Haptic, Visual, Feedbacks, Augmented, Enhanced, Review.

Abstract: Objective: To collect and assess the available evidence for the efficacy of error augmentation in upper limb

robotic rehabilitation.

Methods: A systematic literature search up to May 2013 was conducted in one citation index, the Web of

Knowledge, and in two individual databases: PubMed and Scopus, for publications that utilized error

augmented feedback as practice modality in robotic rehabilitation of the upper limb.

Results: The systematic search returned 12 studies that utilized error augmented feedback in trials to

unimpaired and impaired individuals suffering from stroke, multiple sclerosis and primary dystonia. One

additional study utilizing viscous force fields was included as the authors paid special merit to the effects of

the field in directions where the error was amplified. In the studies that met the inclusion criteria two

different types of error augmented feedback was used that is, haptic and visual feedback which were used

either separately as rehabilitation modalities or in conjunction with each other. All studies but one report

positive outcome regardless of the type(s) of feedback utilized.

Conclusions: Error augmentation in upper limb robotic rehabilitation is a relatively new area of study,

counting almost nine years after the first relevant publication and rather understudied. Error augmentation in

upper limb robotic rehabilitation should be further researched in more practice-intensive studies and with

larger trial groups. The potential of error augmented upper limb rehabilitation should also be explored with

conditions other than the ones described in this review.

1 INTRODUCTION

Neurological impairments resulting from conditions

such as stroke and cerebral palsy are common. For

example, stroke affects 150 000 people in the UK

each year (2005/2006 S.S.C.A., 2001 ) and cerebral

palsy is the commonest cause of childhood disability

in Europe (Reinkensmeyer et al., 2004; Huang and

Krakauer, 2009; Weightman et al., 2011).

Neurological impairment, resulting from these

pathologies, often influences upper limb function

causing weakness, spasticity and loss of selective

muscle activation. These in turn, cause difficulties

with voluntary movements and affect the ability to

reach, grasp transport

and manipulate objects.

Movements in affected individuals are therefore

characterized by increased duration, reduced peak

velocity, increased variability and fewer straight

hand trajectories (Wu et al., 2000).

Improvement in upper limb function can lead to

better performance in activities of daily living,

increased social integration and can thus produce a

better quality of life (Maher et al., 2007; Imms,

2008). Exercise of an impaired limb is known to

improve function (Kluzik et al., 1990), with better

performance observed with increased time and

amount of practice devoted to learning a particular

task (French et al., 2007). Traditionally such

exercises are monitored by a trained clinical

therapist. However, researchers have (recently)

begun to investigate the application of robotics as a

potential modality to support such rehabilitation.

The paradigm of upper limb rehabilitation

robotics is a motivating computer environment,

which promotes therapeutic movements of the

impaired limb with a powered interface

implementing a control algorithm to promote

recovery (Prange et al., 2006; Scott and Dukelow,

167

C. Alexoulis-Chrysovergis A., Weightman A., Hodson-Tole E. and J. A. Deconinck F..

Error Augmented Robotic Rehabilitation of the Upper Limb - A Review.

DOI: 10.5220/0004654101670178

In Proceedings of the International Congress on Neurotechnology, Electronics and Informatics (SensoryFusion-2013), pages 167-178

ISBN: 978-989-8565-80-8

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

2011). Such a system can provide patients with

access to rehabilitation protocols, which do not

require direct, time demanding, supervision of a

clinical therapist. As such they can increase access

to therapy with limited additional burden on

healthcare provision. Furthermore, such systems

enable the logging of valuable data regarding user’s

activity and performance for the therapist to closely

monitor adaptation and provide feedback on

progress to the user. Rehabilitation robotic therapy

has demonstrated statistically significant benefits in

improving upper limb function, with kinematic

analysis revealing benefits in movement time, path

and smoothness of reach (Fasoli et al., 2008; Huang

and Krakauer, 2009; Fluet et al., 2010, Weightman

et al., 2011, Norouzi-Gheidari et al., 2012).

Currently, three types of rehabilitation robot

have been described: i) end point attachment; ii)

multiple point attachment and iii) exoskeletons

(Reinkensmeyer et al., 2004; Jackson et al., 2007,

Scott and Dukelow, 2011; Weightman et al., 2011).

End point attachment robots are limited in that they

can only promote desirable trajectories (spatial and

temporal characteristics) of the hand and cannot

control the corresponding position of the elbow and

shoulder. However, they are likely to be more cost

effective than multiple point of attachment robots

and exoskeletons. Multiple point of attachment

robots and exoskeletons can control the full

kinematics of the arm (end point, elbow, shoulder)

but are usually significantly larger and more

expensive and as such are less likely to be utilized

outside the clinical environment; for example in

home rehabilitation applications where size and

price can be significant consideration factors for

employing such technology.

The control strategy i.e. the manner of

interaction between user and the powered

joysticks/robotics, implemented is critical for the

promotion of improved upper limb function

(Reinkensmeyer et al., 2004; Marchal-Crespo and

Reinkensmeyer, 2009) and different control

strategies have been utilised in the current literature.

Marchal-Crespo et al. (Marchal-Crespo and

Reinkensmeyer, 2009) suggested they can broadly

be divided into three groups. Firstly, assisting

control strategies help to move the impaired upper

limb in aiming type movements, this is similar to the

“active assist” type exercises utilised by therapists

(Marchal-Crespo and Reinkensmeyer, 2009;

Weightman et al., 2011). Secondly, challenge based

control strategies can make movements more

difficult, for example augmenting error between

actual and desired trajectory or promoting increased

effort (resistance training) from the participant.

Thirdly, haptic simulation strategies involve the user

practising activities of daily living within a virtual

haptic environment (Montagner et al., 2007,

Marchal-Crespo and Reinkensmeyer, 2009).

Challenge based algorithms such as, error

augmenting, are based on the concept that errors in

performance and hence results of aiming and

prehensile movements of the upper limb influence

motor adaptation (Wolpert et al., 1995; Patton et al.,

2006b). These strategies have been shown to

improve motor function in adults suffering from

stroke (Morris et al., 2004; Patton et al., 2006b).

Moreover, there have been early indications that

error augmented visual feedback can induce motor

learning in able bodied and possibly in impaired

individuals (Wei et al., 2005). In the last twenty

years, substantial work has been done in robotic

rehabilitation. Error augmentation seems to be a

relatively new modality and to our knowledge there

has not been an attempt to gather and collectively

report the findings of such studies. Therefore, the

purpose of this paper is to present a systematic

literature review of research regarding the use of

error augmented feedback in the robotic

rehabilitation of the upper limb and determine its

potential for promoting improved upper limb

function in those who have suffered a neurological

impairment.

2 METHODS

A systematic literature search up to May 2013 was

conducted in one citation index, the Web of

Knowledge, and in two individual databases:

PubMed and Scopus. In order to ensure that the

search would return as many results as possible two

different sets of keywords were used in each

database. No lower end in year was used in any

search. The keywords for the first set were: robot,

rehabilitation, upper, limb, error and the keywords

for the second set were: rehabilitation, upper, limb,

error. Papers identified in either search were

included for further investigation. To make sure that

significant publications were not missed during the

initial search the references of the retrieved studies

were checked for relevant publications. After

identifying and excluding duplicates, all abstracts

were reviewed and when necessary a full review of

the manuscript was undertaken.

The inclusion criteria for the review were studies

i) with upper limb robotic rehabilitation; ii) utilizing

error augmentation as a training modality, including

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

168

all types of distorted feedback (haptic or visual); iii)

where trials on humans (impaired or able bodied)

were performed. Only papers reporting new

experimental data were included, however it should

be noted that the systematic search returned two

review papers referring to error augmented robotic

therapy in upper limb rehabilitation (Johnson, 2006;

Reinkensmeyer, 2009).

3 RESULTS

Out of 60 papers originally identified 12 met the

inclusion criteria. An exception was made with

study (Patton et al., 2006b) which didn’t meet the set

criteria for the review, because viscous force fields

were used in the study not an error augmentation.

However, the authors discussed the effects of the

treatment in the directions of the movement where

error was amplified. As such the study was

considered suitable for the purposes of this review

and therefore a total of 13 papers were reviewed.

An overview on the contents of the selected

papers can be found in Table 1.

3.1 Overview of Selected Studies

Error augmented robotic therapy for the

rehabilitation of the upper limb is a relatively new

rehabilitation modality, as the first relevant study

was undertaken in 2004 (Patton and Mussa-Ivaldi,

2004). Since then publications regarding this subject

are published with an average rate of 1.5

publications per year (Figure 1).

3.1.1 Clinical Characteristics

of the Participants

All included studies employed human participants

for clinical trials. The conditions that were addressed

varied significantly, with six studies focusing on

upper limb rehabilitation in stroke patients (Patton et

al., 2006a; Patton et al., 2006b; Cesqui et al., 2008;

Rozario et al., 2009; Abdollahi et al., 2011; Molier

et al., 2011), two studies employing participants

with multiple sclerosis (Squeri et al., 2007b; Vergaro

et al., 2010) and one study (Casellato et al., 2012)

employing error augmented robotic therapy in

children with primary dystonia. Furthermore, four

studies experimented in the effects of error

augmented robotic therapy with the participation of

only able bodied, healthy adults (Patton and Mussa-

Ivaldi, 2004; Matsuoka et al., 2007; Wang et al.,

2010; Shirzad et al., 2012).

Figure 1: Number of publications on error augmented

robotic therapy on the upper limb.

3.1.2 Types of Rehabilitation Robots

Interestingly all but two (Patton et al., 2006b; Molier

et al., 2011), studies used single point of attachment

robotic systems (endpoint). In one study two

endpoint robotic devices were utilized to control the

thumb and the index finger of the participants in

pinching movements (Matsuoka et al., 2007) while

in another study a multiple point of attachment

system (exoskeleton) was used for the control of arm

movements (Molier et al., 2011).

3.1.3 Types of Error augmented Feedback

Two different types of feedback, where error was

augmented, were identified among the selected

studies. The approaches can be categorized as: a)

Error augmented haptic feedback, where forces

perturbed upper limb movement when a certain level

of error away from the desired trajectory was

reached (Patton and Mussa-Ivaldi, 2004; Patton et

al., 2006a; Patton et al., 2006b; Squeri et al., 2007b,

Cesqui et al., 2008; Vergaro et al., 2010; Abdollahi

et al., 2011; Molier et al., 2011; Casellato et al.,

2012); b) Error augmented visual feedback, where

the visual output of the system was distorted by a

factor (ε) in order for the actual distance between the

arm and the target, to differ from the one perceived

by the user (Matsuoka et al., 2007; Wang et al.,

2010); c) A combination of a and b where error in

visual and haptic feedback was augmented (Rozario

et al., 2009; Shirzad et al., 2012).

3.2 Intervention Modalities

The main concept of the intervention behind all the

reviewed studies was that a user was positioned in

front of a computer screen while a robotic

manipulandum was attached to/held by the

participant’s upper limb. A target would be

displayed while visual feedback about the current

position of the arm was provided to the user. The

user was asked to perform movements towards

2000 2002 2004 2006 2008 2010 2012

Yearpublished

Numberofpublications

ErrorAugmentedRoboticRehabilitationoftheUpperLimb-AReview

169

predefined targets while the system responded to the

users’ movement by augmenting any error.

In some of the reviewed studies (Patton and

Mussa-Ivaldi, 2004; Squeri et al., 2007b; Cesqui et

al., 2008; Vergaro et al., 2010; Casellato et al., 2012)

haptic error augmenting algorithms were compared

against other types of haptic algorithms namely,

error reducing haptic algorithms. Error reducing

algorithms are adaptive assistive algorithms which

apply forces towards the optimal trajectory when a

threshold of error is reached. In the aforementioned

studies the two different types of haptic algorithms

were either administered to different trial groups or

in the same group but in different stages of the trial

in order for a comparison between the two training

modalities to be feasible. There was one study

(Molier et al., 2011) where restraining forces only

occurred when a certain amount of error was

reached in order to provide position feedback to the

user. In this case the forces were turned off when the

user didn’t exceed a predefined error threshold.

There was great variance in the number of

sessions and the total exercise time the participants

undertook, among the studies. In several cases the

total intervention time was administered in one

session (Johnson, 2006; Patton et al., 2006b;

Matsuoka et al., 2007; Casellato et al., 2012; Shirzad

et al., 2012) while in others the number of sessions

varied from a minimum of 2 sessions (Wang et al.,

2010; Molier et al., 2011) to a maximum 10 sessions

(Cesqui et al., 2008). Moreover, the total time of

exercise administered varied significantly from as

little as 90 minutes (Shirzad et al., 2012) to as much

as 20hours (Cesqui et al., 2008). Additionally, some

studies induced a washout component in the practice

regime either by including a washout cycle in the

practice session where the perturbative forces were

gradually removed (Patton and Mussa-Ivaldi, 2004;

Patton et al., 2006a; Casellato et al., 2012), or by

setting a washout period between trials where no

practice was undertaken (Cesqui et al., 2008;

Rozario et al., 2009, Wang et al., 2010).

Table 2 provides an overview on the practice

schemes administered in the reviewed studies.

3.3 Outcome Measures

The most common clinical measures among the

studies that were used to evaluate outcome on stroke

patients were the Fugl-Meyer scale, the Modified

Ashworth scale for spasticity and the Box and Block

test. Other clinical measures used can be found in

Table 3. In the above-mentioned studies kinematic

data were collected namely, error that is to say the

deviation between the actual and desired trajectory,

jerk index (Squeri et al., 2007a), Jerk (Teulings’)

index (Teulings et al., 1997), and strength (Patton et

al., 2006a; Patton et al., 2006b; Cesqui et al., 2008;

Rozario et al., 2009; Abdollahi et al., 2011; Molier

et al., 2011).

Both studies that performed trials in patients with

multiple sclerosis (Squeri et al., 2007a; Vergaro et

al., 2010) evaluated performance with clinical

measures such as Expanded Disability Status Scale

(EDSS), Ataxia and Tremor scale, Nine Hole Peg

Test, Visual Analogue Scale (VAS) and the Tremor

and Activity of Daily Life (TADL) questionnaire as

well as kinematic such as, lateral deviation (root

mean square value) from the nominal path,

movement duration (seconds), symmetry (ration

between acceleration and deceleration phases) and

smoothness.

Likewise, studies that employed only able-

bodied participants used only kinematic measures

like error (distance between actual and desired

directory, lateral deviation etc.), mean lag (Matsuoka

et al., 2007) and times needed assistance. The times

needed assistance measure was used in one study

(Wang et al., 2010) where visual error augmentation

was utilized. In this studies training scheme the

system would assist movement only when a

threshold in error was reached. In one study

(Shirzad et al., 2012) a

Self-Assessment Manikin

(SAM) affect questionnaire was administered.

Finally, in the study where trials on children with

primary dystonia were performed (Casellato et al.,

2012) only clinical measures were used that is to

say, Burke-Fahn-Marsden Dystonia Rating Scale

(BFMDRS).

A more detailed overview of the outcome

measures used according to condition can be found

in Table 3.

3.4 Impact on Motor Learning

and Upper Limb Function

Out of the 13 reviewed studies 12 report positive

impact on upper limb function, five of which report

conclusive results (Squeri et al., 2007a; Cesqui et al.,

2008; Vergaro et al., 2010; Abdollahi et al., 2011;

Molier et al., 2011) and seven (Patton and Mussa-

Ivaldi, 2004; Patton et al., 2006a; Patton et al.,

2006b; Matsuoka et al., 2007; Rozario et al., 2009;

Wang et al., 2010; Casellato et al., 2012) report

inconclusive but positive results (Table 1).

Inconclusive results were considered as, those

results where the experiment did not have significant

statistical power for definitive conclusions to be

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

170

drawn and the results where the authors couldn’t

definitively link improvements to the error

augmented treatment.

In one study (Shirzad et al., 2012) the authors

concluded that there was no significant impact of the

intervention on motor learning but when the

different training modes employed in the study were

compared, motor learning was improved only when

haptic error augmentation was combined with visual

error augmentation. In the study where a viscous

force field were used (Patton et al., 2006b), the

authors concluded that significant positive effects

were only encountered in the directions where error

was amplified.

4 DISCUSSION

In this review 13 studies were qualitatively analysed

regarding the effects of error augmented feedback on

robotic rehabilitation of the upper limb. The

reviewed studies employed error augmented therapy

either in the form of haptic or visual feedback or a

combination of the two. Trials were conducted on

healthy participants or on adult participants suffering

from the effects of stroke or multiple sclerosis or

children with primary dystonia.

The first identified study utilizing error

augmentation in the robotic rehabilitation of the

upper limb was published in 2004 (Patton and

Mussa-Ivaldi, 2004). In the nine years since this first

study by Patton et. al was published we could only

retrieve twelve additional studies regarding error

augmentation in the rehabilitation of the upper limb.

4.1 Clinical Trial Protocols

The design of the trial protocols implemented in the

reviewed studies varied significantly as did the

intervention time and group formation. Five of the

studies (Squeri et al., 2007a; Cesqui et al., 2008;

Rozario et al., 2009; Vergaro et al., 2010, Abdollahi

et al., 2011) employed a crossover protocol where

the same group was exposed to different training

modalities with a two week washout period between

the two. Furthermore, six studies used single session

trials (Patton and Mussa-Ivaldi, 2004; Patton et al.,

2006a; Patton et al., 2006b; Matsuoka et al., 2007;

Casellato et al., 2012; Shirzad et al., 2012) with the

total practice time spanning from as little as 22 min

(Patton and Mussa-Ivaldi, 2004) to as much as 96

min (Patton et al., 2006a). Interestingly, only one

study utilized a randomized control clinical trial

(RCT) protocol (Patton et al., 2006b).

Although, the reviewed studies have presented

positive indications of the benefits of the error

augmented robotic therapy to the rehabilitation of

the upper limb, many of the authors argue that more

conclusive outcomes could have been produced if

their studies had larger numbers of participants and

provided more sessions with more practice intensive

protocols. Furthermore, the design of the trial

protocols seems to be a significant factor that

influences the trial outcome. As such trials designed

under a Random Control Trial (RCT) protocol,

where a well-established haptic control algorithm

would be compared to an error augmenting haptic

algorithm, could potentially provide more definitive

results (Dobkin, 2004).

4.2 Error Augmented Feedback

in Upper Limb Rehabilitation

4.2.1 Success of Error Augmented Haptic

Feedback Trials

By studying the results of the trials that utilized

haptic error augmentation one can conclude that the

different conditions are affected differently by this

modality. Stroke patients seem to be more positively

affected by haptic error augmentation exercises as

all studies that performed such experiments on

stroke patients conclude that the group that received

error augmented therapy showed improvement in the

function of the paretic limb. However, such a

statement cannot be definitively made as from the

reviewed studies, the ones that performed trials on

participants suffering from primary dystonia and

multiple sclerosis were significantly less than the

trials on stroke patients. Therefore, the reviewed

studies cannot be compared directly in terms of the

outcome for individuals with different conditions.

More specifically, studies (Cesqui et al., 2008,

Abdollahi et al., 2011; Molier et al., 2011)

conclusively report that the patients who received

error augmented therapy were positively affected. In

study (Rozario et al., 2009) the authors report that

while the kinematic measures indicate improvement,

clinical measures did not provide any measurable

change in the performance and they suggest that

results were probably hindered by the small trial

group and the small number of sessions. The

difference in the outcome of kinematic and the

clinical measures may be due to the fact that

kinematic measures in most cases provide better

responsiveness, that is they are more capable of

accurately detecting changes over time, than clinical

scales (Sivan et al., 2011), hence are more sensitive

ErrorAugmentedRoboticRehabilitationoftheUpperLimb-AReview

171

detectors of change.

Both studies that employed participants with

multiple sclerosis report positive outcomes. Study

(Squeri et al., 2007a) concluded that at the end of the

sessions the participants exhibited faster, smoother

and more symmetric movement. On the other hand,

study (Vergaro et al., 2010) presents similar results

but did not indicate significant differences on the

outcome between error reducing and error

augmenting therapy, with the only exception being a

reduction in a tremor related clinical measure which

occurred only after error augmented therapy. As

such, the improvement presented in both studies

may be due to the fact that the participants

experienced the positive effects of adaptation in a

dynamic environment regardless of the conditions

applied within that environment.

With regards to children suffering from primary

dystonia (Casellato et al., 2012) results indicate

improvement in terms of optimal path control which

as the authors suggest may be due to a refinement in

the existing sensorimotor patterns of the impaired

participants rather than due to motor learning. In the

trials involving participation of able bodied

individuals (Patton and Mussa-Ivaldi, 2004; Shirzad

et al., 2012), the participants could adapt their

movement to the altered environment. However, in

(Patton and Mussa-Ivaldi, 2004) there was no clear

difference of the effects of error augmenting therapy

when compared to those from error reducing

therapy. Finally, subjects in (Shirzad et al., 2012)

showed improved in satisfaction, attentiveness and

dominance when they were introduced to augmented

error conditions despite of the type of feedback

where error was augmented, but didn’t show

improvements on their performance. Both trials

utilized a single session training scheme with

relatively small number of repetitions that may have

not allowed significant changes in motor adaptation

to occur.

4.2.2 Success of Error augmented Visual

Feedback Trials

The studies that used error augmentation in visual

feedback, were significantly less than the ones that

made use of error augmented haptic feedback. It

should be noted that in three out four studies where

visual feedback distortion was used, only able

bodied participants were employed as such it is

difficult to draw conclusions on whether the results

would transfer to the motor impaired.

Nevertheless, only one study (Matsuoka et al.,

2007) reports positive outcome when error was

augmented in the visual feedback as it allowed a

new coordination pattern to transfer to the trials with

no feedback distortion and reduced error. Study

(Rozario et al., 2009) didn’t provide statistically

significant results but indicates that for some of the

participants’, error was reduced when they were

exposed to error augmented training.

4.3 Comparison of Haptic Error

Augmented Therapy to other

Haptic Therapy

In the studies where the performance of haptic error

augmented therapy was compared with haptic error

reducing therapy (Squeri et al., 2007a; Cesqui et al.,

2008; Vergaro et al., 2010) all studies report that

there was no clear indication for the prevalence of

one approach over the other. An interesting outcome

came from the study where viscous force fields were

used (Patton et al., 2006b) as the authors conclude

that most of the improvement in function occurred in

the directions of the field where errors where

amplified.

5 CONCLUSSIONS

Error augmentation in upper limb robotic

rehabilitation is a relatively new area of study,

counting almost nine years since the first relevant

publication, and a rather understudied one. Despite

the small number of publications that have employed

this modality, there are some clear indications about

its potential benefits. The evidence gathered from

this review indicate that stroke patients received the

most benefit from haptic error augmented therapy

but no clear conclusions were drawn whether this

training modality has significant benefits on stroke

patients, over other established modalities such as

error reducing or assistive therapy.

We suggest that large scale randomized control

trials be undertaken in order to explore the prospects

of haptic error augmentation and fully evaluate its

effectiveness on upper limb robotic rehabilitation. In

these trials error augmented therapy should be

compared against other, more established training

schemes. Furthermore, we suggest that the impact of

error augmented therapy should be explored in

conditions that share similar symptoms related to

neuromuscular control to stroke, such as cerebral

palsy. Understanding the neurological mechanisms

targeted by different therapies, in terms of both

learning and motor performance, could provide

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

172

greater insight into their potential efficacy in a range

of different pathologies and is an important

consideration for future studies. Likewise, we would

like to encourage scientists to perform trials on

impaired subjects where error augmentation on

visual feedback will be implemented as the results of

this review indicate that this modality hasn’t been

researched to its full capacity.

Guidelines on trial design and dose

administration for rehabilitation of the upper limb in

conditions such as stroke have been presented in

literature (Dobkin, 2004). To the author’s

knowledge, reviews on the outcome measures for

robotic rehabilitation of the upper limb in conditions

such as cerebral palsy, primary dystonia and

multiple sclerosis have not yet been conducted,

while one review regarding such measures has been

undertaken for the rehabilitation for the upper limb

in stroke patients (Sivan et al., 2011).

This review has identified that there is no

uniform condition-specific trial design or evaluation

protocol as different intervention protocols and

different measures have been used in trials with

participants of the same condition. As a result of this

a comparison between trials and their outcomes is

difficult. Adoption of standard outcome measures

would enable inter-study evaluation and help to

progress this area of research significantly. As

robotic rehabilitation of the upper limb is getting

more and more accepted by the scientific community

as a valid rehabilitation modality, we believe that

uniform condition-specific trial protocol guidelines

should be established, in order to enable researchers

to easily evaluate the outcome of relevant studies in

literature and allow them to compare the outcome of

their studies against that of others.

REFERENCES

2005/2006 S.S.C.A., 2001 Stroke incidence and risk

factors in a population based cohort study. ONS

Health Stats Quart (12).

Abdollahi, F., Rozario, S. V., Kenyon, R. V., Patton, J. L.,

Case, E., Kovic, M. & Listenberger, M., 2011. Arm

control recovery enhanced by error augmentation.

IEEE Int Conf Rehabil Robot, 2011, 5975504.

Casellato, C., Pedrocchi, A., Zorzi, G., Rizzi, G., Ferrigno,

G. & Nardocci, N., 2012. Error-enhancing robot

therapy to induce motor control improvement in

childhood onset primary dystonia. Journal of

Neuroengineering and Rehabilitation, 9.

Cesqui, B., Aliboni, S., Mazzoleni, S., Carrozza, M. C.,

Posteraro, F., Micera, S. & Ieee, 2008. On the Use of

Divergent Force Fields in Robot-Mediated

Neurorehabilitation. 2008 2nd Ieee Ras & Embs

International Conference on Biomedical Robotics and

Biomechatronics. 942-949.

Dobkin, B. H., 2004. Strategies for stroke rehabilitation.

Lancet Neurology, 3, 528-536.

Fasoli, S. E., Fragala-Pinkham, M., Hughes, R., Hogan,

N., Krebs, H. I. & Stein, J., 2008. Upper limb robotic

therapy for children with hemiplegia. Am J Phys Med

Rehabil, 87, 929-36.

Fluet, G. G., Qiu, Q., Kelly, D., Parikh, H. D., Ramirez,

D., Saleh, S. & Adamovich, S. V., 2010. Interfacing a

haptic robotic system with complex virtual

environments to treat impaired upper extremity motor

function in children with cerebral palsy. Dev

Neurorehabil, 13, 335-45.

French, B., Thomas, L. H., Leathley, M. J., Sutton, C. J.,

Mcadam, J., Forster, A., Langhorne, P., Price, C. I. M.,

Walker, A. & Watkins, C. L., 2007. Repetitive task

training for improving functional ability after stroke.

Cochrane Database of Systematic Reviews.

Huang, V. S. & Krakauer, J. W., 2009. Robotic

neurorehabilitation: a computational motor learning

perspective. J Neuroeng Rehabil, 6, 5.

Imms, C., 2008. Children with cerebral palsy participate:

A review of the literature. Disability and

Rehabilitation, 30, 1867-1884.

Jackson, A. E., Holt, R. J., Culmer, P. R., Makower, S. G.,

Levesley, M. C., Richardson, R.C., Cozens, J. A.,

Williams, M. M. & Bhakta, B. B., 2007. Dual robot

system for upper limb rehabilitation after stroke: The

design process. Proceedings of the Institution of

Mechanical Engineers, Part C: Journal of Mechanical

Engineering Science, 221, 845-857.

Johnson, M. J., 2006. Recent trends in robot-assisted

therapy environments to improve real-life functional

performance after stroke. Journal of Neuroengineering

and Rehabilitation, 3.

Kluzik, J., Fetters, L. & Coryell, J., 1990. Quantification

of control: a preliminary study of effects of

neurodevelopmental treatment on reaching in children

with spastic Cerebral Palsy. Physical Therapy, 70, 65-

76.

Maher, C. A., Williams, M. T., Olds, T. & Lane, A. E.,

2007. Physical and sedentary activity in adolescents

with cerebral palsy. Developmental Medicine and

Child Neurology, 49, 450-457.

Marchal-Crespo, L. & Reinkensmeyer, D. J., 2009.

Review of control strategies for robotic movement

training after neurologic injury. Journal of

Neuroengineering and Rehabilitation, 6.

Matsuoka, Y., Brewer, B. R. & Klatzky, R. L., 2007.

Using visual feedback distortion to alter coordinated

pinching patterns for robotic rehabilitation. J

Neuroeng Rehabil, 4, 17.

Molier, B. I., Prange, G. B., Krabben, T., Stienen, A. H.

A., Van Der Kooij, H., Buurke, J. H., Jannink, M. J.

A. & Hermens, H. J., 2011. Effect of position

feedback during task-oriented upper-limb training after

stroke: Five-case pilot study. Journal of Rehabilitation

Research and Development, 48, 1109-1117.

ErrorAugmentedRoboticRehabilitationoftheUpperLimb-AReview

173

Montagner, A., Frisoli, A., Borelli, L., Procopio, C.,

Bergamasco, M., Carboncini, M. C. & Rossi, B., Year.

A pilot clinical study on robotic assisted rehabilitation

in VR with an arm exoskeleton deviceed.^eds. Virtual

Rehabilitation, 2007, 57-64.

Morris, S. L., Dodd, K. J. & Morris, M. E., 2004.

Outcomes of progressive resistance strength training

following stroke: a systematic review. Clin Rehabil,

18, 27-39.

Norouzi-Gheidari, N., Archambault, P. S. & Fung, J.,

2012. Effects of robot-assisted therapy on stroke

rehabilitation in upper limbs: systematic review and

meta-analysis of the literature. Journal of Rehabilitation

Research and Development, 49, 479-96.

Patton, J. L., Kovic, M. & Mussa-Ivaldi, F. A., 2006a.

Custom-designed haptic training for restoring reaching

ability to individuals with poststroke hemiparesis. J

Rehabil Res Dev, 43, 643-56.

Patton, J. L. & Mussa-Ivaldi, F. A., 2004. Robot-assisted

adaptive training: custom force fields for teaching

movement patterns. IEEE Trans Biomed Eng, 51, 636-46.

Patton, J. L., Stoykov, M. E., Kovic, M. & Mussa-Ivaldi,

F. A., 2006b. Evaluation of robotic training forces that

either enhance or reduce error in chronic hemiparetic

stroke survivors. Experimental Brain Research, 168,

368-383.

Prange, G. B., Jannink, M. J. A., Groothuis-Oudshoorn, C.

G. M., Hermens, H.J. & Ijzerman, M. J., 2006.

Systematic review of the effect of robot-aided therapy

on recovery of the hemiparetic arm after stroke.

Journal of Rehabilitation Research and Development,

43, 171-183.

Reinkensmeyer, D. J., 2009. Robotic Assistance for Upper

Extremity Training after Stroke. In A. Gaggioli, E. A.

Keshner, P. L. Weiss & G. Riva (eds.) Advanced

Technologies in Rehabilitation: Empowering

Cognitive, Physical, Social and Communicative Skills

through Virtual Reality, Robots, Wearable Systems

and Brain-Computer Interfaces. 25-39.

Reinkensmeyer, D. J., Emken, J. L. & Cramer, S. C.,

2004. Robotics, motor learning, and neurologic

recovery. Annual Review of Biomedical Engineering,

6, 497-525.

Rozario, S. V., Housman, S., Kovic, M., Kenyon, R. V. &

Patton, J. L., 2009. Therapist-mediated post-stroke

rehabilitation using haptic/graphic error augmentation.

Conf Proc IEEE Eng Med Biol Soc, 2009, 1151-6.

Scott, S. H. & Dukelow, S. P., 2011. Potential of robots as

next-generation technology for clinical assessment of

neurological disorders and upper-limb therapy.

Journal of Rehabilitation Research and Development,

48, 335-353.

Shirzad, N., Van Der Loos, H. F. M. & Ieee, 2012. Error

Amplification to Promote Motor Learning and

Motivation in Therapy Robotics. 2012 Annual

International Conference of the Ieee Engineering in

Medicine and Biology Society. 3907-3910.

Sivan, M., O'connor, R. J., Makower, S., Levesley, M. &

Bhakta, B., 2011. Systematic review of outcome

measures used in the evaluation of robot-assisted

upper limb exercise in stroke. J Rehabil Med, 43, 181-

Squeri, V., Vergaro, E., Brichetto, G., Casadio, M.,

Morasso, P. G., Solaro, C. & Sanguineti, V., Year.

Adaptive robot training in the rehabilitation of

incoordination in Multiple Sclerosis: A pilot

studyed.^eds., Noordwijk, 364-370.

Squeri, V., Vergaro, E., Brichetto, G., Casadio, M.,

Morasso, P. G., Solaro, C., Sanguineti, V. & Ieee,

2007b. Adaptive robot training in the rehabilitation of

incoordination in Multiple Sclerosis: a pilot study.

2007 Ieee 10th International Conference on

Rehabilitation Robotics, Vols 1 and 2. 364-370.

Teulings, H. L., Contrerasvidal, J. L., Stelmach, G. E. &

Adler, C. H., 1997. Parkinsonism reduces coordination

of fingers, wrist, and arm in fine motor control.

Experimental Neurology, 146, 159-170.

Vergaro, E., Squeri, V., Brichetto, G., Casadio, M.,

Morasso, P., Solaro, C. & Sanguineti, V., 2010.

Adaptive robot training for the treatment of

incoordination in Multiple Sclerosis. J Neuroeng

Rehabil, 7, 37.

Wang, F., Barkana, D. E. & Sarkar, N., 2010. Impact of

visual error augmentation when integrated with assist-

as-needed training method in robot-assisted

rehabilitation. IEEE Trans Neural Syst Rehabil Eng,

18, 571-9.

Wei, Y., Bajaj, P., Scheidt, R., Patton, J. & Ieee, 2005.

Visual error augmentation for enhancing motor

learning and rehabilitative relearning. 2005 Ieee 9th

International Conference on Rehabilitation Robotics.

New York: Ieee, 505-510.

Weightman, A., Preston, N., Levesley, M., Holt, R., Mon-

Williams, M., Clarke, M., Cozens, A. J. & Bhakta, B.,

2011. Home based computer-assisted upper limb

exercise for young children with cerebral palsy: a

feasibility study investigating impact on motor control

and functional outcome. Journal of Rehabilitation

Medicine, 43, 359-363.

Wolpert, D. M., Ghahramani, Z. & Jordan, M. I., 1995. An

internal model for sensorimotor integration. Science,

269, 1880-1882.

Wu, C. Y., Trombly, C. A., Lin, K. C. & Tickle-Degnen,

L., 2000. A kinematic study of contextual effects on

reaching performance in persons with and without

stroke: Influences of object availability. Archives of

Physical Medicine and Rehabilitation, 81, 95-101.

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

174

APPENDIX

Table 1: Overview of the contents of the reviewed studies.

Study

Year

published

Type of

Condition

Type of

Robot

Type of

control

Type of

feedback

altered

Number of

participants (N)

(E = experimental

group

C = control group)

Time

post-

stroke

(months)

Amount of practice

Impact on

upper limb

function

(Patton and

Mussa-Ivaldi, 2004)

2004 n/a endpoint

error

reducing

error

enhancing

haptic 8 able bodied n/a

Total movements:

871 (1 session)

Total time practiced:

22 min (1 session for

22 min)

positive

inconclusive

(Patton et al.,

2006a)

2006 stroke endpoint

error

enhancing

haptic

15 impaired

E: 12 C: 9

19-132

Total movements: 742

movements (1 session)

Total time practiced:

180 min (1 session of

80 min)

positive

inconclusive

(Patton et al.,

2006b)

2006 stroke endpoint

viscous

force field

haptic

Total: 31

E:27 impaired

C:4 able bodied

16-173

Total movements: 834

(1 session)

Total time practiced:

57 min (1 session of

57 min)

positive

inconclusive

(Matsuoka et al.,

2007)

2007 n/a

endpoint

(fingers)

n/a visual 51 able bodied n/a

Total movements: 920

movements (1 session)

Total time practiced:

Unknown

positive

inconclusive

(Squeri et al.,

2007a)

2007

Multiple

sclerosis

endpoint

error

enhancing

error

reducing

haptic 4 impaired n/a

Total movements:

3680 (4 sessions)

Total time practiced:

240 min (4 sessions of

60 min)

positive

(Cesqui et al., 2008) 2008 stroke endpoint

active

assistive

error

enhancing

haptic 15 impaired n/a

Total movements:

Unknown

Total time practiced:

600 min (10 sessions

of 60 min)

positive

(Rozario et al.,

2009)

2009 stroke endpoint

error

enhancing

haptic/

visual

5 impaired ≥6 months

Total movements:

Unknown

Total time practiced:

240 min (6 sessions of

40 min)

positive

inconclusive

(Vergaro et al.,

2010)

2010

Multiple

sclerosis

endpoint

error

enhancing

error

reducing

haptic 8 impaired n/a

Total movements:

1992 (4 sessions)

Total time practiced:

240 min (4 sessions of

60 min)

positive

(Wang et al., 2010) 2010 n/a endpoint assistive visual 20 able bodied n/a

Total movements:

50 (2 sessions)

Total time practiced:

Unknown

positive

inconclusive

(Abdollahi et al.,

2011)

2011 stroke endpoint

error

enhancing

haptic 19 impaired 6-259

Total movements:

Unknown

Total time practiced:

360 min (6 sessions of

60 min)

positive

(Molier et al., 2011) 2011 stroke

exoskelet

none

(haptic

feedback

for error)

haptic 5 impaired 20-51

Total movements:

Unknown (2 sessions)

Total time practiced:

540 (18 sessions of 30

min)

positive

(Casellato et al.,

2012)

2012

primary

dystonia

endpoint

null

additive

force

constant

disturbing

force

haptic

Total: 22

11 impaired

11 able bodied

n/a

Total movements:

55 (1 session)

positive

inconclusive

(Shirzad et al.,

2012)

2012 n/a Endpoint

error

enhancing

haptic/

visual

10 able bodied n/a

Total movements:

129 (1 session)

Total time practiced:

90 minutes (1 session

of 90 min)

no effect

ErrorAugmentedRoboticRehabilitationoftheUpperLimb-AReview

175

Table 2: Overview of the contents of the reviewed studies.

Study

Number

sessions

Interven

tion time

in a

session

Total

number of

movements

in a session

Number of repetitions under

feedback distortion (haptic, visual

or both) generation in a session

Time trained in error

augmentation in a

session

Washout

(Patton and

Mussa-

Ivaldi,

2004)

21.95

min

871

a) with intermittent perturbations =

298

b) constant exposure = 330

c) random intermittent removal of

the force field

Total = 748

a) 7.50 min

b) 8.33 min

c) 3.00 min

Total = 18.83 min

75 movements (1.83 min) at the end of

the session

(Patton et

al., 2006a)

95.75

min

742

a) machine learning = 200

b) learning (opposite to the learned

forces) = 222

c) aftereffects catch intermittent

removal of the force field = 80

d) sane as c = 80

e) same as b = 2

Total = 584

a) 25.00 min

b) 30.00 min

c) 20.00 min

d) 10.00 min

e) 0.25 min

Total = 85.25 min

50 movements (3.00 min) at the end of

the session

(Patton et

al., 2006b)

57.00

min

834 n/a n/a

120 movements (8.00 min) at the end of

the session

(Matsuoka

et al.,

2007)

1 n/a 920

a) Index-Thumb-Both (ITB)

distortion =120

b) Thumb-Index-Both distortion

(TIB) =120

c) Thumb only condition mirroring

ITB = 40

d) Thumb only condition mirroring

TIB =40

Total = 320

n/a n/a

(Squeri et

al., 2007a)

60.00

min

498

a) Robot learning = 120

b) Trial = 120

c) Training and catch trials = 168

Total = 408

approx. 49.00 min

45 movements at the end of the session

2 weeks after 4 sessions before protocol

change

(Cesqui et

al., 2008)

60.00

min

n/a n/a 60.00 min

2 weeks after 10 sessions before

protocol change

(Rozario et

al., 2009)

40.00

min

n/a n/a 35.00 min

2 weeks after 6 sessions before protocol

change

(Vergaro et

al., 2010)

60.00

min

498

a) Robot training = 120

b) Subject training = 288

approx. 37.00 min

2 weeks after 4 sessions before protocol

change

(Wang et

al., 2010)

2 n/a 25 Total = 25 n/a n/a

(Abdollahi

et al.,

2011)

60.00

min

n/a n/a 30.00 min

2 weeks after 6 sessions before protocol

change

(Molier et

al., 2011)

30.00

min

n/a n/a 30.00 min n/a

(Casellato

et al.,

2012)

1 n/a 55

a) Null additive force = 15

b) Disturbing force =15

c) Deactivation of additive external

force = 15

Total = 45

n/a n/a

(Shirzad et

al., 2012)

90.00

min

129 Total = 65 approx. 45.00 min

10 movements at the beginning every

training block

(5 cycles/session)

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

176

Table 3: Overview of the practice administered in the reviewed studies.

Study

Type of

control

Type of

Condition

Outcome measures Statistical measurable impact Author conclusions

(Patton

and

Mussa-

Ivaldi,

2004)

error

enhancing

n/a

Kinematic:

Error, speed

• The subjects’ trajectories shifted significantly

towards the desired trajectories

(p< 0.05)

• Clinical improvement for adaptive therapy

• No clear difference between error reducing

and error enhancing therapy

(Patton

et al.,

2006a)

error

enhancing

stroke

Clinical:

Fugl-Meyer, MAS

Kinematics: Error

• All but one of the treatment groups movements

showed beneficial aftereffects

• Average reduction in error of –54%, (p < 0.05)

• Treatment group, FM scores marginally increased

an average of 1.6 (p = 0.06). No such

improvement was seen in the control group

(p > 0.27).

• The stroke group: movements showed

beneficial aftereffects after training (error

decreased) that persisted in all but three

patients

• This persistence was twice as long as for

nondisabled people

(Patton

et al.,

2006b)

viscous

force field

stroke

Clinical: Chedoke

stage of Arm, Elbow

modified Ashworth

Spasticity scale, F-M

• The after-effect was significant but 26% smaller

that the healthy subjects (confidence 95%)

• Significant improvements occurred only when the

training forces magnified the original errors

F(1,13) = 4,29 (p<0.001)

• For movement directions that begin with

significant errors, significant improvement

occurred only when the training forces

magnified the original errors

(Matsuo

ka et al.,

2007)

n/a n/a n/a

• The mean total absolute error for at the first

training block was significantly different for the

last block (p = 0.001)

• No significant results for final performance

change at the end of the trial (p = 0.99)

• Training under visual feedback allowed

new coordination pattern to transfer to no-

feedback trials

• Feedback distortion changed the amount

of error reduced for each finger separately,

and altogether

• Distorting individual fingers separately (or

together) did not affect the overall speed

of learning in movement error reduction.

(Squeri

et al.,

2007a)

error

enhancing

error

reducing

Multiple

sclerosis

Clinical: MMSE,

EDSS, NRS, Scripps',

Aswhorth, Ataxia and

Tremor scale, NHPT,

VAS, TADL

Kinematic: Max

lateral deviation,

movement duration,

symmetry, jerk index

• First to last session: highly significant

(p=0,000027) decrease in duration and significant

increase (p=0.031) in speed profile symmetry

• Error enhancing vs error reducing: first-last

sessions profile symmetry (p= 0.006) and

trajectory smoothness (p = 0.05) increased in

error enhancing

• Decrease in NHPT score overall F(1,3)=42.133

(p=0.007)

• Analysis of motor performance reveals

that, at the end of a training session,

movements are faster, smoother and have

a more symmetric speed profile

• Smoothness improved over sessions

(Cesqui

et al.,

2008)

active

assistive

error

enhancing

stroke

Clinical: MSS, MAS,

ROM, Mc-Master

Stroke Assessment

Kinematic:

Smoothness, accuracy,

path length ratio,

movements direction

variability

• Robotic-aided therapy led to a significant

reduction in impairment of the hemiparetic limbs,

as shown by the evolution of the MSS and MAS

throughout the therapy (no p-value was provided)

• Group I: final metric indexes were no different

(F=1.61, p = 0.194)

• Group II: final metric indexes were significantly

different (F=9,46, p = 0.006)

• Post-stroke patients were able to contrast

the perturbation field, i.e., they could reach

the target, and perform the exercise (varies

dependent on the severity of the

impairment)

• Improvements were higher depending on

the admission upper limb severity level

(Rozari

o et al.,

2009)

error

enhancing

stroke

Clinical: FM, WFMT,

FAS, box and blocks

Kinematic: Range of

motion error

• ROM assessment exhibited a floor effect, where

subjects that initially demonstrated fairly low

reaching errors did not significantly improve their

accuracy in reaching to target

• ROM test did reduce for three subjects in the

error augmented and control treatment groups.

• Error for two of the three subjects was

significantly decreased following error

augmentation treatment compared to control

treatment.

• Subjects provided with error augmentation during

the first phase of treatment produced greater

performance improvements

• No significant improvement, deterioration, or

notable trends were demonstrated with the

clinical measures.

(no statistical analysis was performed

• the two week treatment blocks might not

be sufficient to provide any measurable

change clinically, small number of

subjects (five) is not sufficient to draw any

definitive conclusion

• control treatment, while not providing

error-augmentation, still improved

functionality

• MS subjects adapt to unfamiliar dynamic

environments

• Improvements from error measures don’t

correlate with clinical measures

ErrorAugmentedRoboticRehabilitationoftheUpperLimb-AReview

177

Table 4: Overview of the practice administered in the reviewed studies (cont.).

Study

Type of

control

Type of

Condition

Outcome measures Statistical measurable impact Author conclusions

(Vergaro

et al.,

2010)

error

enhancing

error

reducing

Multiple

sclerosis

Clinical: EDSS and

Functional Systems

Score, Scripps’

,NRS, Ashworth

scale, Ataxia and

Tremor scales,

NHPT, VAS,TADL

Kinematic: Lateral

deviation, duration,

Symmetry, Jerk

• Significant effect of period (F(1,6) = 16.004;

p = 0.00283).

• NHPT change from baseline (T0) to the end of

the treatment (T3), irrespective of the training

mode. NHPT score decreased from 61 ± 14 s to

48 ± 20 s, a 24% change (F(1,6) =16.495, p =

0.007);

• Ataxia score decreased from T0 and T3,

irrespective of the training mode (F(1,6) =

6.1935, p = 0.04725). The decrease occurred

during the first four sessions (F(1,6) = 10.500,

p = 0.01768);

• Tremor: TADL score decreased in the first four

sessions, but only with error augmented training

(F(1,6) = 14.087, p = 0.00947);

• TADL secondary outcome that significantly

decreased only in error augmented training

(F(1,6) = 14.087, p = 0.00947).

• Adaptive robot training improves upper limb

function.

• No significant differences- neither short-term

(within session) nor long term (between

sessions) - between error-enhancing and

error-reducing training.

(Wang

et al.,

2010)

assistive n/a

Kinematic: Times

needed assistance,

position error

• Significant improvements were observed in both

AAN Session and INT Session (p < 0.001)

• 19/20 participants needed fewer times of robotic

assistance (no p-value provided)

• Tracking performances improved with error

augmented therapy (p = 0.0014)

• Participants became more capable of

executing the task when the visual error

augmentation training method had been

integrated with the assist-as-needed training

method/no statistically significant difference

in carryover effects was observed between

the two groups

(Abdolla

hi et al.,

2011)

error

enhancing

stroke

Clinical: Fugl-

Meyer, WMFT,

ASFR, Box and

locks

Kinematic: Rom

reach value/rom

error

• Six of nineteen subjects showed significant

improvement in ROM either immediately

following treatment (p = 0.04854) or at the

follow-up phase of error-augmented treatment

(p = 0.07056)

• Error augmentation elicited varied degrees of

performance improvement as measured by the

AMFM scores based on percentage change from

pre-treatment base line values to the follow-up

evaluation (95% confidence intervals)

• Fugl-Meyer score improved (95% confidence

intervals)

• On average, subjects performed better in 1-

week follow-up evaluations than they did at

the end of the two weeks of training. It may

be that the impaired nervous system does not

react to nor does it try to learn from smaller

errors, and the EA approach may promote

learning by simply intensifying the signal-to-

noise ratio for sensory systems, making

errors more noticeable

(Molier

et al.,

2011)

none

(haptic

feedback

for error)

stroke

Clinical: FMA-UL,

Motoricity index

MI/ARAT

Kinematic:

Circulant arm

movements,

isometric strength

• Four subjects improved on the FMA-UL by

between 1.0 and 9.5 points.

• MI, two subjects improved by 8 and 13 points

each

• Four subjects improved on the ARAT by between

0.5 and 5.0 points.

• Three subjects increased workspace by between

20.2% - 63.4%

(no statistical analysis was performed)

• Emphasis on errors at the moment they occur

may possibly stimulate motor learning when

patients perform movement tasks with

sufficiently high difficulty levels.

(Casellat

o et al.,

2012)

null

additive

force

constant

disturbing

force

primary

dystonia

Clinical: BFMDRS

• Disturbing force affected significantly the

movement outcomes in healthy but not in

dystonic subjects

• In the dystonic population the altered dynamic

exposure seems to induce a subsequent

improvement, i.e. a beneficial after-effect in

terms of optimal path control, compared with the

correspondent reference movement outcome

(p = 0.05)

• The short-time error-enhancing training in

dystonia could represent an effective

approach for motor performance

improvement, since the exposure to

controlled dynamic alterations induces a

refining of the existing but strongly

imprecise motor scheme and sensorimotor

patterns.

(Shirzad

et al.,

2012)

error

enhancing

n/a

Clinical: Self-

Assessment Mankin

questionnaire

Kinematic: Absolute

deviation of

trajectory, mean of

max deviation

• Increased satisfaction and attentiveness in error

augmented therapy and even more in visual and

haptic mode (no p-values provided)

• the means of each affect measure are significantly

different between almost all pairs of conditions

(p<0.05)

• High-gain visual plus haptic EA leads to a

significantly larger amount of learning a, in

comparison with both of the visual EA methods

(p<0.1)

• Significant differences in effect (specifically:

satisfaction, attentiveness and dominance)

between progressively more exaggerated

error amplification conditions, even when

presented in random order to subject

NEUROTECHNIX2013-InternationalCongressonNeurotechnology,ElectronicsandInformatics

178