Multidisciplinary stroke rehabilitation delivered by a humanoid robot: Interaction between speech and physical therapies

fencinghuddleΤεχνίτη Νοημοσύνη και Ρομποτική

14 Νοε 2013 (πριν από 3 χρόνια και 9 μήνες)

120 εμφανίσεις

Aphasiology,2013
Vol.27,No.3,252–270,http://dx.doi.org/10.1080/02687038.2012.706798
Multidisciplinary stroke rehabilitation delivered by a
humanoid robot:Interaction between speech and physical
therapies
Yu-kyong Choe
1
,Hee-Tae Jung
2
,Jennifer Baird
3
,
and Roderic A.Grupen
2
1
Department of Communication Disorders,University of Massachusetts
Amherst,Amherst,MA,USA
2
Department of Computer Science,University of Massachusetts Amherst,
Amherst,MA,USA
3
Department of Kinesiology,University of Massachusetts Amherst,Amherst,
MA,USA
Background:A great number of stroke patients pursue rehabilitation services in multiple
domains (e.g.,speech,physical,occupational).Although multidisciplinary and interdisci-
plinary approaches to stroke rehabilitation are considered desirable,it is largely unknown
how the intervention in one domain affects the progress in others.
Aims:The current study investigated the interaction between speech therapy and physical
therapy.Additionally,the feasibility of utilising a humanoid robot in stroke rehabilitation
was described.
Methods & Procedures:A 72-year-old male chronically challenged by aphasia and
hemiparesis completed speech and physical therapy tasks in the sole condition (Speech
Only,Physical Only) and in the sequential condition (Speech & Physical).The therapy
activities were delivered by a humanoid robot.
Outcomes &Results:Greater gains in speech and physical functions were obtained during
the sole condition than in the sequential condition,suggesting a competitive interaction
between speech and physical therapies.
Conclusions:The cross-domain competition can be accounted for by fatigue,participant
characteristics,and task characteristics.Objective data on speech and physical functions
and subjective data on perceived quality of life indicate positive outcomes in this single
case.These findings warrant further research on the feasibility and utility of humanoid
robots in stroke rehabilitation.
Keywords:Aphasia;Hemiparesis;Multidisciplinary;Interdisciplinary;Robotics.
Address correspondence to:Yu-kyong Choe,PhD,CCC-SLP,Department of Communication
Disorders,University of Massachusetts Amherst,358 North Pleasant Street,Amherst,MA 01003-9296,
USA.E-mail:ychoe@comdis.umass.edu
The authors are very grateful to Dr Robert Marshall and two anonymous reviewers for their exceptional
suggestions.This work was supported by Faculty Research Grant and Faculty Research Enhancement
Award (School of Public Health and Health Sciences) from University of Massachusetts Amherst and
by National Clinical Research Program of American Heart Association.Hee-Tae Jung (second author)
acknowledges Robin Popplestone Fellowship and Jeong Song Culture Foundation Scholarship.Special
gratitude is expressed to Melissa Massery,Megan V.Cronin,Jennifer Russell,and members of the
Laboratory for Perceptual Robotics for their assistance during this study.
©2013 Taylor &Francis
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

SPEECH AND PHYSICAL THERAPIES
253
Stroke is the leading cause of severe and complex disability (Adamson,Beswick,&
Ebrahim,2004).A great number of stroke survivors are chronically challenged by
deficits in mobility,dexterity,and ability to communicate (Kelley-Hayes et al.,2003).
Such impairments in multiple domains require rehabilitation fromvarious disciplines
(Miller et al.,2010).To provide coordinated care a team of professionals may take
a multidisciplinary approach (i.e.,individual clinicians with strong autonomy pro-
vide services in parallel) or an interdisciplinary approach (i.e.,clinicians collaborate in
determining treatment plans and conducting intervention) (Korner,2010;Mickan &
Rodger,2000).Past research reported that,compared to routine care,multidisciplinary
and interdisciplinary stroke rehabilitation resulted in less mortality and better func-
tional state (Indredavik et al.,1991),greater independence in medication management
(Purdy,2007),improved activities of daily living (Yagura,Miyai,Seike,Suzuki,&
Yanagihara,2003),and greater emotional support felt by stroke patients and less
strain felt by caregivers (Lincoln,Dixon,&Knights,2004).
Despite the importance of working together with other professionals,it is largely
unknown how the intervention in one domain affects the progress in other domains.
A more specific research question investigated in the current study was:How would
effects of speech-language therapy interact with the effects of physical therapy in a
multidisciplinary approach?To answer this question the current study compared two
treatment schedules:sole (e.g.,only speech therapy was provided for 6 weeks,and then
only physical therapy was provided for 6 weeks),and sequential (e.g.,speech and phys-
ical therapies were provided in back-to-back sessions 6 weeks).At least three different
patterns of potential outcomes were predicted:No interaction,Synergistic interaction,
and Competitive interaction.
If there is no interaction between speech and physical therapies,the sole and
sequential conditions would yield equivalent outcomes.Presumably,speech and phys-
ical therapies would have only independent domain-specific effects without affecting
progress in the other area.This pattern of recovery is supported by the modular
view of cortical and cognitive functions (Fodor,1983).In this view,speech-language
production and limb movements are controlled by two autonomous modules.Thus
impairments and functional recovery in one module do not interact with those in the
other.
A second possibility is that the interventions on speech-language and limb func-
tions may yield a synergistic outcome.As a result stroke patients would make greater
gains from the sequential condition than from the sole condition.Contrary to the
modularity hypothesis,recent research proposes a common cortical area for speech
and limb movements (Rizzolatti & Craighero,2004).Broca’s area,previously known
as the motor speech area,is now considered to play a role in understanding and plan-
ning hand gestures (Baumgaertner,Buccino,Lange,McNamara,& Binkofski,2007;
Binkofski & Buccino,2004).The close link between speech and limb motor control
can be exemplified by an involuntary contraction of lip muscles during precision grips
(Higginbotham,Isaak,& Domingue,2008) and a strong correlation between mouth
and hand movements in velocity and amplitude (Gentilucci,Benuzzi,Gangitano,&
Grimaldi,2001).Furthermore,vocalising “yeah” has been reported to significantly
enhance the velocity and smoothness of hand movements in stroke rehabilitation
(Maitra,Telage,& Rice,2006).Given the tight connection between speech and hand
movements at the cortical and behavioural levels,it is likely that effective intervention
in one domain has a positive impact on the other.Nonetheless,this synergistic interac-
tion between speech and physical therapies may be more relevant to interdisciplinary
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

254
CHOE ET AL.
intervention (i.e.,simultaneous speech and arm movements) than multidisciplinary
intervention (i.e.,sequential therapy sessions as in the current study).
A third possibility states that the speech and physical functions may compete for
limited resources available in the damaged brain.Hence back-to-back sessions of
speech and physical therapies may be detrimental,yielding less gain than the sole
treatment condition.This pattern of recovery is supported by the fact that a great num-
ber of stroke survivors chronically experience physical and mental fatigue,which is
qualitatively differentiated from typical tiredness (Kirkevold,Christensen,Andersen,
Johansen,& Harder,2012).Such persisting fatigue may account for the findings of
Marshall and King (1973) that individuals with stroke-induced aphasia had signifi-
cantly lower scores in the Porch Index of Communicative Ability after 30 minutes of
lower-extremity exercises than after 30 minutes of rest.Similarly,the sequential condi-
tion of speech and physical therapies may cause certain stroke patients to experience
a heightened sense of mental and physical fatigue which,in turn,adversely affects the
treatment outcome.For those individuals the sole condition may be more beneficial.
The secondary aim of this study was to explore the feasibility of utilising a
humanoid robot (i.e.,a robot that resembles the shape of a human body) in stroke
rehabilitation.To address this aim the sole and sequential conditions of speech and
physical therapies were implemented through the use of a humanoid robot.Its mor-
phology and its physical embodiment enabled the robot to present therapy activities in
an interactive manner,promoting the participant’s motivation to complete the given
tasks.This paper reports a single-case study that investigated the interaction between
speech and physical therapies in robot-mediated stroke rehabilitation by comparing
sole and sequential therapy schedules.
METHOD
Participant
A 72-year-old male (Tom)
1
with chronic aphasia and hemiparesis participated in the
study.Tom had a left-hemisphere stroke 9 years prior to participation in the present
study.He was a monolingual English speaker and premorbidly right-handed.He
spoke in short phrases and sentences,and wrote single words and numbers.Tom had
right hemiparesis,and ambulated with a cane.He had received a doctoral degree and
formerly worked as a school superintendent.Tom was provided verbal and written
explanations of the study in the presence of his wife and gave informed consent in writ-
ten form.The procedures for this study were approved by University of Massachusetts
Amherst Institutional Review Board.
During the pre-treatment phase Tom’s speech-language and cognitive functions
were assessed by administering the Western Aphasia Battery – Revised (Kertesz,2007),
the Cognitive Linguistic Quick Test (Helm-Estabrooks,2001),and the Apraxia Battery
for Adults – Second Edition (Dabul,2000).Table 1 provides an overview of the
test results.Tom presented with moderate transcortical motor aphasia,mild verbal
apraxia,and mild cognitive-linguistic deficits.Tom’s upper-extremity functions were
assessed by the Fugl-Meyer Assessment (FMA;Deakin,Hill,& Pomeroy,2003;Fugl-
Meyer,Jaasko,Leyman,Olsson,&Steglund,1975) and the Wolf Motor Function Test
(WMFT;Wolf et al.,2010;Wolf,Lecraw,Barton,& Jann,1989).The FMA and the
1
This is not the participant’s actual name.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

SPEECH AND PHYSICAL THERAPIES
255
TABLE 1
Overview of the standardised test results
Western Aphasia Battery – Revised
Apraxia Battery for Adults –
Second edition
Cognitive Linguistic
Quick Test
Aphasia
quotient
Aphasia type Verbal
apraxia
Limb/Oral
apraxia
Composite severity
rating
73.7 Transcortical
motor
None – Mild None Mild
WMFT are the two most widely used instruments in stroke rehabilitation.These two
tests in combination can provide a comprehensive profile of a stroke survivor’s upper-
extremity functions.The FMA consists of 33 items to evaluate upper limb motor
performance of stroke patients.Items tested include reflex activity,motor synergies,
voluntary movement,grasp,and coordination.Performance on each item(e.g.,Touch
your ear with your weaker hand) is rated by a clinician (e.g.,0 – cannot be performed,
1 – detail partly performed,2 – detail is performed faultlessly).Tom received 31 out
of 66 potential points on the FMA.The WMFT assesses both gross and fine motor
function during performance of 15 timed tasks (e.g.,folding a towel).All 15 items are
highly relevant to daily activities.Some of the tasks require only single-joint control
and one-step actions,whereas other tasks require coordination of multi-joint con-
trol.Up to 120 seconds are given to complete each task.Tom completed 3 out of the
15 tasks,and his mean response time for the 3 tasks was 4.95 seconds.The tasks Tom
completed were considered to be the least difficult of the 15 WMFT items (Woodbury
et al.,2010).Results of the two tests indicated that Tom presented with moderate
right-sided hemiparesis characterised by minimal active motion in the right shoulder
and elbow,and no functional use of the right hand.
Design
Figure 1 outlines the study design that compares two treatment conditions:sole
(Speech Only;Physical Only) and sequential (Speech & Physical).Tom completed
14 speech therapy sessions in 4 weeks during the Speech Only period,and he com-
pleted 14 speech therapy sessions in 5 weeks during the Speech &Physical period.The
Physical Only period and the Speech & Physical period each consisted of 18 sessions
for upper-limb exercises.Tom’s speech and physical functions were assessed before
and after each therapy period.
Speech Only Physical Only Speech & Physical No Therapy
4 weeks 6 weeks 5/6 weeks 4 weeks
Baseline
Assessment
Assessment 1
(Speech Only)
Assessment 3
Assessment 2
Maintenance
Assessment
Sole condition
Sequential condition
Figure 1.Study design.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

256
CHOE ET AL.
Equipment and stimuli
Speech and physical therapy activities were presented through uBot-5,a humanoid
robot developed at the Laboratory for Perceptual Robotics (LPR) at the University
of Massachusetts Amherst.As depicted in Figure 2,the uBot-5 is a bi-manual mobile
manipulator that is 86 cmtall and weighs 16 kg.Each armhas 4 degrees of freedom.
The robot can move and dynamically balance on two wheels.The uBot-5’s behaviours
during the therapy sessions are implemented using control basis and action schema
frameworks,which have been developed by the LPR.These frameworks are especially
powerful at describing interaction between a robot and its environment.
The task for the robot-mediated speech practice was confrontation naming.
To make the task functionally relevant,word stimuli were selected by Tomand his wife.
The word list submitted by them included the names of family members and friends,
food items,and objects related to daily activities (e.g.,Jon,oyster,paint brush).The
baseline assessment on the words consisted of:(1) target picture naming;(2) impor-
tance rating;and (3) articulatory demand score.Tom’s ability to name target pictures
prior to the treatment phase was scored using a modification of the Porch Index of
Communicative Ability scoring system (Porch,2001).Table 2 summarises the multi-
dimensional 16-point scoring system.The scores range from 16,“correct production
without any support or delay” to 0,“no relevant response”.These 16 points are classi-
fied into five groups:Spontaneous (16–12),Cueing (11–7),Modeling (6–4),Tactile (3),
Not produced (2–0).Tom named most of the items once the target words were pre-
sented in written form,and thus his scores were 9 (Whole Word Written Cue) or above.
Any words with scores of 16–12 in the baseline assessment were eliminated fromtarget
word selection to avoid potential ceiling effects in subsequent assessments.
Even if any two items had the same baseline naming score,progress on the
two words could differ depending on functional relevance and motoric demands.
To control for those factors,importance ratings and articulatory demand scores were
obtained.Tom rated each target item (1 – somewhat important,2 – very important,
Figure 2.uBot-5.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

SPEECH AND PHYSICAL THERAPIES
257
TABLE 2
Hierarchical scoring systemfor naming responses
Score Response type Category Description
16 Complete Spontaneous Correctly says the target word without any cue
within 3 seconds.
15 Delayed Spontaneous Correctly says the target word without any support
but after more than 3 seconds’ delay.
14 Phonemic error Spontaneous Incorrect phonemes are pronounced but
spontaneously corrected (e.g.,“tat......cat” for
a target “cat”).
13 Delayed phonemic
error
Spontaneous After more than 3 seconds’ delay,incorrect
phonemes are pronounced but spontaneously
corrected.
12 Self-corrected Spontaneous Responds with a wrong word and then self-corrects
(e.g.,“dog....cat” for a target “cat”).
11 Semantic cue Cueing Correctly says the target word after a phrase or
sentence providing a semantic cue (e.g.,“it
meows” for a target “cat”).
10 Word shape cue Cueing Correctly says the target word when an initial letter
and total number of letters are given in a written
form(e.g.,“c_ _” for a target “cat”).
9 Whole word
written cue
Cueing Correctly says the target word when the whole word
is presented in a written form.
8 Initial sound cue Cueing Requires initial sound cue (e.g.,the/k/-sound for a
target “cat”) before correctly producing the target
word.The cue can be repeated once on request.
7 Lip shape cue Cueing Requires seeing a clinician silently mouth the word
before correctly producing the target word.The
cue can be repeated once on request.
6 Whole word
spoken cue
Modelling Correctly says the target word after it has been
spoken by the clinician.Modelling may be
repeated once by request.
5 Repeated
presentation
Modelling Correctly says the target word after watching the
clinician repeat the word five times.
4 Simultaneous
production
Modelling Correctly produces the target word during five times
of in-unison repetitions with the clinician.
3 Tactile cue Tactile Correctly produces the target word with touch cues
in conjunction with in-unison repetitions fromthe
clinician.
2 Incomplete Not
produced
Produces an approximation but cannot completely
produce the word.
1 Incorrect Not
produced
Produces none of the phonemes in the target word.
0 No response Not
produced
Produces no relevant response (e.g.,stereotypic
utterance).
3 – extremely important) based on the level of functional relevance of the item in
his everyday activities.Articulatory demand scores were obtained by asking Tom to
repeat a syllable consisting of a consonant and the schwa after a clinician’s model.
His responses were scored as follows:1 – correct repetition after 1 or 2 presenta-
tion(s);2 – correct repetition after 3 to 5 presentations;3 – correct repetition with
tactile cueing;4 – incorrect.The articulatory demand score for each word was cal-
culated by adding up the scores of constituting consonants.Thus this measure could
control for word length as well.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

258
CHOE ET AL.
Based on the initial assessments,60 words were divided into three experimental
conditions:sole (Speech Only),sequential (Speech & Physical),and no practice (con-
trol).The three conditions had approximately the same mean and median scores in
baseline naming performance,importance ratings,and articulatory demand scores.
Friedman tests indicated that the three conditions were statistically equivalent in
naming,χ
2
(2) =1,p =.607,importance,χ
2
(2) = 1.351,p =.509,and articulatory
demand,χ
2
(2) =0.8731,p =.646.
The 40 words in the two practice conditions were programmed for computerised
practice using Microsoft PowerPoint software.Detailed descriptions of the practice
program are provided in Choe and Stanton (2011).The main feature of the therapy
programwas the use of video clips of a speaker giving cues (e.g.,semantic,phonemic)
and models for verbal naming.Each target word was presented on six consecutive
slides with increasing levels of support.Figure 3 presents examples of practice slides.
The speech practice program was visually presented on the uBot-5’s monitor screen.
Sound output was played through speakers placed at the level of the robot’s waist.
The robot-mediated physical therapy focused on the dominant right armand hand
functions with the following goals:to enhance ranges of motion,to improve dexterity,
and to increase the use of the impaired arm and hand in daily activities.Through
an extensive discussion among a physical therapist (third author),Tom,and his wife,
it was decided that Tom would engage in three therapy exercises:Task 1 – holding
his hands together and stretching his arms to reach for the robot’s hand presented at
various points on the vertical plane,as demonstrated in Figure 4 (assisted movement);
Task 2 – flexing and extending the elbow joint to touch the robot’s hand presented at
various points on the horizontal plane (unassisted movement);and Task 3 – lifting the
forearmto touch the robot’s hand presented above Tom’s hand and at various points
on the horizontal plane (unassisted movement against gravity).
Video clip: Now say it
with me. Key (x5). What
is this?
[Slide 1]
Sound file: What is this?
[Slide 2]
Video clip: You open a lock
with this.What is this?
[Slide 4]
Video clip:It’s Key.
Say Key.
[Slide 3]
Video clip: It starts with
/k/. What is this?
[Slide 5]
Video clip: Listen to me
say it. Key (x5).What is
this?
[Slide 6]
Figure 3.Example slides of the speech practice program.Adapted from Figure 1 (A) in Choe & Stanton
(2011).Reproduced with permission.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

SPEECH AND PHYSICAL THERAPIES
259
Figure 4.Research assistant demonstrating Task 1.Source:Jung,Baird,Choe,&Grupen (2011a).
Procedure
During the Speech Only period Tomattended fourteen 30-minute sessions.The initial
criterion for completion was either correctly naming 18 out of 20 items at Slide#1
(i.e.,90%spontaneous naming) or completing 18 sessions,whichever came first.Tom
reached 85% at the fourth session,after which his performances continued to hover
around the 65–85%range.After the 14th session Tomand his wife felt that a sufficient
amount of practice had been given to the 20 target items.Thus it was decided that
the speech practice would end.To ensure the equivalent amount of practice in the two
speech-practice conditions,the number of speech sessions in the Speech & Physical
period was set at 14 as well.
During the 6 weeks of the Physical Only period,Tom attended a 1-hour session
three times per week.Each of the three upper-limb tasks was presented in two 5-
minute blocks.The order of the three tasks was always the same (i.e.,Task 1 →Task
2 → Task 3 → Task 1 → Task 2 → Task 3).Throughout the practice session the
robot presented video files of the physical therapist providing various instructions and
remarks (e.g.,“Thanks for coming in today,” “Clasp your hands together and reach
out to touch my hand,” “Nice work!It’s time to take a break.”) Between tasks Tom
took a 5-minute break where the uBot-5 played a short video clip of classical music or
a TVshowof Tom’s choice.Thus a 1-hour session typically consisted of 30 minutes of
practice (= 5 minutes × 3 tasks × 2 repetitions),25 minutes of break (= 5 minutes ×
5 breaks),and 5 minutes for Tomto get ready for tasks and wrap up to leave.
During the 6-week Speech & Physical period,Tom attended a therapy session
three times per week.The first 14 sessions included 30 minutes of speech practice
and 1 hour of upper-limb exercises.The order of speech and physical therapies was
counterbalanced across sessions:upper-limb exercises were completed first in odd-
numbered sessions (e.g.,Sessions 1,3,5) and speech practice was completed first in
even-numbered sessions (e.g.,Sessions 2,4,6).Between speech and physical therapy
activities,Tomtypically took a 5-minute break.The last four sessions of the Speech &
Physical period only included 1 hour of physical therapy.
Throughout the therapy periods Tom primarily interacted with the uBot-5 during
the practice sessions.The researcher quietly monitored Tom’s performance and did
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

260
CHOE ET AL.
not provide any feedback on Tom’s performance,except for general comments (e.g.,
“I appreciate that you completed today’s practice”).
Analysis
Daily probes of naming during the practice phase.As depicted in Figure 3,each
target word was presented in six slides.Because the slides provided increasing levels of
support for naming,Tom’s progress on target items was easily observed.During the
practice the researcher recorded at which of the six slides Tomwas able to produce the
target word.The slide number (e.g.,Slide#3 that provided phonemic cue) quantified
how much support Tom needed from the practice program to name the target item.
This level of required support was monitored throughout the Speech Only period and
the Speech &Physical period.
Daily probes of upper-limb functions during the practice phase.Tom’s performance
in the three upper-limb exercises was measured by frequency (i.e.,number of trials
completed within a set duration)
2
and range of motion (i.e.,distance to the target).
Task difficulty was regularly adjusted by moving the targets further fromTom.At the
same time,feedback fromTomwas used to ensure that the tasks were challenging but
attainable with effort.An increase in the range of motion was directionally defined by
three components (Figure 5):x – forward reach of the movement,y – height of the
movement,and |z| – lateral reach of the movement in both directions.Tom’s progress
in frequency and range of motion was documented during the Physical Only period
and the Speech &Physical period.
Progress in naming over the five assessments.As delineated in Figure 1,Tom’s abil-
ity to name 60 target words was tested five times.Tom’s responses were scored using
the 16-point scale (Table 2).The data were divided into three groups based on prac-
tice condition:20 words practised during the Speech Only period,20 words practised
Figure 5.Range of motion as directionally defined by three components.Source:Jung,Baird,Choe,&
Grupen (2011b).
2
The number of trials was automatically logged by the custom-built software that operated the uBot-5’s
behaviours.At the same time a researcher manually tallied the number as a backup.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

SPEECH AND PHYSICAL THERAPIES
261
during the Speech & Physical period,and 20 words never practised.The three groups
were separately analysed to detect the direct treatment effect on trained items,gen-
eralisation to untrained items,and maintenance of treatment effect.To determine
significance of the gains,a statistical analysis compared baseline performance to scores
in the four subsequent assessments.Given the non-parametric nature of the data
(Table 2),Wilcoxon signed-ranks tests were utilised with a Bonferroni corrected alpha
(p <.0125) for the multiple comparisons.In addition to the statistical analysis based
on the score changes in the 16-point scales,the number of items Tom named spon-
taneously without the clinician’s support was counted.Such spontaneous responses
received scores ranging from16 to 12 (Table 2).
Progress in upper-limb functions over the four assessments.The Fugl-Meyer
Assessment (FMA;Deakin et al.,2003;Fugl-Meyer et al.,1975) and the Wolf Motor
Function Test (WMFT;Wolf et al.,1989,2010) were used to measure the changes in
Tom’s upper-limb functions.
Changes in quality of life over the five assessments.To document Tom’s percep-
tion on his quality of life,the Quality of Communication Life Scale (QCLS;Paul
et al.,2004) and the Stroke and Aphasia Quality of Life Scale – 39 (SAQOL – 39;
Hilari,Byng,Lamping,& Smith,2003) were administered at the five assessments.
Non-parametric data fromthe two scales were analysed using Wilcoxon signed-ranks
tests with a Bonferroni corrected alpha (p <.0125) for the multiple comparisons.
RESULTS
Daily probes of naming during the practice phase
Figure 6 presents the average amount of support required by Tomat each practice ses-
sion.He needed slightly less support (i.e.,named more items spontaneously) during
the Speech Only period than in the Speech &Physical period.Day-to-day fluctuations
in Tom’s performances were noted especially during the Speech & Physical period.
On certain days when speech practice followed the upper-limb exercises (Sessions 5,
0
1
2
3
4
5
6
1 2 3 4 5 6 7 8 9 10 11 12 13 14
Speech-only
Speech & Physical
Amount of Support Required
Sessions
Figure 6.Average level of support required for Tomto produce target words in each practice session.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

262
CHOE ET AL.
7,9,& 11),Tomrequired more support for naming than in the immediately previous
sessions when speech practice preceded the upper-limb exercises (Sessions 4,6,8,&
10).These patterns of Tom’s progress during the practice periods suggest more effi-
cient re-acquisition of target items in the sole condition and an adverse fatigue effect
of physical tasks on speech practice in the sequential condition.
Daily probes of upper-limb functions during the practice phase
Figure 7 presents cumulative increases in the range of motion for the three tasks.Tom
made notable gains in the height he could reach with his upper extremities.The average
increase across the tasks was 14.9 cm during the Physical Only period and 6.4 cm
during the Speech & Physical period.Additionally,the number of trials completed
in each 5-minute block of the tasks was recorded.The average across the three tasks
was 7.0/minute in the Physical Only period and 6.8/minute in the Speech & Physical
period.These data on the frequency and range of motion together suggest that Tom
made greater improvements in the range of motion without sacrificing the frequency
of movements during the Physical Only period,as compared to the Speech &Physical
period.
Progress in naming over the five assessments
Figure 8 depicts the data from the five assessments organised by practice condition.
A visual inspection of the scores suggests treatment-specific gains during the two
speech practice periods (Speech Only,Speech & Physical) with little generalisation.
Table 3 summarises the results of Wilcoxon signed-ranks tests.Tom’s ability to name
the 20 words practised during the Speech Only period significantly improved from
baseline to Assessment 1 (Z =3.312,p =.001,r =.523) and Maintenance Assessment
(Z = 2.625,p =.009,r =.415).His gains in the 20 words practised during the
Speech & Physical period were significant at Assessment 3 (Z = 3.557,p <.001,r
=.566) and Maintenance assessment (Z = 2.692,p =.007,r =.426).Tom’s scores
0
5
10
15
20
25
30
35
40
1 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35
Task 1: y
Task 1: |z|
Task 2: x
Task 2: |z|
Task 3: y
Increase in Range of Motion (cm)
Sessions
Speech &
Physical
period
Physical
Only
period
3
Figure 7.Cumulative increase in range of motion for the three tasks:Task 1 – reaching with both hands;
Task 2 – elbow flexion and extension;Task 3 – forearmlifting.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

SPEECH AND PHYSICAL THERAPIES
263
0
2
4
6
8
10
12
14
16
Baseline 1 2 3 Maintenance
Speech Only
Speech & Physical
Control
Median naming scores
Assessments
Speech
Only
period
Speech &
Physical
period
Physical
Only
period
*
*
* * *
Figure 8.Median naming scores of the words assigned to three speech practice conditions (Speech Only,
Speech &Physical,Control – no practice) at five assessments.

indicates significant at p <.0125 (Bonferroni
corrected alpha).
TABLE 3
Results of Wilcoxon Signed Tests comparing each assessment to baseline performance
Assessments
Practice condition 1 2 3 Maintenance
Z =3.312 Z =1.802 Z =1.807 Z =2.625
Speech Only p =.001

p =.072 p =.071 p =.009

r =.524 r =.285 r =.286 r =.415
Z =0.677 Z =0.962 Z =3.557 Z =2.692
Speech &Physical p =.498 p =.336 p <.001

p =.007

r =.105 r =.152 r =.566 r =.426
Z =1.069 Z =1.134 Z =0.862 Z =2.658
Control p =.285 p =.257 p =.389 p =.008

r =.169 r =.179 r =.136 r =.420

Significant at p <.0125 (Bonferroni corrected alpha).
on the 20 control items,which were not practised during the study periods,were sig-
nificantly better at Maintenance Assessment than at baseline (Z = 2.658,p =.008,
r =.420).
Figure 9 presents the number of spontaneous responses in each practice condi-
tion over the five assessments,as well as incremental changes in the total number of
spontaneous responses in each therapy period.Similar to the statistical analysis stated
above,this examination of spontaneous responses suggests treatment-specific gains,
little generalisation during therapy periods,and maintenance of treatment effect and
delayed generalisation at 4 weeks post-treatment.
Progress in upper-limb functions over the four assessments
The FMAscores showed little changes over the four assessments (Table 4).Tomconsis-
tently completed the same 3 out of the 15 WMFT tasks over the assessments.The three
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

264
CHOE ET AL.
0
2
4
6
8
10
12
14
Baseline 1 2 3 Maintenance
Speech Only
Speech & Physical
Control
Assessments
Number of spontaneous responses
Speech
Only
period
Physical
Only
period
Speech &
Physical
period
(+ 2)(+ 16) (-10) (+ 11)
Figure 9.Number of spontaneous responses to items assigned to three speech practice conditions (Speech
Only,Speech & Physical,Control – no practice) at five assessments.The number in parentheses indicates
the incremental change in total spontaneous responses during each therapy period.
TABLE 4
Results of the Fugl-Meyer Assessment:Scores in total and in the subsections
Assessments
Baseline 2 3 Maintenance
Total (/66) 31 30 31 32
Shoulder/Elbow/Forearm(/36) 18 16 17 17
Wrist (/10) 3 4 4 5
Hand (/14) 7 6 6 6
Coordination/Speed (/6) 3 4 4 4
tasks were:placing the forearmon the table,placing the hand on the table,and pulling
a 1-lb weight across the table.The mean response time of the tasks fluctuated as shown
in Figure 10.Tom’s movements became notably faster over the Physical Only period,
slower over the Speech & Physical period,and again faster over the 4-weeks of No
0
1
2
3
4
5
6
Baseline 2 3 Maintenance
Speech &
Physical
period
Physical
Only
period
Response Time (secs)
Assessments
Figure 10.Average response time of completed tasks in the Wolf Motor Function Test at four assessments.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

SPEECH AND PHYSICAL THERAPIES
265
Therapy.This pattern of progress in upper-limb functions mirrors his improvements
in naming ability:more gains in the sole condition than in the sequential condition.
Changes in quality of life over the five assessments
Figures 11 and 12 present Tom’s responses to the QCLS and the SAQOL – 39 respec-
tively.Compared to baseline,Tom’s responses to the QCLS was significantly higher at
Assessment 1 (Z=2.714,p =.007,r =.480) and Maintenance Assessment (Z=2.970,
p =.003,r =.525).When the incremental changes during each therapy period were
evaluated,significant gains were observed only during the Speech Only period.Tom’s
responses to the SAQOL – 39 significantly improved frombaseline to the Assessment
2 (Z = 2.677,p =.007,r =.303),Assessment 3 (Z = 3.129,p =.002,r =.354),and
0
1
2
3
4
5
Baseline 1 2 3 Maintenance
Overall QCL
General QOL
Speech
Only
period
Speech &
Physical
period
Physical
Only
period
*
*
*
Assessments
Figure 11.Overall quality of communication life (QCL) scores and general quality of life (QOL) scores
fromthe Quality of Communication Life Scale (Paul et al.,2004).
0
1
2
3
4
5
Baseline 1 2 3 Maintenance
Physical Score
Communication Score
Psychosocial Score
Energy Score
SAQOL-39 Mean Score
Speech
Only
period
Speech &
Physical
period
Physical
Only
period
Assessments
*
*
*
*
Figure 12.The mean scores of the four sub-domains (physical,communication,phychosocial,energy) and
the overall mean scores fromthe Stroke and Aphasia Quality of Life Scale – 39 (Hilari et al.,2003).
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

266
CHOE ET AL.
Maintenance Assessment (Z =2.696,p =.007,r =.305).Incremental gains were sig-
nificant only over the Physical Only period (Z =2.684,p =.007,r =.304).These data
from the QCLS and the SAQOL – 39 indicate that the most notable gains in Tom’s
subjective perception of communicative life occurred during the Speech Only period
and that the most notable gains in his perception of overall quality of life occurred
during the Physical Only period.
Tom’s and his wife’s perception of the robot-mediated therapy
Upon completing Assessment 3,Tom and his wife independently responded to a 23-
itemquestionnaire on the robot-mediated therapy.It used 5-point scales for responses
with 1 being the most positive and 5 being the most negative.The items were grouped
into four subsections:Overall Experience (e.g.,I enjoyed working with the robot;
Yes =1,No = 5),Feelings and Impressions (e.g.,Working with the robot was...;
Pleasant =1,Unpleasant = 5),Comments on the Robot (e.g.,I like that the robot
has arms;Like = 1,Dislike = 5),and Future Plans (e.g.,I would come back for more
speech and physical therapies with the robot;Yes = 1,No = 5).The median scores of
Tom’s and his wife’s responses were both 1,indicating overall positive perception of
the robot-mediated intervention.One of the items included in the Overall Experience
subsection was I would rather work with clinicians than with the robot (No =1,Yes =5).
To this item,Tomgave 2,and his wife gave 3.
DISCUSSION
Tom,a stroke survivor chronically experiencing aphasia and right-sided hemiparesis,
completed a robot-mediated rehabilitation program in the sole condition (i.e.,only
speech therapy for 4 weeks and then only physical therapy for 6 weeks) and in the
sequential condition (i.e.,speech and physical therapies for 5/6 weeks).Tom’s naming
scores at the five assessments demonstrate practice effects on trained items during the
therapy periods,and maintenance of the gains and delayed generalisation at 4-week
post-treatment.Tommade notable gains in the frequency and range of the upper-limb
movements.Tom’s responses to the Quality of Communication Life Scale (Paul et al.,
2004) and the Stroke and Aphasia Quality of Life Scale – 39 (Hilari et al.,2003) suggest
significant improvements in perceived quality of life during the practice periods and
at 4-week post-treatment.
To the authors’ knowledge this study was the first attempt to implement a
humanoid robot in delivering speech and physical therapy services to stroke patients.
The objective data on speech and physical functions and the subjective data on quality
of life that have been documented by the current study warrant further investigations
of utilising a humanoid robot in stroke rehabilitation.If future research supports
feasibility and utility of certain therapy robots,potential placements of such robots
at clinical and residential settings can enhance the availability and accessibility of
intensive and interactive therapy services,especially for patients residing in rural areas.
The main objective of the current project was to investigate the interaction between
speech and physical therapies.Data from this single case suggest a competitive inter-
action between the two domains.Tom mastered target items for verbal naming more
efficiently and improved frequency and range of motion to a greater extent in the sole
condition (Speech Only,Physical Only) than in the sequential condition (Speech &
Physical).Daily progress in his naming ability during the Speech & Physical period
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

SPEECH AND PHYSICAL THERAPIES
267
suggests a fatigue effect (Figure 5).The significant practice effect obtained during the
Speech Only period vanished during the Physical Only and Speech & Physical peri-
ods and reappeared at 4-week post-treatment (Table 3,Figure 8,Figure 9).Similarly,
the notable gain in the speed of movement achieved during the Physical Only period
was not maintained during the Speech & Physical period but reappeared at 4-week
post-treatment (Figure 10).This advantage of the sole therapy schedule over the
sequential schedule suggests that speech and physical functions competed for limited
resources.
The adverse effect of upper-limb exercises on subsequent naming tasks observed
during the Speech & Physical period (Figure 5) is consistent with the fatigue effect
that stroke survivors demonstrated in speech-language performances after 30 min-
utes of lower-extremity exercises (Marshall &King,1973).In addition,the significant
performance decline in communicative ability from morning sessions to afternoon
sessions (Marshall,Tompkins,& Phillips,1980) can likewise be attributed to accu-
mulated fatigue that stroke survivors experience fromtheir daily activities.The current
study has replicated these previous findings and expanded themby demonstrating that
the fatigue experienced during practice can ultimately affect the treatment outcomes
in speech-language and physical functions.It is likely that several factors escalated
the mental and physical fatigue and,as a result,amplified the competitive interac-
tion between speech and physical therapies.Such factors include the participants’ age,
severity,and type of speech program.
Advanced age is strongly associated with reduced cognitive capacity (Gilchrist,
Cowan,& Naveh-Benjamin,2008),sensorimotor decline (Lindenberger & Baltes,
1994;Wohlert & Smith,1998),and inefficiency in learning new skills (Daselaar,
Rombouts,Veltman,Raaijmakers,& Jonker,2003).The participant of the current
study was a 72-year-old male,whose reduced capacity in cognitive and motor func-
tioning presumably limited the level of recoveries in speech and physical functions.
It is likely that the adverse effect of age-related deterioration was amplified by the
more demanding therapy schedule of the sequential condition,as compared to the sole
condition.The hypothesised effect of age-related capacity decline warrants a research
study that compares younger and older patient groups in their performances in sole
and sequential therapy schedules.
Severity is another factor that likely contributed to the competitive interac-
tion between speech and physical therapies.The initial assessments revealed that
Tom presented with moderate aphasia,mild verbal apraxia,and moderate-to-severe
hemiparesis.It is possible that patients with milder deficits in speech and physical
functions may not experience the cross-domain competition.
The speech practice program used in this study provided increasing levels of sup-
port (i.e.,least-to-most cues) for naming and promoted effortful learning of target
items.Although effortful learning is as beneficial as errorless learning (Fillingham,
Sage,& Ralph,2005;McKissock & Ward,2007),the effort exerted in word retrieval
and speech production may consume more resources for rehabilitation and cause a
higher level of fatigue.Thus a practice programthat provides decreasing levels of sup-
port (i.e.,most-to-least cues) can reduce the competitive interaction between speech
and physical therapies.Yet another alternative is a lexical-perceptual training program
that requires no effort for overt naming and that has been shown to improve verbal
naming (Fridriksson et al.,2009).Eliminating motor requirements for speech prac-
tice (Fridriksson et al.,2009) can further reduce overall fatigue and enhance treatment
outcomes.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

268
CHOE ET AL.
Potential effects of these participant characteristics (i.e.,age,severity) and task
characteristics (i.e.,effortful vs errorless vs no articulation) should be investigated in
future research.Moreover,therapy activities can be further refined in future studies.
In the current study the uBot-5 dynamically interacted with Tomduring physical ther-
apy sessions whereas it remained static during speech therapy sessions.In a follow-up
study the robot can be programmed to play a board game that incorporates the speech
practice program.The interactive exchange with the humanoid robot can enhance the
participant’s motivation to complete therapy tasks.
The current robotics technology enabled the uBot-5 to deliver therapy activities in
a highly consistent and structured manner across sessions and across treatment con-
ditions.Because of this consistency,objective data of the participant’s progress was
easily and instantly collected.At the same time,the uBot-5 was lacking the flexibility
that a human clinician would demonstrate in response to Tom’s verbal and non-verbal
reactions to therapy activities.This lack of flexibility and adaptability to the par-
ticipant’s immediate needs might have contributed to an increased level of fatigue.
To overcome this shortcoming,next generation of therapy robots need to be equipped
with sophisticated sensor technology,such as a computer vision systemthat monitors
facial expression,eye gaze,and head posture (Ji,Zhu,&Lan,2004;Lang,Wachsmuth,
Wersing,&Hanheide,2010;Sanchez et al.,2011;Tian,Kanade,&Cohn,2001).
This study took a multidisciplinary approach in stroke rehabilitation and observed
a competitive interaction between speech and physical therapies.The treatment sched-
ule of the present study reflected routine clinical practice where speech and physical
therapies would be separately provided in back-to-back sessions.As a result,the
average duration of an entire session was 30–60 minutes for the sole condition and
90 minutes for the sequential condition.To eliminate this confounding factor of
session length and,more importantly,to examine the effect of simultaneously pro-
vided therapies,a future study could take an interdisciplinary approach by embedding
speech-language tasks into physical exercises.Aparticipant could be required to reach
his/her impaired arm for the picture of a target object and simultaneously say the
name of the object.A potential comparison between back-to-back multidisciplinary
sessions and a single interdisciplinary session will yield useful information with clinical
implications.
Manuscript received 19 February 2012
Manuscript accepted 22 June 2012
First published online 31 August 2012
REFERENCES
Adamson,J.,Beswick,A.,&Ebrahim,S.(2004).Is stroke the most common cause of disability?Journal of
Stroke and Cerebrovascular Disease,13,171–177.
Baumgaertner,A.,Buccino,G.,Lange,R.,McNamara,A.,& Binkofski,F.(2007).Polymodal conceptual
processing of human biological actions in the left inferior frontal lobe.European Journal of Neuroscience,
25,881–889.
Binkofski,F.,& Buccino,G.(2004).Motor functions of the Broca’s region.Brain and Language,89,
362–369.
Choe,Y.,& Stanton,K.(2011).The effect of visual cues provided by computerised aphasia treatment.
Aphasiology,25,983–997.
Dabul,B.(2000).Apraxia Battery of Adults (2nd ed.).Austin,TX:Pro-Ed.
Daselaar,S.M.,Rombouts,S.A.R.B.,Veltman,D.J.,Raaijmakers,J.G.W.,&Jonker,C.(2003).Similar
network activated by young and old adults during the acquisition of a motor sequence.Neurobiology of
Aging,24,1013–1019.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

SPEECH AND PHYSICAL THERAPIES
269
Deakin,A.,Hill,H.,& Pomeroy,V.M.(2003).Rough guide to the Fugl-Meyer Assessment:Upper limb
section.Physiotherapy,89,751–763.
Fillingham,J.K.,Sage,K.,& Ralph,M.A.L.(2005).Treatment of anomia using errorless versus errorful
learning:Are frontal executive skills and feedback important?International Journal of Language and
Communication Disorders,40,505–523.
Fodor,J.A.(1983).The modularity of mind.Cambridge,MA:MIT Press.
Fridriksson,J.,Baker,J.M.,Whiteside,J.,Eoute,D.,Moser,D.,Vesselinov,R.,& Rorden,C.(2009).
Treating visual speech perception to improve speech production in nonfluent aphasia.Stroke,40,
853–858.
Fugl-Meyer,A.,Jaasko,L.,Leyman,I.,Olsson,S.,& Steglund,S.(1975).The post-stroke hemiplegic
patient:A method for evaluation of physical performance.Scandinavian Journal of Rehabilitation
Medicine,7,13–31.
Gentilucci,M.,Benuzzi,B.,Gangitano,M.,& Grimaldi,S.(2001).Grasp with hand and mouth:A
kinematic study on healthy subjects.Journal of Neurophysiology,86,1685–1699.
Gilchrist,A.L.,Cowan,N.,& Naveh-Benjamin,M.(2008).Working memory capacity for spoken sen-
tences decreases with adult ageing:Recall for fewer but not smaller chunks in older adults.Memory,16,
773–787.
Helm-Estabrooks,N.(2001).Cognitive Linguistic Quick Test.San Antonio,TX:Harcourt Assessment
Company.
Higginbotham,D.R.,Isaak,M.I.,& Domingue,J.N.(2008).The exaptation of manual dexterity for
articulate speech:An electromyograminvestigation.Experimental Brain Research,186,603–609.
Hilari,K.,Byng,S.,Lambing,D.L.,& Smith,S.C.(2003).Stroke and aphasia quality of life scale – 39
(SAQOL – 39) evaluation of acceptability,reliability,and validity.Stroke,34,944–1950.
Indredavik,B.,Bakke,F.,Solberg,R.,Rokseth,R.,Haaheim,L.L.,&Holme,I.(1991).Benefit of a stroke
unit:Arandomized controlled trial.Stroke,22,1026–1031.
Ji,Q.,Zhu,Z.,&Lan,P.(2004).Real-time nonintrusive monitoring and prediction of driver fatigue.IEEE
Transactions on Vehicular Technology,53,1052–1068.
Jung,H.,Baird,J.,Choe,Y.,& Grupen,R.A.(2011a).Upper-limb exercises for stroke patients through
the direct engagement of an embodied agent.In A.Billard,P.H.Kahn Jr.,J.A.Adams,&J.G.Trafton
(Eds.),Proceedings of the 6th ACM/IEEE International Conference on Human-Robot Interaction (pp.
157–158).Lausanne:Switzerland.
Jung,H.,Baird,J.,Choe,Y.,&Grupen,R.A.(2011b).Upper-extremity physical therapy for stroke patients
using a general purpose robot.In Proceedings of the 2011 IEEE International Symposium on Robot and
Human Interactive Communication (pp.270–275).
Kelley-Hayes,M.,Beiser,A.,Kase,C.S.,Scaramucci,A.,D’Agostino,R.B.,& Wolf,P.A.(2003).The
influence of gender and age on disability following ischemic stroke:The Framinghamstudy.Journal of
Stroke and Cerebrovascular Disease,12,119–126.
Kertesz,A.(2007).Western Aphasia Battery (Rev.ed.).San Antonio,TX:Harcourt Assessment Company.
Kirkevold,M.,Christensen,D.,Andersen,G.,Johansen,S.P.,& Harder,I.(2012).Fatigue after stroke:
Manifestations and strategies.Disability and Rehabilitation,34,665–670.
Korner,M.(2010).Interprofessional teamwork in medical rehabilitation:A comparison of multidisci-
plinary and interdisciplinary teach approach.Clinical Rehabilitation,24,745–755.
Lang,C.,Wachsmuth,S.,Wersing,H.,& Hanheide,M.(2010).Facial expressions as feedback cue in
human–robot interaction – a comparison between human and automatic recognition performances.
Proceedings of the IEEE Computer Society Conference – Computer Vision and Pattern Recognition
Workshops,79–85.
Lincoln,N.B.,Dixon,W.A.,& Knight,P.(2004).Evaluation of a multiprofessional community stroke
team:Arandomized controlled trial.Clinical Rehabilitation,18,40–47.
Lindenberger,U.,& Baltes,P.B.(1994).Sensory functioning and intelligence in old age:A strong
connection.Psychology and Aging,9,339–355.
Maitra,K.K.,Telage,K.M.,& Rice,M.S.(2006).Self-speech-induced facilitation of simple reaching
movements in persons with stroke.American Journal of Occupational Therapy,60,146–154.
Marshall,R.C.,&King,P.S.(1973).Effects of fatigue produced by isokinetic exercise on the communica-
tion ability of aphasic adults.Journal of Speech and Hearing Research,16,222–230.
Marshall,R.C.,Tompkins,C.A.,& Phillips,D.S.(1980).Effects of scheduling on the communicative
assessment of aphasic patients.Journal of Communication Disorders,13,105–114.
McKissock,S.,& Ward,J.(2007).Do errors matter?Errorless and errorful learning in anomia picture
naming.Neuropsychological Rehabilitation,17,355–373.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013

270
CHOE ET AL.
Mickan,S.,& Rodger,S.(2000).The organisational context for teamwork:Comparing health care and
business literature.Australian Health Review,23,179–192.
Miller,E.L.,Murray,L.,Richards,L.,Zorowitz,R.D.,Bakas,T.,Clark,P.,& Billinger,S.A.(2010).
Comprehensive overview of nursing and interdisciplinary rehabilitation care of the stroke patient.
Stroke,41,2402–2448.
Paul,D.R.,Frattali,C.M.,Holland,A.L.,Thompson,C.K.,Caperton,C.J.,& Slater,S.C.(2004).
Quality of Communication Life Scale.Rockville,MD:American Speech-Language-Hearing
Association.
Porch,B.E.(2001).Porch Index of Communicative Ability (4th ed.).Albuquerque,NM:PICAPrograms.
Purdy,M.(2007).Increasing stroke patients’ success in self-medication programs using an interdisciplinary
cognitive rehabilitation approach.Rehabilitation Nursing,32,210–219.
Rizzolatti,G.,& Craighero,L.(2004).The mirror-neuron system.Annual Review of Neuroscience,27,
169–192.
Sanchez,A.,Ruiz,J.V.,Moreno,A.B.,Montemayor,A.S.,Hernandez,J.,& Pantrigo,J.J.(2011).
Differential optical flow applied to automatic facial expression recognition.Neurocomputing,74,
1272–1282.
Tian,Y.,Kanade,T.,& Cohn,J.F.(2001).Recognizing action units for facial expression analysis.IEEE
Transactions on Pattern Analysis and Machine Intelligence,23,97–115.
Wohlert,A.B.,& Smith,A.(1998).Spatiotemporal stability of lip movements in older adult speakers.
Journal of Speech,Language,and Hearing Research,41,41–50.
Wolf,S.L.,Catlin,P.A.,Ellis,M.,Archer,A.L.,Morgan,B.,& Piacentino,A.(2010).Assessing Wolf
Motor Function Test as outcome measure for research in patients after troke.Stroke,32,1635–1639.
Wolf,S.L.,Lecraw,D.E.,Barton,L.A.,& Jann,B.B.(1989).Forced use of hemiplegic upper extremities
to reverse the effect of learned nonuse among chronic stroke and head injured patients.Experimental
Neurology,104,125–132.
Woodbury,M.,Velozo,C.A.,Thompson,P.A.,Light,K.,Uswatte,G.,Taub,E.,& Wolf,S.L.(2010).
Measurements structure of the Wolf Motor Function Test:Implications for motor control theory.
Neurorehabilitation and Neural Repair,24,791–801.
Yagura,H.,Miyai,I.,Seike,Y.,Suzuki,T.,& Yanagihara,Y.(2003).Benefit of inpatient multidisciplinary
rehabilitation up to 1 year after stroke.Archives of Physical Medicine and Rehabilitation,84,1687–1691.
Downloaded by [University of Massachusetts, Amherst], [Yu-kyong Choe] at 15:16 14 March 2013