Content uploaded by Carolee Winstein
Author content
All content in this area was uploaded by Carolee Winstein on Jul 12, 2014
Content may be subject to copyright.
685
Top Stroke Rehabil 2011;18(6):685–700
© 2011 Thomas Land Publishers, Inc.
www.thomasland.com
doi: 10.1310/tsr1806-685
Virtual Reality and Robotics for
Stroke Rehabilitation: Where
Do We Go from Here?
Eric Wade, PhD,1,2 and Carolee J. Winstein, PhD, PT 1,3
1Motor Behavior and Neurorehabilitation Laboratory, Division of Biokinesiology and Physical Therapy at the Herman Ostrow School of
Dentistry, University of Southern California, Los Angeles, California; 2Interaction Laboratory, Department of Computer Science
at the Viterbi School of Engineering, University of Southern California, Los Angeles, California; 3Department of Neurology, Keck School
of Medicine, University of Southern California, Los Angeles, California
An ongoing challenge in the field of
rehabilitation is determining how to promote
functional recovery from motor defi cits due
to neurological impairments such as stroke. There
are 795,000 new or recurring incidences of stroke
annually in the United States; many more than half
of these individuals continue to live with some type
of motor impairment.1 Task-oriented training (TOT)
has emerged as one of the dominant approaches to
restoring functional ability after stroke.2 Although
there are several different interpretations of TOT,
we defi ne it using 3 criteria. First, the training must
be suffi ciently challenging to facilitate learning.
Second, it must be progressive and adaptable so that
users will continue to acquire or refi ne new skills
as the system adjusts to both short- and long-term
time scales. Third, it must be suffi ciently interesting
and meaningful to engage the user in active problem
solving. Based on these criteria, we know the
rehabilitation practice tasks should be specifi c and
data driven. They should have adjustable diffi culty
levels, and the measure of diffi culty should be
quantifi able to allow for systematic progression
and patient-specifi c periodic assessment.
Two technological domains have shown con-
siderable promise for neurorehabilitation, most
markedly for the implementation of task-oriented
training programs. The fi rst, virtual reality (VR)–
assisted rehabilitation, is characterized by systems
that include a user-computer interface, real-time
simulation of an environment, and multimodal
sensory channels to allow for user interaction.3 The
second, robot-assisted rehabilitation, is characte-
rized by the use of devices with sensory, actuation,
and intelligence capabilities to facilitate motor and/
or neurological interaction with the user. In both
cases, users interact with a technological inter-
face that models (or mimics) real-world scenarios.
These technological tools present themselves
as promising options for promoting recovery of
function implemented through such programs as
TOT due to their quantitative, reliable, and adap-
tive nature. In this article, we explore some of the
Promoting functional recovery after stroke requires collaborative and innovative approaches to neurorehabilitation
research. Task-oriented training (TOT) approaches that include challenging, adaptable, and meaningful activities have led
to successful outcomes in several large-scale multisite defi nitive trials. This, along with recent technological advances of
virtual reality and robotics, provides a fertile environment for furthering clinical research in neurorehabilitation. Both virtual
reality and robotics make use of multimodal sensory interfaces to affect human behavior. In the therapeutic setting, these
systems can be used to quantitatively monitor, manipulate, and augment the users’ interaction with their environment,
with the goal of promoting functional recovery. This article describes recent advances in virtual reality and robotics and the
synergy with best clinical practice. Additionally, we describe the promise shown for automated assessments and in-home
activity-based interventions. Finally, we propose a broader approach to ensuring that technology-based assessment and
intervention complement evidence-based practice and maintain a patient-centered perspective. Key words: assessment,
neurorehabilitation, rehabilitation, robotics, task-oriented training, virtual reality
686 TOPICS IN STROKE REHABILITATION/NOV-DEC 2011
physical therapy received additional balance
training through the presentation of various inter-
active scenarios. They were assigned tasks, such
as manipulating objects in the virtual space, that
would engage various balance training activities.
When compared to control participants who re-
ceived only a conventional intervention, partici-
pants receiving the VR training showed increased
dynamic balance and walking velocity. In another
study, Flynn et al used the off-the-shelf Sony Play-
station EyeToy and demonstrated its promise as
a low-cost tool for balance rehabilitation.32,33 The
EyeToy is a projected video capture system that
uses a USB camera to display images of the player
into a game. These researchers showed clinically
relevant improvements in the Dynamic Gait Index
after the use of the EyeToy with an individual with
chronic stroke for 20 one-hour sessions.33 These
examples represent the application of dramatically
different VR systems (eg, a fully immersive cus-
tomized system vs an off-the-shelf game device) to
a problem common in individuals poststroke.
VR has also been used to help wheelchair-bound
persons with spatial attention and upper extremity
(UE) rehabilitation. Buxbaum et al demonstrated
that a virtual wheelchair task could be used to
assess wheelchair navigation abilities.18 Admin-
istration of the VR task was more feasible than
the administration of a real-life navigation assess-
ment; further, participant performance showed
positive correlation with Moss-Magee Wheelchair
Navigation Test scores. VR was also used to ac-
commodate training and exercise for hemiplegic
individuals who required manual wheelchairs for
propulsion.34,35 The use of rewarding activities and
biofeedback has been shown to improve task per-
formance and motivation to practice. O’Connor
et al made use of such biofeedback with GAME-
Wheels, an interface between a portable roller sys-
tem and a computer that enables a wheelchair user
to play commercially available computer video
games during wheelchair propulsion exercises.
They were able to measure individuals’ physiologi-
cal responses and helped the users to reach and
maintain an optimal exercise level.36–38
VR- and game-based systems have also been
applied to the restoration and rehabilitation
of hand and fi nger function in persons with
stroke.39–41 Adamovich et al created a VR-based
issues surrounding the use of these technologies in
rehabilitation.
We begin by describing recent work in the do-
mains of VR and robot-assisted rehabilitation. We
follow this with a brief discussion of 2 promising di-
rections for research in these domains— automated
assessments and the facilitation of in-home function-
al recovery. Finally, we propose a broader approach
to ensuring that such technological interventions
are designed with attention to the interests of both
the patient and clinician and that they be guided
by the most up-to-date evidence in rehabilitation
science.
VR for Stroke Rehabilitation
VR-based rehabilitation is characterized by the
integration of cognitive and physical tasks within
a multimodal sensory environment. VR holds the
promise of being able to enhance physical function
in meaningful activities by harnessing advances
in science and technology with the capability to
systematically enhance (or limit) sensory capacity.
The meaningfulness of tasks (performed in the
virtual environment) to the user can potentially
affect those core processes at the body function
and activities level as defi ned in the International
Classifi cation of Functioning, Disability and Health
(ICF) model.4 VR has been used to address defi cits
associated with upper (UE) and lower extremity
(LE) motor control,5–11 hands and fi ngers,12,13 gait
training,10,14,15 balance,16,17 wheelchair use,17,18
cognition,19,20 mental practice,21,22 community
living,23–25 and spatial neglect.26–29 (The interested
reader is referred to the focused reviews cited in
references 6, 9, 17, and 30.) Here, we point out
a few projects relevant to our discussion, as they
make use of approaches appropriate for a TOT
framework.
A particular defi cit with which VR has shown
promise is impaired postural control. Balance im-
pairment is common in stroke patients; it can lead
to falls and restrict an individual’s normal motor
activities, thereby limiting the sense of indepen-
dence and adversely affecting quality of life.31 Kim
et al designed a system that made use of a monitor,
video camera, and cyber gloves to create the VR
experience designed to improve balance during
ambulation.16 Participants receiving conventional
Virtual Reality and Robotics 687
Lokomat (Hokoma, Inc, Volketswil, Switzerland).
This orthosis is a treadmill with a robotic device
that provides body weight support and varying
levels of gait assistance by guiding the legs through
the gait cycle. In a pilot study, Westlake and Pat-
ten compared the effects of traditional BWSTT to
those of Lokomat-based therapy.70 Although they
found no differences in primary outcomes mea-
sures, they found that participants working with
the Lokomat improved on 4 secondary measures
(including self-selected walk speed). A multisite
randomized clinical trial led by Hidler was incon-
clusive about the relative effi cacy of robot-assisted
BWSTT compared to traditional gait retraining,
but results by Husemann and others have shown
promise for the technology.71,72 One criticism of
the Lokomat orthosis is that it limits mobility of
the user’s trunk and pelvis. This has led to the
investigation of orthoses with additional degrees
of freedom to allow for trunk and pelvic move-
ment and to facilitate “more natural” gait patterns.
Agrawal et al developed a LE orthosis that pro-
vides varying levels of active gait assistance while
allowing for more upper body mobility.73 Their
device has shown promise in early pilot studies
and is currently being developed for a randomized
controlled trial.
The vast majority of robotic devices developed
for individuals poststroke are custom robots for
UE rehabilitation. Many of these rehabilitation
robots are hands-on tools for performing specifi c
exercises with the hand or arm; they are instru-
mented with motors to generate and apply as-
sistive forces, sensors to measure user-applied
forces, and a computer to monitor and display
progress (some examples include devices deve-
loped by Krebs, Reinkensmeyer, and Kahn).74–76
Robot-guided movements range from passive,
whereby the robot moves the more affected limb
in the absence of movement from the patient, to
active-assisted wherein the human makes a con-
tribution to the movement while the robot assists
to complete the movement and active-resisted
wherein the robot provides resistance to the move-
ment. Such guided rehabilitation has included in-
duced rapid-passive elbow and wrist movements
without respect to a target,44 robot-assisted aiming
tasks in which subjects attempt to move toward a
computer displayed target,77–81 and resisted aiming
system for hand rehabilitation that trained fi nger
range of motion, fi nger fl exion speed, indepen-
dence of fi nger motion, and fi nger strength in 8
individuals with chronic stroke.39 All participants
showed improvements on outcome measures in-
cluding kinematic analyses of motion and the
Jebsen-Taylor Test of Hand Function (JTTHF).40
In another study, Szturm et al used an interactive
computer gaming system, coupled with the
manipulation of common objects, as a form of
repetitive, task-specifi c movement therapy for
patients with spinal cord injury and stroke.42 Partic-
ipants selected from 25 different commercial video
games of varying levels of diffi culty. They manip-
ulated instrumented objects in front of the game
screen resulting in improvements on the JTTHF.
The role of VR in rehabilitation continues to be
the subject of much interest and discussion, as
evidenced by the discussion at the recent Medicine
Meets Virtual Reality meeting.43 Much of the dis-
cussion on aging into and with disability focused
on the proliferation of federally funded centers
investigating ongoing research in VR for stroke
rehabilitation. As these systems demonstrate, the
ability to manipulate task meaningfulness, sensory
modality, and the practice environment make
VR-based rehabilitation attractive for individuals
poststroke.
Robotics for Stroke Rehabilitation
Like VR, robotics have been used as a
rehabilitation tool for a number of impairments,
including defi cits in UE and LE motor control,44–50
hands and fingers,51–54 wrist,53,55 gait,49,56–58
position sense,59–61 movement dynamics,61,62
proprioception,63,64 and quality of life65,66 (see
reviews for additional details44,67–69).
Research in robot-assisted rehabilitation for
LE recovery has focused on gait retraining. Body
weight–supported treadmill training (BWSTT) of
the gait pattern often requires 2 to 3 clinicians to
physically move the patients’ legs, while a har-
ness supports some of the patients’ body weight
as they walk on a treadmill. As an alternative to
this physically intensive process, researchers have
used LE orthoses to assist with moving the users’
legs during the intervention. A well-known tool
is the commercial robot-driven gait orthosis,
688 TOPICS IN STROKE REHABILITATION/NOV-DEC 2011
Because human-robot contact involves issues of
safety, hands-on robotics methods can face challenges.
A new approach to rehabilitation using socially
assistive robotics (SAR) has grown in response
to some of these challenges.87 The noncontact
approach of SAR affords the participant the oppor-
tunity to engage in functional therapeutic interac-
tions conveniently and safely within the clinic or
home in a user-friendly manner (in this way, SAR
is similar to noncontact VR strategies). With a pilot
grant from the University of Southern California
Institute for Creative Technologies and the Nation-
al Science Foundation, we developed a noncontact
robot-assisted rehabilitation system that uses light-
weight wearable inertial measurement unit (IMU)
sensors to determine the position of the stroke-
affected limb (see Figure 1).87,88 We then conducted
a pilot study to examine the effectiveness of that
system and to determine the patient’s satisfaction
as a noncontact robot encouraged use of the pa-
retic arm for functional tasks. Data from that pilot
study revealed positive responses from interactions
between people affected by stroke and our non-
contact rehabilitation robot.89
Therefore, we can say that these robot-based
interventions, along with the VR projects dis-
cussed earlier, show a promising direction for the
use of technology to supplement rehabilitation
programs. The next logical and unique approach
after demonstrating technological feasibility is to
develop ways of using VR and robotics-based tech-
nologies as embedded and novel training and as-
sessment components of theoretically defensible,
evidence-based, TOT programs. What follows is a
discussion of how these concepts can be extended
to automated assessments and in-home rehabilita-
tion and recovery.
Automated Assessments
Based on early progress in the domains of VR
and robotics for stroke rehabilitation, we argue
that a promising direction for these technologies
is the automation of clinical assessments. This
conclusion is motivated by (1) the need for
quantitative, standardized objective measures
of function; (2) the desire to obtain measures of
function currently left uncollected; and (3) the
tasks.44,82,83 Reinkensmeyer developed the ARM
Guide to help individuals poststroke learn how
to target an object with their affected arm. Kahn
and colleagues conducted a study where they de-
termined that improvements in participants who
used the ARM Guide were comparable to those of
participants who performed traditional reaching
exercises.76 Outcome measures for the aforemen-
tioned work have been predominantly impair-
ment scale scores; Fugl-Meyer Assessment (FMA)
scores increased following the robotic intervention
periods (ranging from 6 weeks to 2 months) in
4 out of 5 recent studies.77–81,83,84 Two of these stud-
ies included follow-up measures revealing that
these effects were neither maintained at 4 months
nor at 6 months.78,81,84
In a study by Lum et al, strength and reaching
distance of the more-affected limb increased in
participants poststroke who underwent 2 months
of contact robotic training for the upper limb as
compared to controls who received neurodevel-
opmental therapy.78,84 It is not reported whether
these changes were sustainable as follow-up scores
were not reported. More recently, Schweighofer
et al developed the adaptive automated robotic
task practice system (ADAPT) that simulates
real-world UE tasks.85 The ADAPT robot uses a
variety of tools to simulate tasks, such as turning
a doorknob or unscrewing a jar, and automatically
changes the task tool and progresses the partici-
pant according to performance. Early results from
the ADAPT robot have demonstrated techno-
logical feasibility of the device. More recently, a
multisite randomized controlled trial to evaluate
the effectiveness of the MIT Manus robot-assisted
therapy system was completed.86 The trial com-
pared 3 groups; participants receiving intensive
robot-assisted therapy, intensive (human-assisted)
comparison therapy, and usual care. Overall re-
sults indicated that both groups receiving inten-
sive therapy (robot or human) showed modest
improvements on FMA scores. Participants in the
robot-assisted group showed signifi cantly better
results on the Stroke Impact Scale, but there were
no statistically signifi cant differences in primary
or secondary outcomes between the 2 intensive
groups, which were both better than the usual
care group.
Virtual Reality and Robotics 689
sensor data can be mined for valuable metrics
of motor performance (eg, time to peak velocity,
transport/grasp coordination).
In the domain of robotics, the dominant ap-
proach to assessments has been to use technologi-
cal devices to measure quantitative outputs and to
determine their relationships to functional ability.
Mazzoleni et al developed a robotic apparatus for
measuring the performance of activities of daily
living (ADLs).95 The robot was designed to acquire
quantitative measures of force and torque (F/T)
during the performance of 6 ADL tasks. The sys-
tem restricted users to isometric movements and
used forward dynamics models to estimate gen-
erative F/T in the participants’ limbs. In effect, the
robotic system was used to sense the user’s move-
ments (as opposed to acting against them). Bala-
subramanian et al also developed an apparatus to
quantify the required amount of active assistance,
motion smoothness, and movement synergy when
UE exercises were performed.96 They demon-
strated the technological feasibility of their device
and described a technique to estimate motion
ability to perform assessments more often for a
fi ner-grained picture of patient recovery. We will
explain these points in more detail.
Using technology to obtain quantitative, objec-
tive measures of functional ability is becoming
feasible due to the establishment of more valid
and reliable outcome measure instruments and ad-
vances in sensor technologies. Assessments such
as the Wolf Motor Function Test (WMFT) and the
FMA contain both objective (timing) and subjec-
tive (functional ability) scoring methods.91,92 The
subjective scoring depends heavily on the training
and skill of the professional administering the test;
the inherently objective nature of VR and robotic
systems, on the other hand, can reduce interrater
subjectivity and provide standardization. For large
multisite clinical trials and experiments, standardi-
zation is required to ensure that the participants
are evaluated according to the same metrics and
criteria; the importance of such standardization
is the subject of ongoing study.93,94 A quantitative
assessment tool can be used to ensure that the
assessments are consistently administered, and
Figure 1. Socially assistive robot (Bandit) guiding a motor task practice session for an individual poststroke.
The robot provided instructions, feedback, and encouragement as hemiparetic individuals poststroke
practice an upper extremity (UE) wire puzzle practice task (adapted from Wade et al90).
690 TOPICS IN STROKE REHABILITATION/NOV-DEC 2011
with sensors worn on the participants’ hands to
measure proprioception. Individuals were blind-
folded and asked to align the palms of their
hands in parallel 2-dimensional (2D) planes. The
researchers found that the VR system captured dif-
ferences in the 22 stroke-affected participants and
the nondisabled control participants that were not
evident in the standard proprioception test (the
up/down test).100 They showed correlation between
standard measures of proprioception and distance
errors (between the centers of both palms) mea-
sured by their system. In our laboratory, we have de-
veloped systems to evaluate spontaneous UE limb
choice in reaching movements (see Figure 2).101,102
Both systems utilize projected targets and elec-
tromagnetic sensing to stimulate and monitor
reaching movements. Use of the affected limb is
known to be an important measure of recovery for
individuals poststroke; tools to measure such limb
choices give us insight into how individuals incor-
porate their paretic limb into their daily life, and
preliminary results indicate that these tools may
be useful for understanding recovery at the activity
and participation level.
For UE motor assessments, Bonato et al affi xed
triaxial accelerometers to the subjects’ arms and
performed the WMFT. They then compared ac-
celerometer readings for the participants. Initial
results indicated the possibility of determining
different levels of UE functional ability (eg, mod-
erate versus severe impairment) from the raw ac-
celerometer data.103 In our work, we used slightly
“smoothness,” which is one of the important
metrics in assessments such as WMFT and FMA.
Scott et al utilized the 3 degree of freedom (DOF)
KINARM (BKIN Technologies, Kingston, Ontar-
io, Canada) to assess UE position sense.60,97 The
KINARM apparatus was used to guide the nonpa-
retic limb to a set position, and participants were
instructed to move their paretic limb to the mirror
position. They found that their system was suffi -
cient for monitoring position sense.
Another approach to robotic assessment is to
correlate quantitative measures with known assess-
ment instruments. Van Dijck et al utilized posterior
probability profi les in conjunction with results from
Mazzoleni to correlate quantitative F/T measures
with FMA scale scores for study participants.98 Co-
lombo et al developed a system that used an actu-
ated gripper to measure F/T for isometric motions
and characterized the required (active) robotic
assistance, movement velocity, and movement ac-
curacy.99 In their article, they show signifi cant cor-
relation between these measures and FMA scores
for a pilot trial with 16 individuals poststroke.
Automating assessments will also allow us to
capture data that are currently left uncollected.
During the administration of clinical assessments
such as the WMFT or FMA, there are objective,
quantifi able data (eg, joint kinematics) that are
not being measured. Additional quantitative data
can be benefi cial and insightful and will paint a
more complete picture of user functional ability.
Leibowitz et al developed a noninvasive system
Figure 2. Participant working with 2 different virtual reality (VR) setups designed to capture limb choice
and kinematic and dynamic measures of reaching performance. (A) Depicts a VR system used to capture
kinematic measures of reaching and motor planning (adapted from Stewart102). (B) Depicts a VR system
used to determine spontaneous arm choice in reaching movements (adapted from Chen et al101).
A B
Virtual Reality and Robotics 691
Functional Recovery in the Home
Another promising direction for VR and robotics
is the development of home-based interventions.
Introducing rehabilitation tools in the home
has the advantages of (1) providing a view of
short- and long-term effects of interventions,
(2) capturing implicit and explicit intervention
effects, and (3) increasing the quality, intensity,
and duration of in-home task practice. Current
approaches to understanding the effects of
interventions, including clinical evaluations,
standardized outcome measures, and transfer and
retention tests, provide a snapshot of functional
ability and measures of learning on the same, or
similar, tasks to those undertaken in therapeutic
practice.110 However, because they provide a
more instantaneous picture, there is always the
possibility that the participant having a bad
day (or any other temporary condition affecting
performance) can affect the measurement. To fully
determine the effects of clinical interventions, we
need a much more granular understanding of
what happens to the patient outside of the clinical
setting. Acknowledging that learning cannot be
directly assayed, most models of motor learning
include a signifi cant degree of off-line processing
during consolidation.111 These consolidated
memories are expressed by behavior observed
later outside the therapeutic setting; the behavior
refl ects the integration of what has been learned
into the real-world environment. Proper use of
some of the aforementioned devices can move
toward the goal of capturing the longer term and
meaningful benefi ts of motor therapies.
Home-based applications lie within the domain
of telemedicine, broadly defi ned as the delivery of
health care and sharing of medical knowledge over
a distance using tools for telecommunication.112–115
There are examples of the instrumentation of
appliances and of wearable devices for general ap-
plications (that is, not targeted for individuals post-
stroke).116–118 For instance, Junnila et al describe a
system for remote monitoring of pedometers, heart
rate sensors, a bed sensor, an infrared sensor, and a
capacitive fl oor sensor.119 Marshollek et al describe
a technique using accelerometers, galvanic skin
response (GSR), skin temperature sensors, heat
fl ux sensors, and pedometers to monitor ADLs.120
more complex inertial measurement units (consist-
ing of 3 axes each of accelerometers, rate gyros, and
magnetometers) in conjunction with an overhead
camera104,105 (see Figure 3). Our goal was to au-
tomate of the WMFT timing score. Using a single
sensor on the wrist and an overhead camera, we ob-
tained more accurate timing information than that
recorded on a stopwatch (as is done traditionally).
Finally, there are limited examples of standard
assessments administered in the home setting un-
der the remote supervision of a clinician. Riva and
Gamberini describe early work in VR for remote
assessment106 that included an attempt to auto-
mate the administration of the Wisconsin Card
Sorting Test.107 Rovetta et al remotely assessed grip
strength through the use of a dynamometer.108,109
In this last example, the clinician observed the
participant using video monitors.
When these quantitative sensor-based data are
reliably correlated with standard instrument out-
come measures, the data can be used to directly
measure patient capability. Thus, another advan-
tage of automation is that it will enable patients to
perform self-assessments (that is, without the pre-
sence of a clinician). This will lead to assessments
that can be performed in the home (see further
discussion in the next section).
Figure 3. Experimental setup used to obtain
kinematic and dynamic measures of performance
for Wolf Motor Function Test (WMFT) assessment
tasks. The individual performs the WMFT tasks,
guided by a trained physical therapist, while
wearing an inertial measurement unit on the wrist
(adapted from Wade et al105).
692 TOPICS IN STROKE REHABILITATION/NOV-DEC 2011
information is preferred to a self-report. More
information regarding ADLs contributes to a more
complete understanding of the effect and quality
of interventions. An understanding of how data
regarding long-term learning correlate with exist-
ing outcome measures would allow for their use as
predictors of long-term participant recovery.
Finally, the portability and user-friendly interfac-
es of these systems make them ideal for increasing
the quality, intensity, and duration of task pra ctice
in the home. Anecdotal evidence from other do-
mains has indicated that the physical presence of
a robot can help motivate participants to persist
with an activity longer than they would when
being guided either by a computer-based or hand-
written log.87,126 Building on these results, we have
begun an investigation into the ability of our SAR
system to guide and maintain intense, task-specifi c
practice for patients poststroke.89
Testing Evidence-Based Techniques
Building on the evidence of progress in the
use of technologies in rehabilitation, we now
suggest possible goals for current and future
work in this domain. We discussed at length the
potential benefi ts and applications of technological
approaches to rehabilitation. We now caution
that this enthusiasm must be tempered with an
understanding of the limitations of these systems.
Adamovich et al raise an important question:
How do we know that the neural activity present
in VR tasks is similar to that present during the
performance of real-world tasks?3 They argue
that further studies involving imaging are needed
to understand the underlying causes of recovery
when interacting with VR systems. In addition to
translational research, Adamovich et al suggest
a need for imaging studies to evaluate the effects
of sensory manipulation on brain activation
patterns and effects of various training parameters
on long-term changes in brain function. Larger
clinical studies are also needed to establish the
efficacy of sensorimotor rehabilitation using
VR in various clinical populations and, most
important, to identify the training parameters
that are associated with optimal transfer to real-
world functional improvements. Finally, the
application of robust outcome measures across the
3 domains of functioning as defi ned in the ICF,
Medjahed et al take a different approach by de-
scribing how to monitor ADLs through the in-
strumentation of in-home devices including a
telephone, doorbell, sink faucets, coffee machines,
dishwashers, and other appliances.121 Gourlay
et al developed a VR system that includes a fully
functional virtual kitchen for the preparation of
meals using kitchen appliances.122 These systems
are low-cost and safe for the user, allowing for the
practice of ADLs without the risk of physical harm.
These examples indicate the potential for the long-
term use of multimodal sensory tools in the home.
Through sensor fusion techniques, which aggre-
gate data from multiple sources and use observed
patterns to label and classify complex activities,
many important variables can be observed, in-
cluding the use of the affected limb immediately
after a practice session or use of the affected limb
as time from the intervention increases. Although
these data can provide intervention-specifi c infor-
mation, they can also be used to discern general
behaviors of people undergoing a rehabilitation
regimen and provide insight into in-home com-
pliance. Thus, short-term effects (immediately
after a practice session, between sessions, after
a fi nal session) and longer-term effects (1 week,
1 month, or 1 year after) of interventions can be
made accessible to clinicians for the development
of stronger evidence-based practices and better
theoretical models of recovery.
Home-based interventions will also allow us to
capture the relationship between patient-perceived
and actual intervention effects to obtain a measure
of the patient’s self-effi cacy. Patient-perceived out-
comes are often obtained using self-report data.
The specifi city and style of self-reporting prompts
can lead to unintentional variability in responses.
It would be of great interest to compare participant
responses to questions regarding limb use to an ac-
tual measure of relative arm motion. For instance,
the Motor Activity Log asks patients, “Over the
past 3 days, how frequently did you use your af-
fected arm to open the refrigerator door?”123 Com-
paring this response to objective sensor measures
could provide insights into self-effi cacy and recall
bias in individuals poststroke. Further, there is evi-
dence that focused attention to a well-learned or
automatic action can disrupt motor performance
(also known as the Bliss-Boder effect).110,124,125 In
such cases, a passive device obtaining performance
Virtual Reality and Robotics 693
course of a single session and during much longer
recovery periods. The ability of a robot to do so
can only be validated by evaluating the similarity
of the recovery in the test subjects who use the
robot to those receiving therapy from a clinician.
This is the new bar for rehabilitation technologies
that will be required to justify their use in clinical
practice.
Future Directions
We began this discussion by describing TOT, and
we now return to TOT, as it is the most promising
area for direct application of VR and robotics as
described in this article. As we mentioned, TOT
requires tasks that are quantitative, adjustable,
adaptive, and meaningful. The background work
described presents evidence of systems capable
of 1 or more of these requirements. The diffi culty
level of VR and robotic-assisted rehabilitation can
be adjusted according to the challenge level of the
patient. VR and robotics systems can supervise
repetitive task practice simply by using an
appropriate programming loop. The technologies
have been shown to be intrinsically motivating and
interesting for participants. With feedback based
on multimodal sensory capabilities of the systems
we have described, the patient can be an active and
more integral participant in the formulation and
progression of his or her therapy. We have already
begun an investigation into the use of robots to
validate 2 important characteristics of TOT (the
ability to maintain challenge level for the participant,
and its ability to involve the participant in making
decisions that engage active problem solving) and
look forward to the work of other investigators in
this domain. From our experience, the continued
success of VR- and robot-assisted rehabilitation will
require overcoming specifi c challenges.
Partnerships for a client-centered approach
Close partnerships between clinicians and
engineers are necessary to optimally advance
the field. This is a sentiment that has been
expressed elsewhere, but bears repeating in light
of the current state of the fi eld (Campbell et al
describe this very generally for interdisciplinary
collaborations129). These partnerships should be
ongoing collaborations; a specifi c example of
particularly in the participation domain, is vital
for the development of evidence-based guidelines
regarding the effectiveness of VR and robotics-
based interventions that could have the potential to
impact the local and national agenda for the future.
For both VR and robotics, participants are typi-
cally excited by the technology and initially show
an affi nity for the devices and games.127,128 How-
ever, to what extent is this novelty effect due to the
“gee-whiz” nature of the device? Six months after
a robot is introduced into the home, how can we
ensure that its behavior and interactions remain
novel and engaging? Because VR and robotics
simulate real-world situations, we need a better
understanding of how skills acquired in simula-
tion map to skills required for real-world ADLs.
Also, as we alluded to in the Automated Assess-
ments section, we must determine how clinical
outcome measures relate to the quantitative data
obtained by these technological devices. Riva and
Gamberini note the lack of reference standards for
VR systems.106 In other words, many devices are
“one-off,” which can make them diffi cult (or im-
possible) to use with populations other than those
for which they were exclusively designed.
The resolution of these questions necessitates
the use of traditional vetting procedures. Campbell
et al argue that nonpharmacological interventions
need to be evaluated with the same rigor as phar-
macological interventions.129 In our opinion, this
extends to VR and robotics. Much of the research
being published in conferences and performed in
research institutions remains in the preclinical,
modeling, and exploratory phases. These phases
are characterized by validation with participants
outside of the target population (eg, nondisabled,
non–age-matched adults), by small numbers of
participants, or by technological feasibility studies.
Evaluation of how such technological interven-
tions will lead to clinically meaningful outcomes
remains necessary. One promising direction is to
make use of the quantitative outcomes resulting
from evidence-based approaches. Results from the
rehabilitation literature regarding human-human
interactions and best practices can be considered
as control data that can be compared to techniques
used in VR and robotics. For instance, just as TOT
requires repetition and adaptability studies, stud-
ies that use a robot for TOT should thus evaluate
its ability to adapt to user performance over the
694 TOPICS IN STROKE REHABILITATION/NOV-DEC 2011
consider that more than 1 person can be using
VR or a robotic device at the same time while the
clinician attends to another client. Propagation of
these technologies thus depends on decreasing
technology costs and establishing therapies
that can outperform the traditional therapeutic
intervention in 1 or more ways.
Technology augments rehabilitation
Task-specific practice is considered to be
the most important element of any behavioral
training program, particularly when improved
functional skills are sought (eg, cognitive and
physical). In fact, the effects of practice are often
underestimated and, all too often, programs fail
to be effective because either (1) ample practice
time was not prescribed or (2) compliance was
poor. Our work has shown that in addition to
time-on-task practice, the practice structure
(eg, variable or constant) is important for
optimizing consolidation and motor learning.131
As such, we return to the notion raised earlier
where we suggest a departure from the insular view
that these technologies are stand-alone solutions.
Rather, the next step, after the demonstration
of technological feasibility, is to develop ways
of using VR- and robotics-based interventions
as embedded and novel elements for training
and assessments within theoretically defensible,
evidence-based, task-oriented training programs.
The notion is that the added technology should
augment the existing patient-centered program
both in the clinic and in the home after discharge
(Figure 4).
This last point leads to our fi nal conjecture.
Rather than thinking of VR and robotics tools
as stand-alone technologies that might replace
such a partnership was presented in Mazzoleni
et al,95 who followed an iterative design
process that involved consultation among the
engineers, clinicians, and patients. In addition to
conversations with end users, the authors studied
anthropometrical population data, built computer-
based simulations, performed 3D ergonomic
studies, built physical mockups, and constructed
multiple prototypes before arriving at the fi nal
design for their device. Engineers need to witness
therapeutic interventions fi rsthand to understand
the intricacies of the patient-clinician interaction.
At the same time, clinicians need to understand the
strengths (and particularly the limitations) of the
technology. Observations by both the engineers
and clinicians of what worked and did not work
for what specifi c outcomes must then be used to
iterate the design. Because much of this research
is new, the details required to create a system
both technologically feasible and useful for the
target population will require continued contact
among all interested parties. Keshner and Rymer
said it best in the foreword to the book, Advanced
Technologies in Rehabilitation: “The developers and
users must be familiar with the scientifi c rationale
for motor learning and motor control, as well as
the impairments presented by different clinical
populations.”130
Financial costs of VR and robotic systems
Right now, the VR and robotic systems (and to
a lesser extent, video game systems) mentioned
here are prohibitively expensive for individual
purchase. The direct costs of these devices are
not often reported, so it is hard to state a specifi c
dollar amount. However, because VR and robotic
systems are not mass produced, the combination
of the intelligence, imaging, sensing, and actuated
elements can easily range from thousands to tens
of thousands of dollars. To justify their cost as
eventual home-based tools, they must be able to
outperform a human clinician. Much of the work
cited here has preliminarily shown comparable
results between outcomes for participants using
the technological intervention and those receiving
clinician-driven therapies. Therefore, the cost
of these devices cannot be justifi ed outside of
the experimental or research setting unless you
Figure 4. The staging of technological interventions
and assessments relative to stroke recovery.
Automated
assessments
Automated
assessments
In-home
interventions
Acute Subacute Chronic
Virtual Reality and Robotics 695
of the technology need not be purely diagnostic;
rather, it can be introduced at a stage agreed on by
the clinician and the patient (see Figure 4). In this
way, these devices will augment and further em-
power and enable the clinician’s effectiveness and,
at the same time, follow the health care delivery
model into the community where primary care is
accomplished.
We would expect that if the technology is em-
bedded in a therapy program such that the active
ingredients for an effective intervention program
are considered, the technology-enhanced program
would be shown to be superior to a clinician-only
program. The common active ingredients for an
effective intervention program include capacity
building (eg, physiological and cognitive impair-
ment mitigation), skill acquisition using principles
of motor learning (eg, repetition without repeti-
tion, neural substrates of consolidation), and mo-
tivational enhancement (eg, intrinsic motivation,
self-effi cacy, self-management, meaningfulness).94
In cases where the technology was found to be as
effective for achieving meaningful rehabilitation
outcomes as the human-delivered therapy,72,86 we
offer the following checklist of useful questions
that, if addressed, should remedy the problem
and take full advantage of the potential enhanc-
ing capability that the technology offers: (1) Was
the challenge level suffi cient? (2) Did the cli-
ent acquire a new skill? (3) What was learned?
(4) Was the practice task meaningful? (5) What
fundamental capacities were addressed (eg,
strength, fl exibility, control)? (6) Did the client de-
velop self-effi cacy for performing the activity/task?
Conclusion
We are enthusiastic about the role of VR and
robotics in motor neurorehabilitation. With
regard to automated assessment, VR and robotics
stand to address standardization, reliability, and
frequency of administration. The inherently
quantitative nature of the tools used in VR
and robotics means that assessments can be
administered more consistently and within the
training program itself. Of course, this depends
on signifi cant, reliable correlations between the
quantitative data measured by these systems
and those physiological and cognitive measures
humans; we should think of them as tools that
can be used to aid clinicians during the “staging
of therapies.” It makes little sense to perform a
one-to-one comparison between VR and robotic
systems and human clinicians; no existing VR or
robotic system can grasp the complex dynamic
interactions that occur between 2 humans during
rehabilitation. All applications mentioned here are
meant to be utilized in addition to the tools already
available to clinicians. Rather than replace or re-
duce requirements for humans, these applications
should be designed to augment and strengthen
the role and decision making for clinicians and
patients to do what is best for the patients’ func-
tional recovery. For example, a common practice
for treating someone with Parkinson disease who
has just begun a regimen of dopamine is to send
them home to try out the drug. During a follow-
up appointment the physician will ask, “How did
you do at home when you were on the drug? Did
you have any falls? Were you able to move around
better?” The patient reports his or her perception
of how they did, and the drug dosage is adjusted
depending on the subjective report. This practice
could be fi ne-tuned with the introduction of ac-
tivity monitors or social robots in the home that
could monitor movements, activities, and events
(eg, falls) and establish a more fi ne-grained data-
base for future clinical decision making (eg, drug
dosage). This is similar to common practices in
cardiovascular medicine with heart monitors that
are given to patients for an extended period of time
(eg, 48 hours to 1 week). It provides more granular
information than the snapshot that is observed at
the single monthly visit to the clinic. In the typi-
cal case, the appropriateness of technological inter-
ventions may be time, or “stage,” dependent. For
instance, after an individual suffers from a stroke,
they are assessed and receive therapy from a hu-
man clinician during the acute phase of recovery.
At the point when they can show profi ciency in the
clinic, they return home and can begin an auto-
mated therapeutic regimen based on the clinician’s
prescription. At monthly intervals, they perform a
home-based automated assessment that the clini-
cian can use to monitor their progress. If progress
is observed, the automated system adjusts its
challenge level accordingly to enable the patient to
continue learning and progressing. Thus, the role
696 TOPICS IN STROKE REHABILITATION/NOV-DEC 2011
activity regimens. Combining these technologies
with useful motor and neurological rehabilitative
tasks will lead to a deeper understanding of
the relationship between clinical interventions and
the patient’s functional and meaningful life.
Acknowledgments
This work was supported in part by funding
from the National Institutes of Health, the National
Institute of Neurological Disorders and Stroke,
and the National Center for Medical Rehabilitation
Research award number U01 NS056256 for
the Interdisciplinary Comprehensive Arm
Rehabilitation Evaluation Stroke Initiative;
the National Institute for Disability and
Rehabilitation Research through the Rehabilitation
Engineering Research Center on Successful Aging
with a Disability, grant H133E080024; NSF CNS-
0709296 grant for CRI: IAD-Computing Research
Infrastructure for Human-Robot Interaction and
Accuracy, and the extension of our system to the
remaining Socially Assistive Robotics and NSF
IIS-0713697 grant HRI: Personalized Assistive
Human-Robot Interaction: Validation in Socially
Assistive Robotics for Post-Stroke Rehabilitation.
familiar to clinicians, and, ultimately, to the
patients. Therefore, it is critical that we begin to
understand the relationship between quantitative
measures and motor neurological recovery.
This logical step is necessary to move the fi eld
forward. The results of such work are also critical
to the development and refi nement of in-home
rehabilitation applications. These technologies will
shift traditional clinical applications, including
assessments and monitored task practice, into the
home and community, which will benefi t both
clinicians and users. For clinicians, this means
more data and understanding of what participants
are doing outside the clinical setting. Particularly
valuable is the knowledge of the use of affected
limbs in the long- and short-term following stroke.
This will provide critical information regarding
the real, lasting effects of interventions and help
establish the relationship between assessment
instruments and a person’s functional recovery.
With respect to task practice, both game-like
VR and social robots have been demonstrated
to be extrinsically motivating for participants.
This desire to perform well at VR games or to
work with robots has the potential to drastically
increase a patient’s compliance with exercise/
REFERENCES
1. American Heart Association. Heart Disease and
Stroke Statistics—2010 Update: A Report From the
American Heart Association. Dallas, TX: AHA; 2010.
2. National Institute of Neurological Disorders and
Stroke. Report of the Stroke Progress Review Group —
September 2006. Bethesda, MD: National Institutes
of Health; 2006.
3. Adamovich SV, Fluet GG, Tunik E, Merians AS,
et al., Sensorimotor training in virtual reality: a
review. NeuroRehabilitation. 2009;25(1):29–44.
4. World Health Organization. International
Classifi cation of Functioning, Disability and Health.
Geneva: WHO; 2010.
5. Piron L, Turolla A, Agostini M, et al. Exercises for
paretic upper limb after stroke: a combined virtual-
reality and telemedicine approach. J Rehabil Med.
2009;41(12):1016–1020.
6. Lucca LF. Virtual reality and motor rehabilitation
of the upper limb after stroke: a generation of
progress? J Rehabil Med. 2009;41(12):1003–1006.
7. Trotti C, Menegoni F, Baudo S, Bigoni M, Galli
M, Mauro A. Virtual reality for the upper limb
motor training in stroke: a case report. Stud Health
Technol Inform. 2009;144:257–260.
8. Rand D, Katz N, Weiss PL. Intervention using the
VMall for improving motor and functional ability of
the upper extremity in post stroke participants. Eur
J Phys Rehabil Med. 2009; 45(1):113–121.
9. Crosbie JH, Lennon S, Basford JR, McDonough
SM. Virtual reality in stroke rehabilitation:
still more virtual than real. Disabil Rehabil.
2007;29(14):1139–1146; discussion 1147–1152.
10. Mirelman A, Patritti BL, Bonato P, Deutsch
JE. Effects of virtual reality training on gait
biomechanics of individuals post-stroke. Gait
Posture. 2010;31(4):433–437.
11. Dunning K, Levine P, Schmitt L, Israel S, Fulk G.
An ankle to computer virtual reality system for
improving gait and function in a person 9 months
poststroke. Top Stroke Rehabil. 2008;15(6):602–610.
12. Adamovich SV, Fluet GG, Mathai A, Qiu Q, Lewis
J, Merians AS. Design of a complex virtual reality
simulation to train fi nger motion for persons with
hemiparesis: a proof of concept study. J Neuroeng
Rehabil. 2009;6:28.
13. Merians, AS, Tunik E, Adamovich SV. Virtual
reality to maximize function for hand and arm
rehabilitation: exploration of neural mechanisms.
Stud Health Technol Inform. 2009;145:109–125.
14. BaramY, Aharon-Peretz J, Lenger R. Virtual reality
feedback for gait improvement in patients
with idiopathic senile gait disorders and
Virtual Reality and Robotics 697
29. Katz N, Ring H, Naveh Y, Kizony R, Feintuch U,
Weiss PL. Interactive virtual environment training
for safe street crossing of right hemisphere stroke
patients with unilateral spatial neglect. Disabil
Rehabil. 2005;27(20):1235–1243.
30. Deutsch JE, Mirelman A. Virtual reality-based
approaches to enable walking for people
poststroke. Top Stroke Rehabil. 2007;14(6):45–53.
31. Nichols DS. Balance retraining after stroke
using force platform biofeedback. Phys Ther.
1997;77(5):553–558.
32. Rand D, Kizony R, Weiss PT. The Sony PlayStation II
EyeToy: low-cost virtual reality for use in rehabilitation.
J Neurol Phys Ther. 2008;32(4):155–163.
33. Flynn S, Palma P, Bender A. Feasibility of using
the Sony PlayStation 2 gaming platform for an
individual poststroke: a case report. J Neurol Phys
Ther. 2007;31(4):180–189.
34. Nelson DL, Konosky K, Fleharty K, et al. The
effects of an occupationally embedded exercise
on bilaterally assisted supination in persons
with hemiplegia. Am J Occup Ther. 1996;50(8):
639–646.
35. Cunningham D, Krishack M. Virtual reality: a
wholistic approach to rehabilitation. Stud Health
Technol Inform. 1999;62:90–93.
36. O’Connor TJ, Cooper RA, Fizgerald SG, et al.,
Evaluation of a manual wheelchair interface to
computer games. Neurorehabil Neural Repair.
2000;14(1):21–31.
37. O’Connor TJ, Fizgerald SG, Cooper RA, Thorman
TA, Boninger ML. Does computer game play aid
in motivation of exercise and increase metabolic
activity during wheelchair ergometry? Med Eng
Phys. 2001;23(4):267–273.
38. O’Connor TJ, Fizgerald SG, Cooper RA, Thorman
TA, Boninger ML. Kinetic and physiological analysis
of the GAME(Wheels) system. J Rehabil Res Dev.
2002;39(6):627–634.
39. Adamovich SV, Merians AS, Boian R, et al. A virtual
reality based exercise system for hand rehabilitation
post-stroke: transfer to function. Conf Proc IEEE Eng
Med Biol Soc. 2004;7:4936–4939.
40. Boian R, Sharma A, Han S, et al. Virtual reality-
based post-stroke hand rehabilitation. Stud Health
Technol Inform. 2002;85:64–70.
41. Brewer BR, Klatzky R, Matsuoka Y. Visual
feedback distortion in a robotic environment for
hand rehabilitation. Brain Res Bull. 2008;75(6):
804–813.
42. Szturm T, Peters JF, Otto C, Kapadia N, Desai A.
Task-specifi c rehabilitation of fi nger-hand function
using interactive computer gaming. Arch Phys Med
Rehabil. 2008;89(11):2213–2217.
43. Rizzo A, Requejo P, Winstein CJ, et al. Virtual reality
applications for addressing the needs of those
aging with disability. Stud Health Technol Inform.
2011;163:510–516.
44. Hesse S, Schmidt H, Werner C, Bardeleben A.
Upper and lower extremity robotic devices for
rehabilitation and for studying motor control. Curr
Opin Neurol. 2003;16(6):705–710.
patients with history of stroke. J Am Geriatr Soc.
2010;58(1):191–192.
15. Walker ML, Ringleb SI, Maihafer GC, et al. Virtual
reality-enhanced partial body weight-supported
treadmill training poststroke: feasibility and
effectiveness in 6 subjects. Arch Phys Med Rehabil.
2010;91(1):115–122.
16. Kim JH, Jang SH, Kim CS, Jung JH, You JH.
Use of virtual reality to enhance balance and
ambulation in chronic stroke: a double-blind,
randomized controlled study. Am J Phys Med
Rehabil. 2009;88(9):693–701.
17. Holden MK. Virtual environments for motor
rehabilitation: review. Cyberpsychol Behav. 2005;
8(3):187–211; discussion 212–219.
18. Buxbaum LJ, Palermo MA, Mastrogiovanni D,
et al. Assessment of spatial attention and neglect
with a virtual wheelchair navigation task. J Clin Exp
Neuropsychol. 2008;30(6):650–660.
19. Kizony R, Levin MF, Hughey L, Perez C, Fung
J. Cognitive load and dual-task performance
during locomotion poststroke: a feasibility study
using a functional virtual environment. Phys Ther.
2010;90(2):252–260.
20. Kang YJ, Ku J, Han K, et al. Development and
clinical trial of virtual reality-based cognitive
assessment in people with stroke: preliminary
study. Cyberpsychol Behav. 2008;11(3):329–339.
21. Gaggioli A, Meneghini A, Morganti F, Alcaniz M,
Riva G. A strategy for computer-assisted mental
practice in stroke rehabilitation. Neurorehabil
Neural Repair. 2006;20(4):503–507.
22. Gaggioli A, Morganti F, Meneghini A, et al.
Computer-guided mental practice in
neurorehabilitation. Stud Health Technol Inform.
2009;145:195–208.
23. Yip BC, Man DW. Virtual reality (VR)-based
community living skills training for people with
acquired brain injury: a pilot study. Brain Inj.
2009;23(13–14):1017–1026.
24. Broeren J, Claesson L, Goude D, Rydmark M,
Sunnerhagen KS. Virtual rehabilitation in an activity
centre for community-dwelling persons with
stroke. The possibilities of 3-dimensional computer
games. Cerebrovasc Dis. 2008;26(3):289–296.
25. Yang YR, Tsai MP, Chuang TY, Sung WH, Wang RY.
Virtual reality-based training improves community
ambulation in individuals with stroke: a randomized
controlled trial. Gait Posture. 2008;28(2):
201–206.
26. Fordell H, Bodin K, Bucht G, Malm J. A virtual reality
test battery for assessment and screening of spatial
neglect. Acta Neurol Scand. 2011;123(3):167–174.
27. Tsirlin I, Dupierrix E, Chokron S, Coquillart S,
Ohlmann T. Uses of virtual reality for diagnosis,
rehabilitation and study of unilateral spatial
neglect: review and analysis. Cyberpsychol Behav.
2009;12(2):175–181.
28. Kim J, Kim K, Kim DY, et al. Virtual environment
training system for rehabilitation of stroke patients
with unilateral neglect: crossing the virtual street.
Cyberpsychol Behav. 2007;10(1):7–15.
698 TOPICS IN STROKE REHABILITATION/NOV-DEC 2011
61. Hidler J, Neckel N. Inverse-dynamics based
assessment of gait using a robotic orthosis. Conf
Proc IEEE Eng Med Biol Soc. 2006;1:185–188.
62. Colombo R, Sterpi O, Mazzone A, Delconte
C, Minuco G, Pisano F. Measuring changes
of movement dynamics during robot-aided
neurorehabilitation of stroke patients. IEEE Trans
Neural Syst Rehabil Eng. 2010;18(1):75–85.
63. Sanguineti V, Casadio, M, Vergaro E, Squeri
V, Giannoni P, Morasso PG. Robot therapy for
stroke survivors: proprioceptive training and
regulation of assistance. Stud Health Technol
Inform. 2009;145:126–142.
64. Casadio M, Giannoni P, Masia L, et al. Robot
therapy of the upper limb in stroke patients:
preliminary experiences for the principle-based
use of this technology. Funct Neurol. 2009;24(4):
195–202.
65. Kutner NG, Zhang R, Butler AJ, Wolf SL, Alberts
JL. Quality-of-life change associated with robotic-
assisted therapy to improve hand motor function
in patients with subacute stroke: a randomized
clinical trial. Phys Ther. 2010;90(4):493–504.
66. Kanamori M, Suzuki M, Tanaka M. [Maintenance
and improvement of quality of life among elderly
patients using a pet-type robot]. Nippon Ronen
Igakkai Zasshi. 2002;39(2):214–218 [in Japanese].
67. Brewer BR, McDowell SK, Worthen-Chaudhari LC.
Poststroke upper extremity rehabilitation: a review
of robotic systems and clinical results. Top Stroke
Rehabil. 2007;14(6):22–44.
68. Kwakkel G, Kollen BJ, Krebs HI. Effects of robot-
assisted therapy on upper limb recovery after
stroke: a systematic review. Neurorehabil Neural
Repair. 2008;22(2):111–121.
69. Teasell RW, Kalra L. What’s new in stroke
rehabilitation. Stroke. 2004;35(2):383–385.
70. Westlake KP, Patten C. Pilot study of Lokomat
versus manual-assisted treadmill training for
locomotor recovery post-stroke. J Neuroeng
Rehabil. 2009;6:18.
71. Husemann B, Muller F, Krewer C, Heller S, Koenig
E. Effects of locomotion training with assistance of
a robot-driven gait orthosis in hemiparetic patients
after stroke: a randomized controlled pilot study.
Stroke. 2007;38(2):349–354.
72. Hidler J, Nichols D, Pelliccio M, et al. Multicenter
randomized clinical trial evaluating the effectiveness
of the Lokomat in subacute stroke. Neurorehabil
Neural Repair. 2009;23(1):5–13.
73. Banala SK, Kim SH, Agrawal SK, Scholz JP. Robot
assisted gait training with active leg exoskeleton
(ALEX). IEEE Trans Neural Syst Rehabil Eng.
2009;17(1):2–8.
74. Krebs HI, Volpe BT, Ferraro M, et al. Robot-aided
neurorehabilitation: from evidence-based to
science-based rehabilitation. Top Stroke Rehabil.
2002;8(4):54–70.
75. Reinkensmeyer DJ, Kahn LE, Averbuch M, McKenna-
Cole A, Schmit BD, Rymer WZ. Understanding and
treating arm movement impairment after chronic
brain injury: progress with the ARM guide. J Rehabil
Res Dev. 2000;37(6):653–662.
45. Kakou H, Shitama H, Kimura Y, et al. [Walking
assist robot and its clinical application]. J UOEH.
2009;31(2):207–218. [in Japanese]
46. Lewek MD. The value of overground gait training for
improving locomotion in individuals with chronic
stroke. J Neurol Phys Ther. 2009;33(4):187–188.
47. Merians AS, Tunik E, Fluet GG, Qiu Q, Adamovich SV.
Innovative approaches to the rehabilitation of upper
extremity hemiparesis using virtual environments.
Eur J Phys Rehabil Med. 2009;45(1):123–133.
48. Edwards DJ, Krebs HI, Rykman A, et al. Raised
corticomotor excitability of M1 forearm area
following anodal tDCS is sustained during robotic
wrist therapy in chronic stroke. Restor Neurol
Neurosci. 2009;27(3):199–207.
49. Mirelman A, Bonato P, Deutsch JE. Effects of training
with a robot-virtual reality system compared with a
robot alone on the gait of individuals after stroke.
Stroke. 2009;40(1):169–174.
50. Neckel ND, Bionien N, Nichols D, Hidler J
Abnormal joint torque patterns exhibited by
chronic stroke subjects while walking with a
prescribed physiological gait pattern. J Neuroeng
Rehabil. 2008;5:19.
51. Schabowsky CN, Godfrey SB, Holley RJ, Lum PS.
Development and pilot testing of HEXORR: Hand
EXOskeleton Rehabilitation Robot. J Neuroeng
Rehabil. 2010;7:36.
52. Cruz EG, Kamper DG. Use of a novel robotic
interface to study fi nger motor control. Ann Biomed
Eng. 2010;38(2):259–268.
53. Miller LC, Ruiz-Torres R, Stienen AH, Dewald JP.
A wrist and fi nger force sensor module for use
during movements of the upper limb in chronic
hemiparetic stroke. IEEE Trans Biomed Eng.
2009;56(9):2312–2317.
54. Rosati G, Gallina P, Masiero S. Design,
implementation and clinical tests of a wire-based
robot for neurorehabilitation. IEEE Trans Neural Syst
Rehabil Eng. 2007;15(4):560–569.
55. Krebs HI, Volpe BT, Williams D, et al. Robot-aided
neurorehabilitation: a robot for wrist rehabilitation.
IEEE Trans Neural Syst Rehabil Eng. 2007;15(3):
327–335.
56. Khanna I, Roy A, Rodgers MM, Krebs HI, Macko
RM, Forrester LW. Effects of unilateral robotic limb
loading on gait characteristics in subjects with
chronic stroke. J Neuroeng Rehabil. 2010;7:23.
57. Kim SH, Banala SK, Brackbill EA, Agrawal SW,
Krishnamoorthy V, Scholz JP. Robot-assisted
modifi cations of gait in healthy individuals. Exp
Brain Res. 2010;202(4):809–824.
58. Dickstein R. Rehabilitation of gait speed after
stroke: a critical review of intervention approaches.
Neurorehabil Neural Repair. 2008;22(6):649–660.
59. Vergaro E, Casadio M, Squeri V, Giannoni P,
Morasso P, Sanquineti V. Self-adaptive robot
training of stroke survivors for continuous tracking
movements. J Neuroeng Rehabil. 2010;7:13.
60. Dukelow SP, Herter TM, Moore KD, et al.
Quantitative assessment of limb position sense
following stroke. Neurorehabil Neural Repair.
2010;24(2):178–187.
Virtual Reality and Robotics 699
92. Wolf SL, Caitlin PA, Ellis M, Archer AL, Morgan B,
Piacentino A. Assessing Wolf Motor Function Test
as outcome measure for research in patients after
stroke. Stroke. 2001;32(7):1635–1639.
93. Blanton S, Morris DM, Prettyman MG, et al. Lessons
learned in participant recruitment and retention: the
EXCITE trial. Phys Ther. 2006;86(11):1520–1533.
94. Wolf SL, Winstein CJ, Miller JP, et al. Retention of
upper limb function in stroke survivors who have
received constraint-induced movement therapy:
the EXCITE randomised trial. Lancet Neurol.
2008;7(1):33–40.
95. Mazzoleni S, Toth A, Munih M, et al. Whole-
body isometric force/torque measurements for
functional assessment in neuro-rehabilitation:
platform design, development and verifi cation.
J Neuroeng Rehabil. 2009;6:38.
96. Balasubramanian S, Wei R, Herman R, He J.
Robot-measured performance metrics in stroke
rehabilitation. In: 2009 ICME International
Conference on Complex Medical Engineering;
2009:330–335.
97. Ball SJ, Brown IE, Scott SH. A planar 3DOF robotic
exoskeleton for rehabilitation and assessment.
In: 2007 Annual International Conference of the
IEEE Engineering in Medicine and Biology Society.
2007;1–16:4024–4027.
98. Van Dijck G, Van Vaerenbergh J, Van Hulle MM.
Posterior probability profi les for the automated
assessment of the recovery of patients with stroke
from activity of daily living tasks. Artif Intell Med.
2009;46(3):233–249.
99. Colombo R, Pisano F, Micera S, et al. Robotic
techniques for upper limb evaluation and
rehabilitation of stroke patients. IEEE Trans Neural
Syst Rehabil Eng. 2005;13(3):311–324.
100. Leibowitz N, Levy N, Weingarten S, et al. Automated
measurement of proprioception following stroke.
Disabil Rehabil. 2008;30(24):1829–1836.
101. Chen SY, Han CE, Parikh N, et al. BART: a novel
instrument to measure spontaneous arm use
in patients after stroke. Paper presented at:
Neuroscience 2008; November 15–19, 2008;
Washington, DC.
102. Stewart J. Planning of unconstrained reach actions
after unilateral sensorimotor stroke. Doctoral
dissertation. Division of Biokinesiology and
Physical Therapy, University of Southern California,
Los Angeles; 2010.
103. Knorr B, Hughes R, Sherrill D, Stein J, Akay M,
Bonato P. Quantitative measures of functional
upper limb movement in persons after stroke.
In: 2nd International IEEE/EMBS Conference on
Neural Engineering; 2005:252–255.
104. Parnandi AR, Wade E, Matari MJ. Motor function
assessment using wearable inertial sensors.
Presented at: 32nd Annual International
Conference of the IEEE Engineering in Medicine
and Biology Society (EMBC 2010); 2010; Buenos
Aires, Argentina.
105. Wade E, Parnandi AR, Matari ĆMJ. Automated
administration of the Wolf Motor Function Test
for post-stroke assessment. Presented at: ICST
76. Kahn, LE, et al. Comparison of robot-assisted
reaching to free reaching in promoting recovery
from chronic stroke. Integr Assist Technol Information
Age. 2001;9:39–44.
77. Krebs HI, Hogan N, Aisen ML, Volpe BT. Robot-
aided neurorehabilitation. IEEE Trans Rehabil Eng.
1998;6(1):75–87.
78. Lum P, Reinkensmeyer D, Mahoney R, Rymer WZ,
Burgar C. Robotic devices for movement therapy
after stroke: current status and challenges to clinical
acceptance. Top Stroke Rehabil. 2002;8(4):40–53.
79. Fasoli SE, Krebs HI, Stein J, Frontera WR, Hogan N.
Effects of robotic therapy on motor impairment
and recovery in chronic stroke. Arch Phys Med
Rehabil. 2003;84(4):477–482.
80. Fasoli SE, Fragala-Pinkham M, Krebs HI, Hogan
N, Stein J. Robotic therapy for chronic motor
impairments after stroke: follow-up results. Arch
Phys Med Rehabil. 2004;85(7):1106–1111.
81. Fasoli SE, Krebs HI, Ferraro M, Hogan N, Volpe
BT. Does shorter rehabilitation limit potential
recovery poststroke? Neurorehabil Neural Repair.
2004;18(2):88–94.
82. Mahoney, R.M., Van der Loos HFM, Lum PS, Burgar
C. Robotic stroke therapy assistant. Robotica.
2003;21:33–44.
83. Stein J, Krebs HI, Frontera WR, Fasoli SE, Hughes R,
Hogan N. Comparison of two techniques of robot-
aided upper limb exercise training after stroke. Am
J Phys Med Rehabil. 2004;83(9):720–728.
84. Lum PS, Burgar CG, Shor PC, Majmundar M,
Van der Loos M. Robot-assisted movement
training compared with conventional therapy
techniques for the rehabilitation of upper-limb
motor function after stroke. Arch Phys Med Rehabil.
2002;83(7):952–959.
85. Choi Y, Qi F, Gordon J, Schweighofer N. An
adaptive automated robotic task-practice system
for rehabilitation of arm functions after stroke. IEEE
Trans Robotics. 2009;25(3):556–568.
86. Lo AC, Guarino PD, Richards LG, et al. Robot-assisted
therapy for long-term upper-limb impairment after
stroke. N Engl J Med. 2010;362(19):1772–1783.
87. Matari Ć MJ, Eriksson J, Feil-Seifer DJ, Winstein
CJ. Socially assistive robotics for post-stroke
rehabilitation. J Neuroeng Rehabil. 2007;4:5.
88. Eriksson J, Matari Ć MJ, Winstein CJ. Hands-off
assistive robotics for post-stroke arm rehabilitation.
In: IEEE 9th International Conference on
Rehabilitation Robotics; 2005:21–24.
89. Mead R, et al. An architecture for rehabilitation
task practice in socially assistive human-robot
interaction. In: IEEE International Symposium in
Robot and Human Interactive Communication
(Ro-Man 2010); 2010; Viareggio, Italy.
90. Wade E, et al., Assessing the quality and quantity of
social interaction in a socially assistive robot-guided
therapeutic setting. In: International Conference on
Rehabilitation Robotics; 2011; Zurich.
91. Sanford J, Moreland J, Swanson LR, Stratford PW,
Gowland C. Reliability of the Fugl-Meyer assessment
for testing motor-performance in patients
following stroke. Phys Ther. 1993;73(7):447–454.
700 TOPICS IN STROKE REHABILITATION/NOV-DEC 2011
and alteration of diabetes patients. J Diabetes Sci
Technol, 2009;3(1):141–148.
119. Junnila S, Kailanto H, Merilahti J, et al. Wireless,
multipurpose in-home health monitoring platform:
two case trials. IEEE Trans Inf Technol Biomed.
2010;14(2):447–455.
120. Marschollek M, et al. Multimodal home monitoring
of elderly people—fi rst results from the LASS study.
In: 21st International Conference on Advanced
Networking and Applications Workshops/
Symposia, Proceedings. 2007;2:815–819.
121. Medjahed H, Istrate D, Broudy J, Dorizzi B. Human
activities of daily living recognition using fuzzy
logic for elderly home monitoring. In: 2009 IEEE
International Conference on Fuzzy Systems.
2009;1–3:2001–2006.
122. Gourlay D, Lun KC, Liya G. Virtual reality and
telemedicine for home health care. Computers &
Graphics-UK. 2000;24(5):695–699.
123. Uswatte G, Taub E, Morris D, Light K, Thompson
PA. The Motor Activity Log-28: assessing daily use
of the hemiparetic arm after stroke. Neurology.
2006;67(7):1189–1194.
124. Boyd LA, Winstein CJ. Impact of explicit
information on implicit motor-sequence learning
following middle cerebral artery stroke. Phys Ther.
2003;83(11):976–989.
125. Wulf G, Shea C, Park JH. Attention and motor
performance: preferences for and advantages of an
external focus. Res Q Exerc Sport. 2001;72(4):335–344.
126. Kidd C, Breazeal C. A robotic weight loss coach.
In: 22nd Conference on Artificial Intelligence.
Vancouver, British Columbia: AAAI; 2007.
127. Bickmore T, Gruber A. Relational agents in clinical
psychiatry. Harv Rev Psychiatry. 2010;18(2):119–130.
128. Breazeal C. Role of expressive behaviour for robots
that learn from people. Philos Trans Rl Soc Lond B
Biol Sci. 2009;364(1535):3527–3538.
129. Campbell M, Fitzpatrick R, Haines A, et al.
Framework for design and evaluation of
complex interventions to improve health. BMJ.
2000;321(7262):694–696.
130. Gaggioli, A, Kesher EA, Weiss PL, Riva G, eds.
Advanced technologies in rehabilitation. In: Studies
in Health Technology and Informatics. IOS Press:
2009;145.
131. Kantak SS, Sullivan KJ, Fisher BE, Knowlton BJ,
Winstein CJ. Neural substrates of motor memory
consolidation depend on practice structure. Nat
Neurosci. 2010; 13(8):923–925.
4th International ICST Conference on Pervasive
Computing Technologies for Healthcare 2010;
2010; Munich, Germany.
106. Riva G, Gamberini L. Virtual reality in telemedicine.
Telemed J E Health. 2000;6(3):327–340.
107. Pugnetti, L, Meehan M, Mendozzi L.
Psychophysiological correlates of virtual reality:
a review. Presence Teleoper Virtual Environ.
2001;10(4):384–400.
108. Grimes GJ, Dubois H, Grimes SJ, Greenleaf WJ,
Rothenburg S, Cunningham D. Telerehabilitation
services using web-based telecommunication. Stud
Health Technol Inform. 2000;70:113–118.
109. Rovetta A, Lorini F, Canina M. A new project for
rehabilitation and psychomotor disease analysis
with virtual reality support. Stud Health Technol
Inform.1998;50:180–185.
110. Schmidt RA, Lee TD. Motor Control and Learning :
A Behavioral Emphasis. 4th ed. Champaign, IL:
Human Kinetics; 2005:vi, 537.
111. Cahill L, Gorski L, Le K. Enhanced human memory
consolidation with post-learning stress: interaction
with the degree of arousal at encoding. Learn
Mem. 2003;10(4):270–274.
112. Pattichis CS, Kyriacou E, Voskarides S, Pattichis MS,
Istepanian RSH, Schizas CN. Wireless telemedicine
systems: an overview. IEEE Antennas Propagation
Mag. 2002;44(2):143–153.
113. Mehairi SO, Barada H, Qutayri M. Integration
of technologies for smart home application.
In: 2007 IEEE/ACS International Conference on
Computer Systems and Applications. 2007;1,
2:241–246.
114. Gatzoulis L, Iakovidis I. Wearable and portable
eHealth systems. IEEE Eng Med Biol Mag.
2007;26(5):51–56.
115. Gaddam A, Mukhopadhyay SC, Sen Gupta
G, Guesgen H. Wireless sensors networks
based monitoring: review, challenges and
implementation issues. In: Proceedings of the Third
International Conference on Sensing Technology;
2008:533–538.
116. Roy SH, Cheng MS, Chang SS, et al. A combined
sEMG and accelerometer system for monitoring
functional activity in stroke. IEEE Trans Neural Syst
Rehabil Eng. 2009;17(6):585–594.
117. Mark T. Wearable health monitors: real-time,
patient-friendly data collection. Telemed J E Health.
2010;16(2):134–138.
118. Helal A, Cook DJ, Schmalz M. Smart home-
based health platform for behavioral monitoring
