CASE REPORT

Gait Training After Stroke: A Pilot Study Combining a
Gravity-Balanced Orthosis, Functional Electrical Stimulation,
and Visual Feedback
Vijaya Krishnamoorthy, PT, PhD, Wei-Li Hsu, PT, PhD, Trisha M. Kesar, MS, PT,
Daniel L. Benoit, PhD, Sai K. Banala, PhD, Ramu Perumal, PhD, Vivek Sangwan, BTech,
Stuart A. Binder-Macleod, PT, PhD, FAPTA, Sunil K. Agrawal, PhD, and John P. Scholz, PT, PhD

Rationale: This case report describes the application of a novel gait
retraining approach to an individual with poststroke hemiparesis.
The rehabilitation protocol combined a specially designed leg orthosis (the gravity-balanced orthosis), treadmill walking, and functional electrical stimulation to the ankle muscles with the application
of motor learning principles.
Case: The participant was a 58-year-old man who had a stroke more
than three years before the intervention. He underwent gait retraining over a period of five weeks for a total of 15 sessions during
which the gravity compensation provided by the gravity-balanced
orthosis and visual feedback about walking performance was gradually reduced.
Outcomes: At the end of five weeks, he decreased the time required
to complete the Timed Up and Go test; his gait speed increased
during overground walking; gait was more symmetrical; stride
length, hip and knee joint excursions on the affected side increased.
Except for gait symmetry, all other improvements were maintained
one month post-intervention.
Conclusions: This case report describes possible advantages of
judiciously combining different treatment techniques in improving
the gait of chronic stroke survivors. Further studies are planned to
evaluate the effectiveness of different components of this training in
both the subacute and chronic stages of stroke recovery.

Physical Therapy Department (V.K., W.-L.H., T.M.K., R.P., S.A.B.-M.,
J.P.S.), Mechanical Engineering Department (D.L.B., S.K.B., V.S.,
S.K.A.), Biomechanics and Movement Science Program (W.-L.H.,
T.M.K., S.A.B.-M., S.K.A., J.P.S.), and Center for Biomedical Engineering Research (R.P., S.A.B.-M., S.K.A., J.P.S.), University of Delaware,
Newark, Delaware.
Vijaya Krishnamoorthy is currently at the Division of Physical Therapy,
Department of Rehabilitation Medicine, Emory University, Atlanta,
Georgia.
Daniel L. Benoit is currently at the School of Rehabilitation Sciences,
University of Ottawa, Ottawa, ON, Canada.
Wei-Li Hsu is currently at the Institute of Neuroscience, University of
Oregon, Eugene, Oregon.
Address correspondence to: John P. Scholz, E-mail: jpscholz@udel.edu
Copyright © 2008 Neurology Section, APTA
ISSN: 1557-0576/08/3204-0192
DOI: 10.1097/NPT.0b013e31818e8fc2

192

Supplemental information (videos) for this article can be found
at www.jnptextra.org
Key words: gait, walking, recovery of function, robotic training,
physical therapy, rehabilitation, stroke
(JNPT 2008;32: 192–202)

INTRODUCTION

S

troke is a leading cause of disability among adults in the
United States.1 Individuals who have had a stroke typically undergo intensive gait retraining in the first few months
after the stroke. However, even those stroke survivors that are
able to walk independently after a course of physical therapy
frequently have residual gait deficits, and many individuals
never achieve community ambulation status.2 Individuals
with hemiparesis walk slower and have a slower cadence and
shorter stride length compared with nonimpaired control
subjects.3,4 These individuals tend to spend more time in the
double support phase of the gait cycle and less time in stance
on the affected limb. Such deficits result in inefficient gait
patterns that limit an individual’s independence at home and
in the community5,6 and predispose them to secondary problems such as injuries from falls.7,8
Recent studies have emphasized the need for intensive
walking practice after a stroke.9,10 However, supervised intensive gait training is usually too expensive to be practical.
Moreover, in the acute and subacute phases after a stroke,
trunk instability and poor balance may make walking safety
a particular concern. Recent studies have tested a variety of
gait retraining interventions intended to provide the benefits
of safe, intensive walking practice to participants with stroke.
These include body weight–supported training,11 motor imagery,12 gait training with a robot-driven active gait orthosis,13 and a mechanized gait trainer.14,15
In a recent editorial, Whitall16 suggested that individuals with a chronic stroke (more than nine months) benefit
from further training and future research on stroke rehabilitation should focus on specific questions that maximize training benefits leading to efficient and effective rehabilitation
strategies. Some of the directions suggested in this editorial
were combining different training programs and targeting
JNPT • Volume 32, December 2008

JNPT • Volume 32, December 2008

specific programs toward particular patient groups. Only a
few studies have evaluated the effects of combining interventions such as functional electrical stimulation (FES) and the
mechanized gait trainer,17,18 virtual reality coupled with robotic therapy,19 and robotic therapy along with application of
motor learning principles.20 These studies suggest that combining different treatment strategies could make poststroke
rehabilitation more efficient and effective.
In this article, we describe the findings of an initial case
report to test the potential usefulness of the combination of
training with a specially designed gravity-balanced orthosis
(GBO), treadmill walking, FES, and principles of motor
learning. Each component had a specific role in the training
protocol that is described below.

GRAVITY-BALANCED ORTHOSIS
Robotic devices that are currently available for gait
training mostly move the patient’s limb through a prescribed
path. This allows the patient to experience a “normal” pattern
of movement without having to generate the movement on his
or her own. The GBO, on the other hand, allows the patient
to produce movement on his or her own while making the
motion easier by reducing/eliminating the effect of gravity on
the limb.
The GBO was designed to primarily assist motion at the
hip and knee joints by fully or partially compensating for the

Gait Training After Stroke

weight of the limb segments. Although gravity plays less of
a role than dynamic forces during fast walking, its contribution is more important at the walking speeds produced by
most individuals after a stroke.5 Conceptually, removing the
effects of gravity early in rehabilitation may allow patients to
activate weakened muscles while they produce more normal
walking patterns, limiting the typical compensations.
Details about the GBO design and applications have
been reported elsewhere.21,22 Briefly, the GBO is designed to
assist persons with hemiparesis to walk by using a specially
designed spring configuration that can completely or partially
reduce the effect of gravity (Figure 1). The principle involved
is as follows: (1) the center of mass of the subject’s lower
limb and the attached orthosis is geometrically located using
a parallelogram mechanism and (2) springs are placed at
suitable positions so that they completely or partially balance
the effect of gravity over the entire range of joint motion. The
device can be adjusted to the geometry and inertia of each
subject to achieve the desired level of gravity balancing
throughout the range of motion. A key feature of the device
is that it is passive, eliminating the need for motors. Benefits
of such a system are reduced safety concerns and drastically
reduced equipment costs. From a rehabilitation perspective,
the device is entirely patient driven and serves only to
supplement or facilitate the movements initiated by the user.

FIGURE 1. The gravity-balanced orthosis (GBO). A. A subject is shown at rest in the gravity balanced orthosis, while standing
on a split-belt treadmill. Note the attachment of the orthosis to the large frame at the back of the subject and the treadmill.
B. Participant walking on a treadmill with the gravity-balanced orthosis during a training session. For the video of the GBO
go to jnptextra.org.
© 2008 Neurology Section, APTA

193

JNPT • Volume 32, December 2008

Krishnamoorthy et al

Because the GBO provided complete or partial elimination of the influence of gravity on the participant’s paretic
limb, thereby reducing the amount of muscle activation
required to attain adequate joint excursions,22,23 more extended training sessions were possible than for unassisted
walking. Moreover, combining this assistance with treadmill
walking allows for a high number of repetitions in a functional context, known to be required for plastic brain changes
to occur.24 Continuous treadmill walking can also assist
cardiovascular fitness in chronic stroke survivors, which is an
added benefit of the training.25
With improvements in walking performance, the
amount of gravity balancing could be gradually reduced or
eliminated altogether, which is an ultimate goal along with
improved independent overground walking.

FES
FES has been applied to assist functional movements
such as standing,26,27 walking,28 –30 and grasping.31 In individuals with hemiparesis after stroke, stimulation of ankle
dorsiflexors for the treatment of footdrop is the most common
application of FES.32– 40 In contrast to most studies that use
FES for gait training, we delivered FES not only to the ankle
dorsiflexor muscles for swing phase training but also to the
plantarflexor muscles to assist with push-off and the initial
swing phase of knee flexion.

Feedback for Motor Learning
Provision of adequate feedback is essential for motor
learning.41 During training sessions, a video monitor was
placed in front of the participant at eye level to provide online
knowledge of performance. Feedback consisted of real-time
plots of the relative motion between the hip and knee angle
throughout the gait cycle or the sagittal plane foot position
(ie, instantaneous anterior-posterior vs vertical position plot).
The “desired” or “ideal” joint angle motion or footpath was
also displayed for the subject. Initially, this allowed the
participant to visualize in real time his deviations from the
desired gait pattern. However, during training sessions, this
feedback was gradually withdrawn as continuous concurrent
feedback is known to degrade learning.42 Instead, summary
feedback was provided after a trial of multiple gait cycles (see
Methods section).
In the current approach, a team of professionals, including physical therapists, a biomechanist, and engineers worked
together to design and build the GBO and FES system used
for training. The physical therapists took the lead in designing
and implementing the training, whereas the engineers took
the lead in developing and assembling the hardware and
software used during training. Our goal in combining GBO,
FES, and motor learning principles was to determine whether
the patient could achieve an increased stride length on the
hemiparetic side, improved gait symmetry, increased motion
of both hip and knee joints, faster gait speed, and increased
weight bearing on his impaired right side.

194

CASE REPORT
Participant
The participant in this case report was a 58-year-old
man who was left-hand dominant before a left middle cerebral artery infarction. He had a single stroke three years and
five months before this training. He gave informed consent to
participate in this training based on procedures approved by
the Human Subjects Review Board at the University of
Delaware.
The participant’s stroke led to infarcts in the left basal
ganglia, internal capsule, corona radiata, and posterior left
temporoparietal lobe. Immediately after the stroke, he spent
five days in acute care followed by three weeks of rehabilitation including physical, occupational, and speech therapy.
He continued physical therapy on an out-participant basis two
to three times a week for the first year after the stroke, with
some breaks in between. For three years after the stroke, the
participant also volunteered in the laboratory section of a
neurological rehabilitation class for physical therapy students, where he received additional motor training once per
week for 10 weeks each fall semester.
Before his stroke, the participant held a full-time engineering job. His height and weight were 1.75 m and 74.8 kg,
respectively. At the time of this report, he had retired and was
no longer undergoing therapy. His exercise activity included
regular walks in the neighborhood around his home without
the assistance of any orthotic device or a walking aid. He
reported being independent in all activities of daily living.
However, observational analysis revealed a gait that was
slower than normal for his age and asymmetrical. He had
difficulty clearing the ground with the affected foot, and used
compensatory patterns such as hip elevation and circumduction. Because of the increased effort required to walk, he
reported limiting his walking to less than he would like. The
participant’s goal from this intervention was to improve his
ability to walk on uneven outdoor surfaces so that he could
take longer walks for exercise.

Baseline Examination
Clinical Assessment
A detailed baseline clinical assessment was performed
before the start of training. The participant had good sensation on the affected side and no major cognitive deficits. He
had normal perceptual abilities as determined by the motorfree visual perception test.43 His performance score on other
tests performed at baseline are listed in Table 1. The MiniMental State Examination44 and the Modified Ashworth
TABLE 1. Scores of Evaluations Administered Before
Training
Evaluations

Scores

Mini-Mental State Examination score
Modified Ashworth Scale: ankle
Timed Up and Go
Modified Fugl-Meyer, lower extremity portion
Berg Balance Scale

30/30
1⫹
15 sec
28/36
56/56

© 2008 Neurology Section, APTA

JNPT • Volume 32, December 2008

Scale45 were administered only at baseline assessment. The
Berg Balance Scale (BBS),46 lower extremity portion of the
modified Fugl-Meyer assessment,47 and the Timed Up and
Go test48 were repeated at the end of training.

Gait Analysis
Gait analysis was performed before training (pretraining), midway through training (midtraining), at the end of
training (post-training), and one month after training ended
(follow-up). Gait analysis was performed both during overground walking (without the GBO) and during walking on the
treadmill (both with and without the GBO). This report
focuses on the overground evaluation without the GBO because of its greater functional significance.
Gait data were collected at 120 Hz using a six-camera
Motion Analysis Corporation system and processed using an
automated real-time motion analyses system (Eva real-time
version 4.3). A modified version of the Helen Hayes marker
set consisting of 23 tracking retroreflective markers was used
during testing. Ground reaction force data were collected
using two force plates embedded in an instrumented custombuilt split-belt treadmill (Bertec Corporation, Columbus,
OH), both during treadmill and overground walking. Analog
data were collected at 2400 Hz, and kinetic data were normalized to body mass. Foot switches were placed under the
heel and the forefoot to capture heel strike and toe-off during
walking.
For the overground walking trials, the participant
walked at his comfortable walking speed over a walkway that
included the force plates embedded in the treadmill that was
at the center of the walkway. The trials that had good force
plate contact by at least one foot were noted and used for
further processing. Five “good” trials with force plate contact
by each foot were collected. Each of these trials included two
to three complete gait cycles, although ground reaction force
information was limited to one cycle.

Gait Training After Stroke

plantarflexor muscles. First, a twitch with 600-␮sec pulse
duration and 135-V amplitude was delivered to the muscle as
the participant relaxed and the peak force produced recorded
as the resting twitch force. Next, the participant volitionally
contracted his plantarflexor muscles to generate a maximum
voluntary isometric contraction (MVIC). During the MVIC,
another twitch with the same pulse duration and amplitude as
the resting twitch was delivered to the muscle. The increment
in the MVIC force produced by the superimposed twitch was
recorded as the interpolated twitch force. The resting and
interpolated twitch forces were used to obtain a measure of
volitional activation, the central activation ratio (CAR), as
follows:

[

CAR ⫽ 1 ⫺

Interpolated twitch amplitude
Resting twitch amplitude

]

Also, the maximum voluntary isometric force was divided by the CAR to obtain the estimated maximum force
generating ability of the muscle.

Intervention
The participant trained three days per week for five
weeks. Each session lasted approximately two hours, including setup time and rest periods. The participant walked about
40 minutes during each session, which included four 10minute blocks of treadmill walking. After each block, the
participant was allowed to rest while sitting on a stool and his
blood pressure was monitored.

GBO
During the training sessions, the participant walked on
a treadmill (TR 2500-HRC, LifeSpan Fitness, Park City, UT)
with the GBO. The GBO was attached to the participant by

Isometric Muscle Performance Testing
The participant also underwent isometric muscle performance testing before and after training. The ankle dorsiflexor and plantarflexor muscles were evaluated with the
subject positioned on an isokinetic dynamometer (KinCom
III 500-11, Chattecx Corporation, Chattanooga, TN) with the
hips and knees extended. The axis of the ankle joint was
aligned with the axis of the force transducer. The ankle was
positioned in 15 degrees of plantarflexion for testing the
dorsiflexors and at 0 degree (neutral ankle position) for
testing the plantarflexors. The distal portion of the foot, the
distal and proximal portions of the leg, and the distal portion
of the subject’s thigh were stabilized using inelastic Velcro
straps (Figure 2). Self-adhesive surface electrodes were attached on the skin over the dorsiflexor and plantarflexor
muscle groups. Details about electrode placement are provided in the “Setup for FES” subsection. After ensuring
appropriate electrode placement, the twitch interpolation
technique49,50 was used to estimate the degree of central
activation. Because the participant was unable to produce a
measurable volitional isometric force from the dorsiflexors,
the twitch interpolation technique was tested only on the
© 2008 Neurology Section, APTA

FIGURE 2. Isometric ankle muscle performance testing.
Testing position for the dorsiflexor and plantarflexor muscles
on the KinCom apparatus. The subject was seated with the
hip and knee extended and the ankle at 15 degrees of plantarflexion for testing the dorsiflexors and at 0 degree (neutral ankle position) for testing the plantarflexors.

195

JNPT • Volume 32, December 2008

Krishnamoorthy et al

cuffs strapped to the pelvis, thigh, and lower leg, and by an
insert that fits into the user’s shoe (Figure 1). The GBO is
mounted on a walking frame behind the treadmill. A harness
was attached overhead to the walking frame to prevent falling
should the patient lose his or her balance. A large metal plate
connected the pelvic band to the walking frame to counter the
forces generated by the patient’s movements. The participant’s right (affected) arm was suspended in a sling because
there was not adequate room for him to swing the arm during
walking. However, the participant showed very little arm
swing during overground walking, as is often the case with
individuals after a stroke. The participant held the treadmill’s
handrails only when the treadmill speed was changed.

Setup for FES
Both dorsiflexor and plantarflexor muscle groups were
stimulated using a real-time FES system consisting of a
real-time controller (NI cRIO-9004), analog input module
(NI9210), digital input/output module (NI9401), and Grass
S8800 stimulator. Foot switches placed on the heel and toe of
each foot were used to detect the different phases of gait
during treadmill walking. Custom written software (LabVIEW, National Instruments, Austin, TX) was used to acquire the foot-switch data. Electrical stimulation was delivered via self-adhesive electrodes (TENS Products, Grand
Lake, CO; two ⫻ two–inch square foam for dorsiflexors;
three-inch Round Tricot for plantarflexors). For the dorsiflexors, the cathode electrode was placed over the motor point of
the tibialis anterior (TA) and the anode was placed over the
muscle belly at the distal portion of the anterolateral aspect of
the leg. TA electrode placement was adjusted to ensure that
negligible eversion/inversion ankle movements were produced during stimulation. For the plantarflexors, the cathode
was placed over both the medial and lateral heads of the
gastrocnemius (MG-LG); the anode was placed over the
distal portion of the gastrocnemius muscle belly. Stimulation
amplitude for the TA was set with the participant seated with
his leg suspended so that the ankle was hanging freely in a
plantar-flexed position. The train duration was set at 300
msec and the pulse duration at 300 ␮sec. The stimulation
train began with an initial triplet (ie, three pulses, each
separated by five milliseconds) and the remaining pulses were
separated by a 33-msec interpulse interval (30 Hz). The initial
high frequency burst was used to take advantage of the
catchlike property of the muscle.51 The stimulation amplitude
was then increased gradually until a dorsiflexion movement
to neutral ankle position (0 degree) was produced without
significant eversion and inversion. Stimulation amplitude for
MG-LG was set with the participant standing and bearing
weight on both limbs. Similar stimulation parameters to those
used for the TA were used, except the train frequency was
100 Hz and pulse duration was 600 ␮sec. The stimulation
amplitude was then increased until we observed the heel
lifting off the ground or the subject identified the maximum
tolerable amplitude. In the present report, the participant was
able to tolerate the levels of stimulation required to lift the
heel off the ground. Because the pulse amplitude was not
varied during the session and the frequency and pulse duration used to set the amplitude were the highest frequency and

196

pulse duration used for activation of either muscle during the
training sessions, we were assured that the subject would be
able to tolerate the stimulation. For both muscles, the 30-Hz
train frequency and 300-␮sec pulse duration were the initial
stimulation parameters. The stimulation frequency and intensity (via pulse duration) were increased in the training session
to generate the desired responses from both muscles and to
compensate for any fatigue that developed during the session.
The participant had no difficulty tolerating the stimulation
either when setting the amplitude or when the stimulation
parameters were changed during treadmill walking. The dorsiflexor muscle group was stimulated from toe-off to heel
strike of the paretic limb. Plantar-flexor stimulation was
timed to start at heel strike of the nonparetic limb, which
coincided with terminal stance for the paretic leg. The FES
system terminated plantarflexor stimulation when toe-off of
the paretic leg was detected.

Visual Feedback
In the current report, the desired motions were based on
patterns exhibited by five healthy volunteers walking in the
GBO with 0% balancing (ie, no assistance from the device).
The data were scaled to the participant’s segment lengths to
produce an appropriate foot trajectory template. The template
was used as a target that the participant attempted to match
during practice. Visual feedback was provided either in the
form of plots of hip angle versus knee angle motion (in
degrees) or of the sagittal plane foot motion (in centimeters)
of the hemiparetic leg. During early training, continuous
real-time visual feedback was provided to the participant.
Over training sessions, the continuous feedback was reduced
and gradually taken away. At the end of each block of
walking for all training sessions, summary feedback was
provided about the performance over several gait cycles
during the previous block relative to the hip-knee angle or
sagittal plane footpath. At the beginning of every training
session, the participant was shown a figure with his performance in the previous sessions (including baseline) versus the
required template. This way, the participant could see
whether there was an improvement in performance over
successive sessions.

Protocol
Initial training had the participant walking at his selfselected walking speed (1.5 miles/hr or 0.67 m/sec), determined during the initial gait evaluations. The speed was
gradually increased over sessions to the participant’s tolerance, reaching a maximum of 1.9 miles/hr (0.85 m/sec) by the
end of training. Training began with full (100%) gravity
balancing of both the leg and device. Gradually, over several
sessions, the gravity balancing of the leg was reduced to 75%,
50%, 25%, and finally 0% gravity balancing, whereas the
device was always 100% gravity balanced (ie, the device
itself was always gravity neutral so that the user would not
“feel” the mass of the device). Table 2 shows the schedule of
training in detail, including the level of gravity balancing, use
of FES, and visual feedback at each training session, and the
timing of the evaluation sessions. The follow-up evaluation
© 2008 Neurology Section, APTA

JNPT • Volume 32, December 2008

Gait Training After Stroke

TABLE 2. Training Protocol
Feedback

Evaluation
Session

Training
Day

% Gravity
Balanced

FES

Continuous

Pre-training

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15

100
100
100
100
100
100
75
75
50
50
50
25
25
0
0

—
—
—
PF
PF/DF
PF/DF
PF/DF
PF/DF
PF
PF/DF
PF/DF
PF/DF
PF/DF
PF/DF
PF/DF

x
x

Mid-training

Intermittent

x (early)

x (early)

x
x (late)
x
x
x
x
x (late)
x

Summary

Foot/Joint

x
x
x
x
x
x
x
x
x
x
x
x
x
x
x

Foot
Joint
Foot
Foot
Joint
Foot/joint
Foot/joint
Joint
Joint
Foot/joint
Joint
Joint
Joint
Joint
Joint

Post-training
Abbreviations: FES, functional electrical stimulation; PF, plantarflexors (medial and lateral gastrocnemius); DF, dorsiflexors (tibialis anterior).
Early and late refer to the first half (blocks 1 and 2) and second half (blocks 3 and 4) of each training session. Please note that the device always gravity-balanced the mass of
the gravity-balanced orthosis, so that the user would not “feel” the mass of the device.

session is not shown as it was performed a month after the
final training session.

Outcomes
Clinical Assessment
The only clinical measure that showed improvement
from the pretraining evaluation to the posttraining evaluation
was the participant’s score on the Timed Up and Go test. By
the end of training, the participant’s time had dropped to nine
seconds compared with his time of 15 seconds before training, a 40% improvement in performance. The posttraining
time was within the normal range for adults without neurological impairment who are independent with balance and
mobility skills.

FIGURE 3. Gait speed in miles per hour across the four
evaluation sessions. SD bars across gait cycles are shown.
© 2008 Neurology Section, APTA

Gait Evaluation
Overground walking
Gait speed showed a marginal improvement over training, increasing by 10% from the pretraining to post-training
(Figure 3). This improvement was maintained at the onemonth post-training follow-up evaluation.
To test for the effect of training on gait symmetry, we
calculated the ratio of the stance time to swing time for each
leg. As expected, the ratio was very asymmetrical pretraining,
with the stance to swing ratio being much lower on the
impaired side (Table 3; 1.2 on affected side vs 2.0 on the
unaffected side). Immediately post-training, a more symmetrical gait pattern was seen (1.6 vs 1.8), which was due to
decreases in the ratio for the unimpaired side as well as an
increase in the ratio on the impaired side. However, this
training effect was lost completely by the one-month follow-up evaluation, with a substantial increase in the ratio on
the unimpaired limb. This result suggests that although there
was an initial adaptation to the training, the changes that lead
to greater gait symmetry were not learned.
There was an increase of approximately 20% in excursion of the hip (from 33.2 to 39.7 degrees) and approximately
9% of the knee (from 24.8 to 27.2 degrees) post-training
during the swing phase of gait (Figure 4), which continued to
increase at the one-month follow-up evaluation. In addition,
the amount of plantarflexion at initial foot contact decreased
over the training period, although the ankle was still less than
neutral at the end of training (8.3 to 4.6 degrees of plantarflexion from pre- to post-training, Table 4). The amount of
plantarflexion during swing also decreased by the one-month
follow-up evaluation (2.8 degrees of plantarflexion). Other
changes in joint motion were less systematic or not main-

197

JNPT • Volume 32, December 2008

Krishnamoorthy et al

TABLE 3. Gait Symmetry Across Training Sessions
Pretraining

Stance time (% of gait cycle)
Swing time (% of gait cycle)
Stance/swing ratio

Midtraining

Follow-up

Affected

Unaffected

Affected

Unaffected

Affected

Unaffected

Affected

Unaffected

0.55
0.45
1.2

0.67
0.33
2.0

0.61
0.39
1.6

0.65
0.35
1.9

0.62
0.38
1.6

0.64
0.36
1.8

0.58
0.42
1.4

0.71
0.29
2.4

tained. Finally, step length of the affected foot during the
swing phase was estimated by calculating the distance traveled by the ankle marker in the sagittal plane. This measure
was chosen because the greatest effect of the GBO was
expected to be on the hip and knee of the affected side during
swing. This distance not only increased post-training by
about 10% (from 101.2 to 111.2 cm), but also remained
higher at the one-month follow up, suggesting that the participant had learned to take longer steps on the affected side.

Changes in Ground Reaction Forces
The vertical ground reaction forces on the affected side
in the posttraining and follow-up evaluations were higher
than at the pretraining evaluation, suggesting that the participant learned to take more weight on the affected limb (Figure
5). Although the peak ground reaction force increased from
830.67 N pre-training to 937.95 N mid-training, it decreased
again at the posttraining evaluation (858.9 N), and then
increased again by the one-month posttraining evaluation
(899.83 N). There was, therefore, a general upward trend.
However, because the highest value occurred midtraining, the
reliability of this improvement is uncertain.

Changes in Isometric Muscle Performance
After training, the MG-LG MVIC showed a 46.4%
increase (pre-training ⫽ 207.8 N, post-training ⫽ 304.24 N);
the CAR showed a 9.21% increase (pre-training ⫽ 0.51,

FIGURE 4. Average hip and knee angle motion (⫾SD) during the stance and swing phases during each of the four
evaluation sessions. Positive angles correspond to flexion
motion and negative angles correspond to extension motion.

198

Post-training

post-training ⫽ 0.55); the estimated maximum force generating ability of MG-LG showed a 34% increase (pre-training ⫽ 410.49 N, post-training ⫽ 549.97 N).

Subjective Feedback from Participant After
Training
The participant reported tolerating the training protocol
well. During the training sessions, the participant reported
being more “aware” of his gait and was “conscious” of
whether his foot was clearing the ground adequately, bending
his knee adequately during swing, or transporting his affected
foot farther forward. After training, the participant reported
feeling that he walked faster during overground walking and
that his foot was catching less on the ground during normal
activities. He specifically reported that he was more confident
walking in his neighborhood over uneven terrain and was
spending more time doing so.

DISCUSSION
This case report provides a preliminary evaluation of
the effectiveness of a poststroke gait retraining intervention
that involved the simultaneous application of a gravity-balanced orthotic device, FES, and visual feedback. Overall, the
training regimen was well tolerated by the participant. Some
of our predictions were supported by quantitative changes in
gait during this report. The time required to perform the
Timed Up and Go decreased; there was an increase in the
distance covered by the impaired side during swing, the speed
of walking, weight-bearing on the impaired limb, and knee
and hip joint excursions. All these parameters were also
maintained during the one-month follow-up. There was also
46% increase in MVIC of MG-LG. Unfortunately, although
the participant’s gait was more symmetrical after the fiveweek training, this symmetry was not maintained at the
one-month evaluation post-training.
These results are encouraging for an initial case report
evaluating the effectiveness of our novel approach. For most
of our outcome measures, the minimal clinically important
difference are not known. One standardized test that did show
a difference post-training was the Timed Up and Go. Although minimal clinically important difference or cutoff
scores are not known for the stroke population, a score of
ⱖ14 seconds is suggestive of a high risk of falls in community-dwelling older adults.52 Adults without neurological impairments who are independent with balance and mobility
skills are able to perform the test in ⬍10 seconds.53 Thus, at
the beginning of this case report, the participant was at risk of
falls based on this criterion (15 seconds), but after the
training, he was able to perform the test in the same time as
adults without neurological impairments (nine seconds).
© 2008 Neurology Section, APTA

JNPT • Volume 32, December 2008

Gait Training After Stroke

TABLE 4. Overground Walking Parameters
Evaluations

Hip flexion excursion—swing (degree)
Hip abduction excursion—swing (degree)
Peak hip extension angle (⫺)—stance (degree)
Knee flexion excursion—swing (degree)
Ankle position [PF(⫺)/DF(⫹)] at initial
contact (degree)
Path length of foot—swing (cm)

Pre-training

Mid-training

Post-training

Follow-up

33.2
2.3
10.7
1.3
⫺7.9
0.8
24.8
4.7
⫺8.3
2.3
101.2
4.4

38.0
3.8
13.2
0.5
⫺9.9
2.5
30.7
2.9
⫺7.1
0.8
109.2
3.1

39.7
7.6
14.1
1.4
⫺13.3
0.8
27.2
2.7
⫺4.6
1.0
111.2
2.2

42.9
2.1
9.5
0.6
⫺9.0
1.9
40.9
2.7
⫺2.8
2.2
109.9
2.2

Mean
SD
Mean
SD
Mean
SD
Mean
SD
Mean
SD
Mean
SD

The bold values indicate the measures that showed the most change from the pre-training to the post-training measurements and were maintained during follow-up.
Abbreviations: PF, plantarflexors (medial and lateral gastrocnemius); DF, dorsiflexors (tibialis anterior).

Results of this preliminary investigation compare favorably with those of previous studies examining the effects
of novel gait retraining programs.11–14,18 Studies that used a
single treatment intervention showed an improvement in
some parameters such as gait speed and balance.11,12,14,18 This
was sometimes accompanied by other improvements such as
a decrease in double support time or increase in stance
time.12,13 However, some of these studies were performed
within the first few months after stroke14,18 making it difficult
to rule out spontaneous recovery as one of the factors contributing to improvement in performance. Further, another
difference of our approach is to address specific clinical
problems with targeted treatments. For example, patients who
have difficulty in producing adequate hip and knee flexion
during swing would benefit from using the GBO and those
with inadequate plantar- and/or dorsiflexor strength would
likely benefit from addition of FES. An important difference
to point out between the GBO and other robotic devices used
in other studies is that the GBO does not move the patient’s

FIGURE 5. Average ground reaction forces measured from
the force platform during the stance phase of walking at
each of the four evaluation sessions. The inset shows the
peak ground reaction force with SD bars.
© 2008 Neurology Section, APTA

limb through a predetermined pattern, but instead makes it
easier for the patient to move by reducing the effects of
gravity.
In our report, the participant’s balance before training
was normal based on his performance on the BBS, so we did
not expect changes in balance with training. Recently, a
ceiling effect has been noted with the BBS in communitydwelling older adults54 and chronic stroke survivors.55 Thus,
in future studies, we would consider using alternate tests for
balance, such as the Postural Assessment Scale for Stroke
Patients, which did not show such a ceiling effect.55
The participant was an independent ambulator before
the intervention, but his gait was awkward. Although we did
not specifically measure energy cost, the participant reported
fatiguing quickly after walking for short distances, which
restricted his walking. In addition, he had developed several
gait compensations. After the training, he voluntarily reported
that he was able to walk longer distances without fatiguing
and that he tripped less. The reported decrease in fatigue is
consistent with previous studies showing that treadmill training improves cardiovascular fitness in chronic stroke survivors.25 The results of this case report are promising in that
changes in walking of this chronic stroke survivor occurred
after a five-week period of intensive training. This report adds
support to the recent idea that with appropriate intervention,
even chronic participants may improve performance.11 An
important future issue is to determine whether individuals
with subacute stroke can also achieve improvements in their
walking with the present training. We believe that this mode
of treatment may be more effective in such persons if applied
intensively because it may help limit the gait compensations
that such persons typically develop when attempting to ambulate, whereas selective muscle weakness is a significant
problem.
Another issue that needs to be studied further is why
some parameters such as gait symmetry were not retained in
the one month after training. It may be that the asymmetries
developed by the participant were too ingrained over the three

199

JNPT • Volume 32, December 2008

Krishnamoorthy et al

plus years of functional walking after his stroke to be permanently affected without more intensive training than the
five weeks provided here. Spending more time training at the
lower levels of gravity balancing (25% and 0%) at the end of
training may have been beneficial in this regard. We were
unable to extend the training beyond this period because of
the participant’s personal constraints.
In this report, the GBO was used to train motion of the
hip and knee joints. Problems in controlling the ankle were
addressed by adding FES to the ankle muscles. We took this
approach because patients after a stroke frequently have
greater impairment of muscles of the distal limb. A long-term
goal of this work is to build a portable stimulator so that
participants who require FES to substitute for the lack of
voluntary muscle activation at the ankle will have a device
that they can use permanently. Portable FES systems could
provide a means for controlling ankle muscles during daily
activities.38,56 –59 In contrast, a portable system to help control
the hip or knee joints is more difficult, particularly regarding
comfort (eg, stimulating the hamstrings is typically very
uncomfortable). Thus, the GBO in combination with motor
learning through feedback and structured practice was designed to help the participant improve the control of the hip
and knee joints, as well as the overall foot trajectory. Improvement of knee and hip flexion during swing, in turn, will
reduce the amount of ankle dorsiflexion required to clear the
foot during walking.
A recent meta-analysis of literature showed that FES
produces improvements in muscle strength post-stroke.60 Interestingly, the MG-LG MVIC, CAR, and estimated maximum force generating all showed considerable increases with
training, suggesting that our training approach probably increased the muscle’s force generating ability, both volitionally and in response to stimulation. Because no specific
muscle training was performed in this report, we speculate
that the functional strength improvements were related to the
practice with gradually increasing difficulty as gravity balancing was reduced, in combination with the FES. The
question remains, however, whether similar increases in muscle strength could be obtained by FES alone or in combination with a similar intensity of treadmill training alone.
Similar questions can be raised about the measured
changes in overground walking speed and movement kinematics. However, it is noteworthy that each time the gravity
compensation was reduced, it took some time for the participant to recover the ability to approximate the required
templates, suggesting that practice in the reduced gravity
environment helped the participant develop improved control
of his foot trajectory.
One of the strengths of this training approach is that
continuous practice of walking can be provided along with on
line visual feedback and summary feedback after multiple
walking cycles. The participant was able to see his performance with respect to the prescribed trajectories and his
improvement in performance within and across training sessions. This was motivational for the participant. We used
known principles of motor learning, in combination with
practical considerations, to design the practice and feedback
provided during training. However, future work will need to

200

determine how to optimize the training and feedback to
achieve better results. For example, although summary feedback has been shown to be better than continuous feedback to
enhance learning in healthy subjects learning to parameterize
relatively simple tasks,61 it is unclear whether summary,
faded or continuous feedback might be more effective to
enhance learning of motor patterns by patients with motor
impairments, particularly in the early phases of learning.62,63
This case report is an initial test of a novel training
protocol. Although several of the measured variables showed
a change in the direction of improvement, none except the
Timed Up and Go has established norms that can be used to
demonstrate clinical change. As already discussed, there
could have been a ceiling effect of the BBS in this patient.
This is a limitation of this report and future studies should
include other tests that are more sensitive. Another limitation
of the report is that the GBO in its current form is not yet
ready for clinical use. The current prototype is a research
prototype. It requires the engineers to make adjustments in
settings of the springs and links to adjust the device for a
given subject and also for different levels of gravity balancing. However, it serves as a proof of concept. Future versions
of the GBO will address these issues to make it user friendly.
The advantage of the device is that it makes it easier for the
patient to voluntarily control their impaired leg without the
typical compensations, allowing the practice of movements
that better approximate “normal.” From a financial point of
view, it would be much cheaper than many motor-driven
robotic devices because it does not consist of any powered
elements. By making therapy more automated, gravity-balanced gait training may allow therapists to train several
patients simultaneously with less physical effort by the therapist. These predictions obviously require more extensive
investigation.
The current case report served to show the feasibility of
using the GBO in conjunction with FES and visual feedback
as a gait-training paradigm, thereby providing a framework
for designing future studies. Future, controlled studies are
planned with small groups of patients receiving training for
similar lengths of time with only the GBO or FES, with both
GBO and FES, and a control group receiving walking practice on a treadmill without the GBO or FES. Another question
that needs to be resolved in future work is whether the
improvements observed were due the specific training or
resulted from a general increase in strength that could be
achieved with methods involving intensive gait training. The
data from the current report cannot answer this question.
However, the participant was a relatively young, active individual. Thus, it is less likely that strength improvements per
se played the primary role in his gait improvements. One way
that we plan to control for this in future work is to have
patients engage in an intensive strength training protocol
before commencing gait training.

SUMMARY
In this case report, we applied a novel training approach
to a chronic stroke survivor that included combining a GBO,
FES, and motor learning principles to improve walking performance. This approach was found to be feasible and
© 2008 Neurology Section, APTA

JNPT • Volume 32, December 2008

showed positive effects on the participant’s gait after a
five-week training period. Advantages of this training approach are that training parameters including feedback can be
precisely computer controlled and accurate data can be generated that the therapist can use to monitor progress within
and across sessions. Future research will be required to
determine the effectiveness of different components of this
training in both the subacute and chronic stages of stroke
recovery.

ACKNOWLEDGMENTS
The project described was supported by grant number
HD38582 from the National Center for Medical Rehabilitation Research. The content is solely the responsibility of the
authors and does not necessarily represent the official views
of the National Center for Medical Rehabilitation Research.
REFERENCES
1. Rosamond W, Flegal K, Friday G, et al. Heart disease and stroke
statistics—2007 update: a report from the American Heart Association
Statistics Committee and Stroke Statistics Subcommittee. Circulation.
2007;115:e69 – e171.
2. Mayo NE, Wood-Dauphinee S, Ahmed S, et al. Disablement following
stroke. Disabil Rehabil. 1999;21:258 –268.
3. Brandstater ME, de Bruin H, Gowland C, et al. Hemiplegic gait: analysis
of temporal variables. Arch Phys Med Rehabil. 1983;64:583–587.
4. Olney SJ, Richards C. Hemiparetic gait following stroke, part I: characteristics. Gait Posture. 1996;4:136 –148.
5. Perry J, Garrett M, Gronley JK, et al. Classification of walking handicap
in the stroke population. Stroke. 1995;26:982–989.
6. Detrembleur C, Dierick F, Stoquart G, et al. Energy cost, mechanical
work, and efficiency of hemiparetic walking. Gait Posture. 2003;18:47–
55.
7. Diller L, Weinberg J. Evidence for accident-prone behavior in hemiplegic patients. Arch Phys Med Rehabil. 1970;51:358 –363.
8. Nyberg L, Gustafson Y. Patient falls in stroke rehabilitation. A challenge
to rehabilitation strategies. Stroke. 1995;26:838 – 842.
9. Malouin F, Potvin M, Prevost J, et al. Use of an intensive task-oriented
gait training program in a series of patients with acute cerebrovascular
accidents. Phys Ther. 1992;72:781–789; discussion 789 –793.
10. Richards CL, Malouin F, Wood-Dauphinee S, et al. Task-specific
physical therapy for optimization of gait recovery in acute stroke
patients. Arch Phys Med Rehabil. 1993;74:612– 620.
11. Miller EW, Quinn ME, Seddon PG. Body weight support treadmill and
overground ambulation training for two patients with chronic disability
secondary to stroke. Phys Ther. 2002;82:53– 61.
12. Dunsky A, Dickstein R, Marcovitz E, et al. Home-based motor imagery
training for gait rehabilitation of people with chronic poststroke hemiparesis. Arch Phys Med Rehabil. 2008;89:1580 –1588.
13. Husemann B, Muller F, Krewer C, et al. Effects of locomotion training
with assistance of a robot-driven gait orthosis in hemiparetic patients
after stroke: a randomized controlled pilot study. Stroke. 2007;38:349 –
354.
14. Hesse S, Uhlenbrock D. A mechanized gait trainer for restoration of gait.
J Rehabil Res Dev. 2000;37:701–708.
15. Hesse S, Werner C, Uhlenbrock D, et al. An electromechanical gait
trainer for restoration of gait in hemiparetic stroke patients: preliminary
results. Neurorehabil Neural Repair. 2001;15:39 –50.
16. Whitall J. Stroke rehabilitation research: time to answer more specific
questions? Neurorehabil Neural Repair. 2004;18:3– 8; author reply 9 –11.
17. Tong RK, Ng MF, Li LS. Effectiveness of gait training using an
electromechanical gait trainer, with and without functional electric
stimulation, in subacute stroke: a randomized controlled trial. Arch Phys
Med Rehabil. 2006;87:1298 –1304.
18. Tong RK, Ng MF, Li LS, et al. Gait training of patients after stroke using
an electromechanical gait trainer combined with simultaneous functional
electrical stimulation. Phys Ther. 2006;86:1282–1294.
19. Mirelman A, Bonato P, Deutsch JE. Effects of training in robot virtual

© 2008 Neurology Section, APTA

Gait Training After Stroke

reality system vs a robot alone on the gait of individuals post stroke.
Stroke. In press.
20. Ferraro M, Palazzolo JJ, Krol J, et al. Robot-aided sensorimotor arm
training improves outcome in patients with chronic stroke. Neurology.
2003;61:1604 –1607.
21. Agrawal SK, Banala SK, Fattah A, et al. Assessment of motion of a
swing leg and gait rehabilitation with a gravity balancing exoskeleton.
IEEE Trans Neural Syst Rehabil Eng. 2007;15:410 – 420.
22. Banala SK, Agrawal SK, Fattah A, et al. Gravity-balancing leg orthosis
and its performance evaluation. IEEE Trans Robotics. 2006;22:1228 –
1239.
23. Benoit D, Krishnamoorthy V, Banala SK, et al. An integrated approach
for improving gait in a stroke population: combining robotics, FES and
neuromusculoskeletal modeling. Presented at: Annual Meeting of American Society of Biomechanics, Blacksburg, VA, 2006.
24. Nudo RJ. Mechanisms for recovery of motor function following cortical
damage. Curr Opin Neurobiol. 2006;16:638 – 644.
25. Macko RF, Ivey FM, Forrester LW, et al. Treadmill exercise rehabilitation improves ambulatory function and cardiovascular fitness in patients with chronic stroke: a randomized, controlled trial. Stroke. 2005;
36:2206 –2211.
26. Bajd T, Kralj A, Sega J, et al. Use of a two-channel functional electrical
stimulator to stand paraplegic patients. Phys Ther. 1981;61:526 –527.
27. Bajd T, Kralj A, Turk R. Standing-up of a healthy subject and a
paraplegic patient. J Biomech. 1982;15:1–10.
28. Agarwal S, Kobetic R, Nandurkar S, et al. Functional electrical stimulation for walking in paraplegia: 17-year follow-up of 2 cases. J Spinal
Cord Med. 2003;26:86 –91.
29. Kralj A, Bajd T, Turk R. Enhancement of gait restoration in spinal
injured patients by functional electrical stimulation. Clin Orthop Relat
Res. 1988;233:34 – 43.
30. Wieler M, Stein RB, Ladouceur M, et al. Multicenter evaluation of
electrical stimulation systems for walking. Arch Phys Med Rehabil.
1999;80:495–500.
31. Mangold S, Keller T, Curt A, et al. Transcutaneous functional electrical
stimulation for grasping in subjects with cervical spinal cord injury.
Spinal Cord. 2005;43:1–13.
32. Binder-Macleod SA, Lee SCK. Assessment of efficacy of functional
electrical stimulation in patients with hemiplegia. Top Stroke Rehabil.
1997;3:88 –98.
33. Burridge JH, Taylor PN, Hagan SA, et al. The effects of common
peroneal stimulation on the effort and speed of walking: a randomized
controlled trial with chronic hemiplegic patients. Clin Rehabil. 1997;11:
201–210.
34. Chae J, Yu DT. Neuromuscular electrical stimulation for motor restoration in hemiparesis. Top Stroke Rehabil. 2002;8:24 –39.
35. Johnson CA, Burridge JH, Strike PW, et al. The effect of combined use
of botulinum toxin type A and functional electric stimulation in the
treatment of spastic drop foot after stroke: a preliminary investigation.
Arch Phys Med Rehabil. 2004;85:902–909.
36. Kottink AI, Oostendorp LJ, Buurke JH, et al. The orthotic effect of
functional electrical stimulation on the improvement of walking in
stroke patients with a dropped foot: a systematic review. Artif Organs.
2004;28:577–586.
37. Ring H, Rosenthal N. Controlled study of neuroprosthetic functional
electrical stimulation in sub-acute post-stroke rehabilitation. J Rehabil
Med. 2005;37:32–36.
38. Taylor PN, Burridge JH, Dunkerley AL, et al. Clinical use of the
Odstock dropped foot stimulator: its effect on the speed and effort of
walking. Arch Phys Med Rehabil. 1999;80:1577–1583.
39. Voigt M, Sinkjaer T. Kinematic and kinetic analysis of the walking
pattern in hemiplegic patients with foot-drop using a peroneal nerve
stimulator. Clin Biomech (Bristol, Avon). 2000;15:340 –351.
40. Weber DJ, Stein RB, Chan KM, et al. BIONic WalkAide for correcting
foot drop. IEEE Trans Neural Syst Rehabil Eng. 2005;13:242–246.
41. Schmidt RA, Lee TD. Motor Control and Learning: A Behavioral
Emphasis. Champaign, IL: Human Kinetics, Inc.; 2005.
42. Schmidt RA, Wulf G. Continuous concurrent feedback degrades skill
learning: implications for training and simulation. Hum Factors. 1997;
39:509 –525.
43. Colarusso RP, Hammill DD. Motor-Free Visual Perception Test
(MVPT-3). Novato, CA: Academic Therapy Publications; 2006.

201

Krishnamoorthy et al

44. Folstein MF, Folstein SE, McHugh PR. “Mini-mental state.” A practical
method for grading the cognitive state of patients for the clinician.
J Psychiatr Res. 1975;12:189 –198.
45. Bohannon RW, Smith MB. Interrater reliability of a modified Ashworth
scale of muscle spasticity. Phys Ther. 1987;67:206 –207.
46. Berg KO, Wood-Dauphinee SL, Williams JI, et al. Measuring balance in
the elderly: validation of an instrument. Can J Public Health. 1992;
83(suppl 2):S7–S11.
47. Lindmark B, Hamrin E. Evaluation of functional capacity after stroke as
a basis for active intervention. Presentation of a modified chart for motor
capacity assessment and its reliability. Scand J Rehabil Med. 1988;20:
103–109.
48. Podsiadlo D, Richardson S. The timed “Up & Go”: a test of basic
functional mobility for frail elderly persons. J Am Geriatr Soc. 1991;
39:142–148.
49. Herbert RD, Gandevia SC. Twitch interpolation in human muscles:
mechanisms and implications for measurement of voluntary activation.
J Neurophysiol. 1999;82:2271–2283.
50. Norregaard J, Lykkegaard JJ, Bulow PM, et al. The twitch interpolation
technique for the estimation of true quadriceps muscle strength. Clin
Physiol (Oxford, England). 1997;17:523–532.
51. Binder-Macleod SA, Kesar T. Catchlike property of skeletal muscle:
recent findings and clinical implications. Muscle Nerve. 2005;31:681– 693.
52. Shumway-Cook A, Brauer S, Woollacott M. Predicting the probability
for falls in community-dwelling older adults using the Timed Up & Go
Test. Phys Ther. 2000;80:896 –903.
53. Shumway-Cook A, Woollacott M. Motor Control: Theory and Practical
Applications. Baltimore, MD: Williams & Wilkins; 1995.

202

JNPT • Volume 32, December 2008

54. Bogle Thorbahn LD, Newton RA. Use of the Berg Balance Test to
predict falls in elderly persons. Phys Ther. 1996;76:576 –583; discussion
584 –585.
55. Mao HF, Hsueh IP, Tang PF, et al. Analysis and comparison of the
psychometric properties of three balance measures for stroke patients.
Stroke. 2002;33:1022–1027.
56. Chae J, Yu D. A critical review of neuromuscular electrical stimulation
for treatment of motor dysfunction in hemiplegia. Assist Technol. 2000;
12:33– 49.
57. de Kroon JR, van der Lee JH, IJzerman MJ, et al. Therapeutic electrical
stimulation to improve motor control and functional abilities of the
upper extremity after stroke: a systematic review. Clin Rehabil. 2002;
16:350 –360.
58. Peckham PH, Knutson JS. Functional electrical stimulation for neuromuscular applications. Annu Rev Biomed Eng. 2005;7:327–360.
59. Stein RB, Chong S, Everaert DG, et al. A multicenter trial of a footdrop
stimulator controlled by a tilt sensor. Neurorehabil Neural Repair.
2006;20:371–379.
60. Glanz M, Klawansky S, Stason W, et al. Functional electrostimulation in
poststroke rehabilitation: a meta-analysis of the randomized controlled
trials. Arch Phys Med Rehabil. 1996;77:549 –553.
61. Sherwood DE, Lee TD. Schema theory: critical review and implications
for the role of cognition in a new theory of motor learning. Res Q Exerc
Sport. 2003;74:376 –382.
62. Hanlon RE. Motor learning following unilateral stroke. Arch Phys Med
Rehabil. 1996;77:811– 815.
63. Winstein CJ, Merians AS, Sullivan KJ. Motor learning after unilateral
brain damage. Neuropsychologia. 1999;37:975–987.

© 2008 Neurology Section, APTA