Training for Walking Efficiency With a Wearable Hip-Assist

Robot in Patients With Stroke

A Pilot Randomized Controlled Trial
Hwang-Jae Lee, PT, PhD; Su-Hyun Lee, PT, MSc; Keehong Seo, PhD; Minhyung Lee, PhD;
Won Hyuk Chang, MD, PhD; Byung-Ok Choi, MD, PhD; Gyu-Ha Ryu, PhD; Yun-Hee Kim, MD, PhD

Background and Purpose—The purpose of this study was to investigate the effects of gait training with a newly developed
wearable hip-assist robot on locomotor function and efficiency in patients with chronic stroke.
Methods—Twenty-eight patients with stroke with hemiparesis were initially enrolled, and 26 patients completed the
randomized controlled trial (14 in the experimental and 12 in the control groups). The experimental group participated in
a gait training program over a total of 10 sessions, including 5 treadmill sessions and 5 over-ground gait training sessions
while wearing a hip-assist robot, the Gait Enhancing and Motivating System (GEMS, Samsung Advanced Institute of
Technology, Suwon, Republic of Korea). The control group received gait training without Gait Enhancing and Motivating
System. Primary outcome measured locomotor function and cardiopulmonary metabolic energy efficiency. Also,
secondary outcome measured motor function and balance parameter.
Results—Compared with the control group, the experimental group had significantly greater improvement in spatiotemporal
gait parameters and muscle efforts after the training intervention (P<0.05). The net cardiopulmonary metabolic energy
cost (mL·kg−1·min−1) was also reduced by 14.71% in the experimental group after the intervention (P<0.01). Significant
group×time interactions were observed for all parameters (P<0.05). Cardiopulmonary metabolic efficiency was strongly
correlated with gait symmetry ratio in the experimental group (P<0.01).
Conclusions—Gait training with Gait Enhancing and Motivating System was effective for improving locomotor function
and cardiopulmonary metabolic energy efficiency during walking in patients with stroke. These findings suggest that
robotic locomotor training can be adopted for rehabilitation of patients with stroke with gait disorders.
Downloaded from http://ahajournals.org by on November 1, 2020

Clinical Trial Registration—URL: https://clinicaltrials.gov. Unique identifier: NCT02843828.   (Stroke. 2019;50:3545-

3552. DOI: 10.1161/STROKEAHA.119.025950.)
Key Words: gait ◼ locomotion ◼ robotics ◼ stroke rehabilitation ◼ walking

See related article, p 3337 negative impacts on quality of life.3,4 Therefore, restoration of

S troke is the primary cause of adult neurological disability,

affecting all aspects of life including physical function,
balance, and activities of daily living. Gait deficits con-
tribute significantly to functional disability after stroke, and
of all stroke-related impairments, improvement in gait func-
tion is the goal most often requested by patients with stroke.1
Asymmetrical gait patterns are major characteristics of hem-
iplegic gait observed in individuals poststroke. Impairment
in gait symmetry can result in compensatory movement pat-
terns, slowed gait speed, limited functional mobility, and
increased risk of falls.2 As a result, gait asymmetry in patients
with stroke leads to higher energy consumption and can have
gait symmetry is one of the important gait rehabilitation goals
for patients with stroke.5
Recent technological advances in stroke gait rehabilita-
tion have made it possible for robotic devices to provide safe,
intensive training through accurate repetitive motion.6–9 The
primary reasons for using robotic rehabilitation are that ap-
propriately designed robots are able to reduce the burden on
therapists while providing extensive task-specific practice
in novel dynamic environments, and they allow continuous
monitoring of patient performance and progression.10 Robotic
devices providing mechanically assisted functional gait train-
ing were recently adapted for patients with neuromuscular gait

Received August 28, 2018; final revision received May 15, 2019; accepted May 17, 2019.
From the Department of Physical and Rehabilitation Medicine (H.-J.L., S.-H.L., W.H.C., Y.-H.K.), Department of Neurology, Neuroscience Center
(B.-O.C.), and Department of Medical Device Management and Research, SAIHST (G.-H.R.), Center for Prevention and Rehabilitation, Heart Vascular
Stroke Institute, Samsung Medical Center, Sungkyunkwan University School of Medicine, Seoul, Republic of Korea; Department of Health Sciences
and Technology (H.-J.L.) and Department of Health Sciences and Technology, Department of Medical Device Management and Research, Department
of Digital Health (Y.-H.K.), SAIHST, Sungkyunkwan University, Seoul, Republic of Korea; and Samsung Advanced Institute of Technology, Samsung
Electronics, Gyeonggi-do, Republic of Korea (K.S., M.L.).
The online-only Data Supplement is available with this article at https://www.ahajournals.org/doi/suppl/10.1161/STROKEAHA.119.025950.
Correspondence to Yun-Hee Kim, MD, PhD, Department of Physical and Rehabilitation Medicine, Center for Prevention and Rehabilitation, Heart
Vascular Stroke Institute, Samsung Medical Center, Sungkyunkwan University School of Medicine, Samsung Advanced Institute for Health Science and
Technology, Sungkyunkwan University, 81 Irwon-ro, Gangnam-gu, Seoul, 06351, Republic of Korea. Email yun1225.kim@samsung.com
© 2019 American Heart Association, Inc.
Stroke is available at https://www.ahajournals.org/journal/str DOI: 10.1161/STROKEAHA.119.025950

3545
 3546  Stroke  December 2019
disorders. Representatively, Lokomat produced by Hocoma
(Zurich, Switzerland) was the most widely used robotic gait-
assist device for patients with stroke. One of the advantages
of Lokomat training is that virtual reality with constant audio,
and visual feedback is offered. However, conventional gait
training is focused on using autonomous movement generated
by the robot primarily on treadmills. Clinical applications for
wearable robots are being actively explored to overcome the
limitations of fixed-type robots.9,11–13 Wearable robotic gait
training enables personalized gait training with active user
participation in a variety of environments.14
We used the Gait Enhancing and Motivating System
(GEMS), a wearable hip-assistance robot, developed by
Samsung Advanced Institute of Technology (Suwon, Republic
of Korea) to deliver locomotion training that interactively
assists motion according to the user's voluntary efforts. In
this randomized controlled trial, we aimed to investigate the
effects of locomotion training with the GEMS on improve-
ment in locomotor function, cardiopulmonary metabolic en-
ergy efficiency, motor function and balance in patients with
chronic stroke.
Methods
Data Availability Statement
The data that support the findings of this study are available from the
corresponding author on reasonable request.
Participants
Downloaded from http://ahajournals.org by on November 1, 2020
Individuals were eligible for inclusion if they met the following require-
ments: (1) hemiplegic stroke more than 3 months prior, (2) age >20 years,
(3) ability to walk without personal assistance (Functional Ambulatory
Category [FAC],15 scale 3 and 4, moderate to mild impairment), (4) no
excessive spasticity in hip, knee, and ankle joints (Modified Ashworth
Scale <3), (5) sufficient cognition to follow simple instructions and to
understand the content and purpose of the study (the Korean version
of the Montreal Cognitive Assessment score ≥ 23 points), and (6) no
musculoskeletal conditions such as lower extremity fracture or painful
arthritis of the joints. Potential participants were excluded if they had
severe heart disease or uncontrolled hypertension, or if they had any
neurological disease that might interfere with the study. Eligible par-
ticipants were randomly placed in either the experimental group (gait
training with GEMS) or control group (gait training without GEMS)
by an administrator who used a computer program to generate random
assignments after baseline assessment. Each subject provided written
informed consent by signing a form that was approved by the Samsung
Medical Center Research Institutional Review Board.
After the pretraining test, all participants were randomly assigned
to treadmill or over-ground gait training by a random number table.
If a patient was initially selected for treadmill gait training, 5 over-
ground gait training sessions were performed after completing 5
treadmill gait training sessions.
The protocol in our Clinicaltrials.gov was designed as par-
allel 2 experiments for the elderly (n=28) and patients with stroke
(n=28) with the same experimental design. For both experiments,
participants were randomized by experimental (training with robot)
or control (training without robot) group, respectively. The data of 2
experiments were separately analyzed, and the results obtained from
stroke participants were reported in this article. Analysis of data from
the elderly participants will be reported in the subsequent article.
Gait Enhancing and Motivating System
The GEMS is a lightweight (2.8 kg), slim, comfortable, and powerful
active assistance robot developed to improve locomotor function
in patients with stroke with gait disorders (Figure 1A and 1B). The
GEMS torque assistance units consist of angular sensors and actua-
tors that work on bilateral hip joints. The GEMS can provide assist
torque and power around the bilateral hip joints for both extension
and flexion during walking. A research administrator used a tablet
PC (Galaxy Tab 3 8.0 with Android 4.2 OS, Samsung Electronics Co,
Ltd, Republic of Korea) to specify user weights and assistance levels
for experiments. For more information about the assistive algorithms
used for the GEMS, please see Supplement I in the online-only Data
Supplement.
Intervention Protocol
The training intervention was conducted for 4 weeks with 3 sessions
of training per week. All participants completed a gait training pro-
gram with a total of 10 sessions comprising 5 treadmill sessions and
5 over-ground gait training sessions with GEMS in the experimental
group or without GEMS in the control group. The order of tread-
mill or over-ground gait training was assigned by a random number
table. Duration of one training session was 45 minutes comprised of
a 5 minutes warm-up, a 35 minutes gait practice including 5 minutes
resting time, and a 5 minutes cool-down. If participants missed any
training session, additional session was offered at another time during
the week or during an additional week at the end of the intervention
period. For selecting speed of the first treadmill gait training, partici-
pants walked on the treadmill at their most comfortable speed with
GEMS in the experimental group or without GEMS in the control
group with wearing a harness without any body weight support. For
the next 4 sessions, the treadmill speed was gradually increased by
5% per session. In the same manner, the first over-ground gait training
was performed at a comfortable speed for each participant in a cor-
ridor of the laboratory. For the next 4 over-ground training sessions,
the examiner encouraged participants to walk as fast as possible while
maintaining their balance. Therefore, the training intensity was cus-
tomized to participant's physical function level by gradually increas-
ing the gait speed according to their subjective comforts.
Outcome Measures
Primary outcome of this study was locomotor function measured by
spatiotemporal parameters and gait symmetry ratio while walking.
Spatiotemporal parameters (gait speed, cadence, and stride length)
and gait symmetry (temporal symmetry ratio and spatial step sym-
metry ratio) were measured using a 3D motion capture system with
6 infrared cameras (Motion Analysis Corporation, Santa Rosa, CA).
Nineteen reflective markers in the Helen-Hayes marker set con-
figuration were used for this study.16 In addition, the muscle effort
symmetry ratio between the affected and unaffected sides was meas-
ured using an 8-channel wireless surface electromyography system
(sEMG; Noraxon Inc, Scottsdale, AZ). Gait symmetry ratios were
calculated using the following equations:
paretic� swing� time/� stance� time
Temporal symmetry ratio17 =
nonparetic� swing� time/stance� time
nonparetic step length
Spatial step symmetry ratio18 =
paretic� step� length
paretic� muscle� effort
Muscle effort symmetry ratio 4 =
nonparetic� muscle� effort
To obtain sEMG signals, electrodes were positioned bilaterally on
the rectus femoris (RF), biceps femoris, tibialis anterior, and the me-
dial gastrocnemius muscles in accordance with the recommendations
of the Non-Invasive Assessment of Muscles Project (SENIAM).17
Each site was prepared by shaving, abrading, and cleaning the area
with alcohol to reduce surface impedance. In addition, foot-switch
sensors were placed on the plantar surfaces of the toes and heels of
both feet. The signals from the foot-switch sensors recorded data for
the stance and swing phases while walking. We applied the method
described in a previous study that used sEMG muscle effort defined
as % maximum voluntary contraction (% MVC) averaged across
 Lee et al   Robotic Gait Training for Stroke Rehabilitation   3547

Figure 1. The robot configuration and controller. A, Configuration of the Gait Enhancing and Motivating System (GEMS). B, Overall control architecture for
assistance timing controller along with phase detection and torque generation (see Supplement I in the online-only Data Supplement). PASO indicates partic-
ularly shaped adaptive oscillators.

all participants.11,18 To normalize the sEMG signal amplitude, three
5-second maximal muscle contraction measurements were collected
from each muscle to determine each participant's MVC. sEMG sig-
nals were bandpass filtered between 10 and 350 Hz, and the root
mean squared (RMS) of the signals were calculated using a sliding
100 ms window for analysis. To verify the effect of locomotor func-
tion according to pre- and post-tests, we measured the MVC of all
muscles for all participants. After MVC assessment, participants
were instructed to walk at a self-selected speed along an 8 m length
walkway. Before formal measurements, practice sessions were per-
Downloaded from http://ahajournals.org by on November 1, 2020
group, paired t tests were used to compare data from pre- and post-
tests. To examine the main effects of the intervention, a 2 (group)×2
(time) ANOVA with repeated measures was performed. Relationships
between gait symmetry ratios and cardiopulmonary metabolic effi-
ciency were examined using the Pearson correlation coefficient (r).
All statistical analyses were undertaken using SPSS version 22.0
(IBM, Armonk, NY), and the level of significance was set at P<0.05.
Table 1.  Baseline Characteristics of Participants

formed to familiarize the participants with the procedure. Then, 5 Experimental Control Group
trials of walking data were acquired per participant over a walkway. Characteristics Group (n=14) (n=12) P Value
In addition, we measured cardiopulmonary metabolic energy effi-
ciency during 6 minutes of treadmill walking using a valid and reliable Gender (male/female) 7/7 7/5 0.722*
portable telemetric gas analyzer system (Cosmed K4B2, Rome, Italy) Age, y 61.85 (7.87) 62.25 (6.36) 0.573†
which was worn on the back, to measure breath-by-breath metabolic
energy costs. Heart rate was monitored via a wireless chest-strap dur- Height, cm 161.07 (6.01) 164.41 (6.72) 0.218†
ing treadmill walking. In the pretraining test, baseline cardiopulmo- Weight, kg 61.00 (9.56) 64.25 (8.95) 0.374†
nary metabolic energy efficiency was measured for 5 minutes in the
standing position before metabolic energy cost (mL·kg−1·min−1) was Stroke onset duration, d 1486 (264.12) 1536 (311.54) 0.088†
recorded during the treadmill gait test (6-minute walk test) without
Stroke type
GEMS at each participant's most comfortable gait speed. Participants
identified their most comfortable speeds by walking at their own pace  Ischemic/Hemorrhagic 10/4 8/4 0.559*
for 3 minutes while wearing a safety harness without using the hand-
rails. The post-training test was performed on the same treadmill at Stroke location
the same speed and using the same measurement method as the pre-  Cortical/Subcortical 12/2 10/2 0.311*
training test. The metabolic energy cost (mL·kg−1·min−1) was calcu-
lated as the average oxygen uptake during the last 2 minutes of each Side of stroke
test condition. Net metabolic energy cost was calculated by subtract-  Right/left 5/9 6/6 0.458*
ing standing oxygen demand from the average oxygen uptake during
the last 2 minutes of each test condition. All clinical assessments were MAS grade (0/1/1+/2)
performed at baseline and immediately after the intervention during  Hip (flexion/extension) 0/0/0/0 0/0/0/0 …
each of 10 sessions by the same examiners.
Secondary outcome includes motor and balance function assessed  Knee (flexion/extension) 0/2/0/0 0/1/0/0 …
by Fugl-Meyer Assessment scale—lower limb and Motricity Index
 Ankle (dorsi-/plantar-flexion) 0/5/1/0 0/4/2/0 …
of affected lower extremity, Berg Balance Scale, and Korean version
Fall Efficacy Scale. Detailed information of assessment method is K-MoCA, score 27.16 (1.25) 27.04 (1.03) 0.895†
described in Supplement II in the online-only Data Supplement.
FAC grade (3/4) 6/8 7/5 0.719*
Self-selected treadmill speed, m/s 0.68 (0.25) 0.62 (0.37) 0.653†
Statistical Analysis
The Shapiro-Wilk test was used to confirm that all outcome variables Values are expressed as mean (SD). FAC indicates functional ambulation
were normally distributed. We used independent t tests for contin- categories; K-MoCA, Korean version of the Montreal Cognitive Assessment; and
uous variables, Mann-Whitney U tests for ordinal variables, and ×2 MAS, Modified Ashworth Scale.
tests for categorical variables to compare participants' baseline char- *χ2 test.
acteristics between the experimental and control groups. Within each †Independent t test.
 3548  Stroke  December 2019

Table 2.  Locomotor Function Presenting Spatiotemporal Gait Parameters and Gait Symmetry in the Experimental and Control Groups

Experimental Group (n=14) Control Group (n=12) Between Groups

Variable Pretest Post-Test Pretest Post-Test P Value Effect Size
Spatiotemporal gait parameters
 Gait speed, m/s 0.62 (0.28) 0.91 (0.15)* 0.60 (0.14) 0.78 (0.11)† 0.001 0.190
 Cadence, step/min 78.6 (21.54) 106.88 (29.48)* 74.97 (25.67) 88.01 (22.82)† 0.001 0.095
 Stride length, cm 60.54 (15.97) 92.64 (15.46)* 61.15 (14.26) 80.64 (16.44)† 0.013 0.089
Gait symmetry ratio
 Temporal symmetry ratio 1.65 (0.28) 1.19 (0.23)† 1.62 (0.19) 1.35 (0.38) 0.031 0.088
 Spatial step symmetry ratio 1.82 (0.24) 1.22 (0.28)† 1.84 (0.30) 1.44 (0.27) 0.038 0.076
 Muscle effort (% MVC) symmetry ratio
  RF 2.68 (0.92) 1.47 (0.57)† 2.59 (1.02) 1.88 (0.64)† 0.017 0.052
  BF 2.45 (0.51) 1.27 (0.47)† 2.53 (0.49) 1.72 (0.57) <0.001 0.063
  TA 3.45 (0.46) 2.27 (0.76)† 3.62 (0.72) 3.22 (0.65) <0.001 0.058
  GCM 3.38 (0.33) 2.21 (0.42)† 3.26 (0.35) 3.10 (0.42) <0.001 0.022
Values are presented as mean (SD). BF indicates biceps femoris; GCM, medial of gastrocnemius; MVC, maximum voluntary contraction; RF, rectus femoris; and TA,
tibialis anterior.
*P<0.01 for within-group comparisons.
†P<0.05.

Results reduced by 14.73% in the experimental group (P<0.05) but

Participant baseline characteristics are presented in Table 1.
There were no significant differences between groups. A
CONSORT diagram showing excellent participant compli-
ance with the study protocol is presented in Supplement III in
Downloaded from http://ahajournals.org by on November 1, 2020
only by 3.05% in the control group. This demonstrated a sig-
nificant repeated measures group×time interaction, which
meant that training with GEMS was better than without
GEMS for improving cardiopulmonary metabolic efficiency

the online-only Data Supplement. after gait training (P<0.01).

Primary Outcome: Locomotor Function and
Cardiopulmonary Metabolic Energy Efficiency
The specific values for locomotor function before and after the
training intervention are presented in Table 2 and Supplement
IV in the online-only Data Supplement. After completion of
10 training sessions, spatiotemporal gait parameters including
gait speed, cadence, and stride length, were significantly
improved in both groups. In addition, significant group×time
interactions were found for all spatiotemporal gait parameters,
and the experimental group demonstrated greater improve-
ment than the control group (P<0.05).
Particularly, muscle efforts (% MVC) on the affected
side during the gait cycle were significantly improved after
gait training for the experimental group (Figure 2), but not
in the control group. In addition, positive results were seen
in all gait symmetry ratios including temporal step symmetry
by 27.88%, spatial step symmetry by 32.97% and a decrease
of muscle effort symmetries in RF (45.15%), biceps femo-
ris (48.16%), tibialis anterior (34.20%), and gastrocnemius
(34.62%) in the experimental group (P<0.05). In contrast,
only RF muscle effort symmetry was improved by 27.41%
in the control group (P<0.05). Significant group×time inter-
actions were observed for all gait symmetry ratios (P<0.05;
Table 2).
Changed values for net metabolic energy cost during
gait training were shown in Figure 3. After 10 intervention
sessions the net metabolic energy cost (mL·kg−1·min−1) was
Secondary Outcome: Motor and Balance
Changed values for motor and balance are shown in Supplement
V in the online-only Data Supplement. After completion of 10
training sessions, experimental group demonstrated greater
improvement in Fugl-Meyer Assessment scale—lower limb
and K-FES than the control group (P<0.05). Significant
group×time interactions were also observed for Fugl-Meyer
Assessment scale - lower limb and K-FES (P<0.05).
Relationship Between Gait Symmetry Ratio and
Cardiopulmonary Metabolic Energy Efficiency
The relationship between gait symmetry ratio and cardiopul-
monary metabolic efficiency is presented in Table 3. The re-
duction in net metabolic energy cost during gait training with
GEMS was significantly positively correlated with all gait
symmetry ratios (P<0.01).
Discussion
The findings from the current study suggest that gait training
with GEMS has several key advantages for locomotor func-
tion and efficiency compared with conventional gait training.
The spatiotemporal gait parameters including gait speed, ca-
dence, and stride length as well as muscle effort symmetries of
both legs were more improved after gait training with GEMS
than without GEMS. Furthermore, net metabolic energy costs
during walking were remarkably improved after gait training
with GEMS in patients with hemiparetic stroke.
 Lee et al   Robotic Gait Training for Stroke Rehabilitation   3549
Downloaded from http://ahajournals.org by on November 1, 2020

Figure 2. Diagrams showing differences in mean normalized muscle efforts (% MVC) during the gait cycle (%) measured by surface electromyography
(sEMG) pre- and post-training in the experimental (A) and control groups (B). BF indicates biceps femoris; GCM, medial of gastrocnemius; GEMS, Gait
Enhancing and Motivating System; MVC, maximum voluntary contraction; RF, rectus femoris; and TA, tibialis anterior.

Gait asymmetry is associated with potential negative fac-
tors in patients with hemiparetic stroke and includes unbal-
anced gait control, increased compensation by the nonparetic
side, increased metabolic energy consumption, and poorer
activities of daily living.3,19 Therefore, gait asymmetry is a
major factor influencing gait efficiency and is a meaningful
surrogate for biomechanical assessments of neuromotor
impairment and monitoring the effect of rehabilitation pro-
grams.20 A previous study demonstrated that normal persons
usually have a symmetrical gait pattern, and their spatiotem-
poral gait symmetry ratio ranges from 0.9 to 1.1.19 According
to the results of our study, the spatiotemporal gait symmetry
 3550  Stroke  December 2019

Figure 3. Metabolic energy cost result after gait training according to the intervention method. A, Changes in net metabolic energy cost (mL·kg−1·min−1) during
6 minutes of gait training with (experimental group) or without (control group) Gait Enhancing and Motivating System (GEMS). B, Different effects of gait train-
ing on net metabolic energy cost between the experimental and control groups. **P<0.01.

ratio of patients with stroke became closer to normal range
after gait training with GEMS (temporal symmetry ratio 1.19,
spatial step symmetry ratio 1.22). In contrast, the control
group showed higher symmetry ratios even after gait train-
ing; temporal symmetry ratio 1.35 and spatial step symmetry
ratio 1.44. These results lead us to postulate that gait training
with GEMS resulted in improvements of the temporal sym-
metry ratio by 27.88% and the spatial step symmetry ratio
Downloaded from http://ahajournals.org by on November 1, 2020
in previous studies.22–24 In particular, improvements in gait
speed and spatial step asymmetry after gait training contrib-
uted to improved cardiopulmonary metabolic energy effi-
ciency in patients with stroke.3 Recent studies have provided
insight into why improvements in energy expenditure and gait
compensatory gait pattern result from reductions in the mag-
nitude of spatial and temporal asymmetry after gait training
in patients with stroke.3,25 In our study, participants showed

by 32.99% in patients with stroke. Moreover, in the current a continuous increase in cardiopulmonary metabolic energy
study, improvement in gait symmetry was verified by the cost during a 6-minute treadmill walking before the interven-
muscle effort (% MVC) symmetry ratio. In the experimental tion. Interestingly, energy cost increased more rapidly from
group, gait training with GEMS reduced the hip extensor (bi- the third minute of walking. In contrast, in the experimental
ceps femoris) to flexor muscle (RF) effort symmetry ratio. group, energy cost did not show such a rapid increase from
More importantly, it induced changes in ankle strategy dur- third minute of walking to the completion of the interven-
ing walking, that is, improvement in ankle dorsi-flexor (tibi- tion. These results may suggest that gait training with GEMS
alis anterior) to ankle plantar flexor (gastrocnemius) muscle is effective for improving walking endurance by decreasing
effort symmetry ratio. We believe that GEMS assisted in energy depletion during sustained walking in patients with
creating a symmetrical pattern of hip flexor and extensor ac-
tivities during gait training and was more effective than con- Table 3.  Correlation between Cardiopulmonary Metabolic Energy Efficiency
ventional training for regaining muscle effort symmetry in and Gait Symmetry Ratios
the lower limbs in patients with hemiparetic stroke. In addi- ∆ Cardiopulmonary Metabolic Energy
tion, GEMS rhythmic assistance for hip movement of the Cost (r)
affected side might decrease the compensatory over-activity
Experimental Group Control Group
of the unaffected hip muscles, thereby leading to enhanced
gait symmetry. In fact, the MVC of the lower limb muscles Temporal symmetry ratio 0.667* 0.352
of the affected side was significantly improved after gait Spatial step symmetry ratio 0.611* 0.310
training with GEMS compared with the unaffected side in Muscle effort (% MVC) symmetry ratio
our study (please see Supplement VI in the online-only Data
 RF 0.715* 0.701†
Supplement). More interestingly, muscle efforts on the af-
fected side during walking were significantly improved after  BF 0.836* 0.299
gait training with GEMS as shown in Figure 2. Therefore, it  TA 0.701* 0.192
can be postulated that GEMS promoted use of the affected
 GCM 0.675* 0.158
lower limb during walking and facilitated paretic muscle ac-
tivities that helped to restore a symmetrical gait pattern after Values are Pearson correlation coefficients (r). ∆ Cardiopulmonary metabolic
energy cost: difference in metabolic energy cost values pretraining time and
gait training. post-training. BF indicates biceps femoris; GCM, medial of gastrocnemius; MVC,
Cardiopulmonary metabolic energy cost is a representa- maximum voluntary contraction; RF, rectus femoris; and TA, tibialis anterior.
tive indicator of locomotor efficiency.11,21 Reductions in en- *Significant correlation (P<0.01).
ergy cost have been associated with use of robotic devices †Significant correlation (P<0.05).
 Lee et al   Robotic Gait Training for Stroke Rehabilitation   3551
stroke (Figure 3A). Our results indicated that the GEMS
reduced net cardiopulmonary metabolic cost (mL·kg−1·min−1)
by ≈14.7% after gait training for four weeks (Figure 3B).
More importantly, net cardiopulmonary metabolic costs
exhibited strong relationships with all gait symmetry ratio
parameters. However, conventional gait training in the con-
trol group showed that net cardiopulmonary metabolic costs
exhibited meaningful relationships only with the RF muscle
effort ratio. According to the inverted pendulum model of lo-
comotion, improvements in metabolic energy efficiency result
from reductions in spatial step asymmetry if such reductions
facilitate equalization of collision and push-off forces during
the double support phase of gait cycle. Indeed, we found that
the muscle effort symmetry ratio for gastrocnemius associated
with push-off force improved after gait training with GEMS
and showed a strong correlation with improved metabolic en-
ergy efficiency.
Also, our study is the significant improvement observed
in only motor function (Fugl-Meyer Assessment scal—lower
limb) and fall efficacy such (K-FES) in the experimental
group. These results suggest that short-term gait training
with 10 sessions improved fall efficacy as well as motor and
locomotor function, but did not have a significant effect on
physical balance (Berg Balance Scale). Thus, more research
is required to investigate the effect of functional movement
training (sit to stand and stair ascent movement) with GEMS.
Recently, wearable robots were developed as a new
strategy to facilitate rehabilitation by providing gait training
in various environments including over ground, stairs, and up-
Downloaded from http://ahajournals.org by on November 1, 2020
hill with accurate and repetitive movements and motivation
benefits in patients with gait disorders.26 Previous interven-
tion studies with wearable robotic devices have demonstrated
improved gait function in patients with stroke.12,27,28 In our
study, training with GEMS was associated with remarkable
improvements in gait parameters, gait symmetry, and an ex-
cellent reduction of cardiopulmonary metabolic energy con-
sumption in patients with chronic stroke.
All participants in this study completed ten intervention
sessions with GEMS without adverse events. This indicates
that gait training with GEMS is safe and does not pose risks
to patients. In addition, participant attitudes regarding train-
ing with GEMS were mostly positive. However, several fe-
male participants complained that the robot was a little heavy.
There are some limitations to this study. First, the statistical
power was low because of the small number of participants,
therefore, our results cannot be generalized to the entire stroke
population. Second, long-term effects of gait training with
GEMS were not considered. Long-term training with fol-
low-up in larger participant cohorts will be conducted in the
future. Finally, further studies are needed to demonstrate the
effects of GEMS gait training in the early stages of stroke.
The feasibility of using this robot in home settings must be
investigated as well.
To our knowledge, this is the first randomized controlled
trial to consider locomotor function and efficiency after gait
training with a wearable hip-assist robot, the GEMS. Our find-
ings provide evidence that supports implementing the GEMS
in locomotion rehabilitation programs for patients with hemi-
paretic stroke.
Sources of Funding
This study was supported by Samsung Medical Center (PHO0171341)
and by grants from the National Research Foundation (NRF; NRF-
2016R1A6A3A11930931, NRF-2017R1A2A1A05000730, and
NRF-2017M3A9G5083690), funded by the Korean government.
Disclosures
None.
References
1. Franceschini M, Carda S, Agosti M, Antenucci R, Malgrati D,
Cisari C; Gruppo Italiano Studio Allevio Carico Ictus. Walking after
stroke: what does treadmill training with body weight support add to
overground gait training in patients early after stroke?: a single-blind,
randomized, controlled trial. Stroke. 2009;40:3079–3085. doi:
10.1161/STROKEAHA.109.555540
2. Veerbeek JM, Koolstra M, Ket JC, van Wegen EE, Kwakkel G. Effects of
augmented exercise therapy on outcome of gait and gait-related activities
in the first 6 months after stroke: a meta-analysis. Stroke. 2011;42:3311–
3315. doi: 10.1161/STROKEAHA.111.623819
3. Awad LN, Palmer JA, Pohlig RT, Binder-Macleod SA, Reisman DS.
Walking speed and step length asymmetry modify the energy cost of
walking after stroke. Neurorehabil Neural Repair. 2015;29:416–423.
doi: 10.1177/1545968314552528
4. Patterson KK, Gage WH, Brooks D, Black SE, McIlroy WE. Evaluation
of gait symmetry after stroke: a comparison of current methods and rec-
ommendations for standardization. Gait Posture. 2010;31:241–246. doi:
10.1016/j.gaitpost.2009.10.014
5. Zadravec M, Olensek A, Rudolf M, Bizovicar N, Goljar N, Matjacic Z.
A novel robot-assisted training approach for improving gait symmetry
after stroke. IEEE Int Conf Rehabil Robot. 2017;2017:222–227. doi:
10.1109/ICORR.2017.8009250
6. Schwartz I, Sajin A, Fisher I, Neeb M, Shochina M, Katz-Leurer M, et
al. The effectiveness of locomotor therapy using robotic-assisted gait
training in subacute stroke patients: a randomized controlled trial. PM R.
2009;1:516–523. doi: 10.1016/j.pmrj.2009.03.009
7. Bertolucci F, Di Martino S, Orsucci D, Ienco EC, Siciliano G, Rossi B,
et al. Robotic gait training improves motor skills and quality of life in
hereditary spastic paraplegia. NeuroRehabilitation. 2015;36:93–99. doi:
10.3233/NRE-141196
8. Kim SY, Yang L, Park IJ, Kim EJ, JoshuaPark MS, You SH, et al. Effects
of innovative WALKBOT robotic-assisted locomotor training on balance
and gait recovery in hemiparetic stroke: a Prospective, Randomized,
Experimenter Blinded Case Control Study with a four-week follow-
up. IEEE Trans Neural Syst Rehabil Eng. 2015;23:636–642. doi:
10.1109/TNSRE.2015.2404936
9. Wall A, Borg J, Palmcrantz S. Clinical application of the Hybrid Assistive
Limb (HAL) for gait training-a systematic review. Front Syst Neurosci.
2015;9:48. doi: 10.3389/fnsys.2015.00048
10. Chen G, Chan CK, Guo Z, Yu H. A review of lower extremity assistive
robotic exoskeletons in rehabilitation therapy. Crit Rev Biomed Eng.
2013;41:343–363.
11. Lee HJ, Lee S, Chang WH, Seo K, Shim Y, Choi BO, et al. A wearable
hip assist robot can improve gait function and cardiopulmonary meta-
bolic efficiency in elderly adults. IEEE Trans Neural Syst Rehabil Eng.
2017;25:1549–1557. doi: 10.1109/TNSRE.2017.2664801
12. Buesing C, Fisch G, O'Donnell M, Shahidi I, Thomas L,
Mummidisetty CK, et al. Effects of a wearable exoskeleton stride man-
agement assist system (SMA®) on spatiotemporal gait characteristics
in individuals after stroke: a randomized controlled trial. J Neuroeng
Rehabil. 2015;12:69. doi: 10.1186/s12984-015-0062-0
13. Raab K, Krakow K, Tripp F, Jung M. Effects of training with the
ReWalk exoskeleton on quality of life in incomplete spinal cord in-
jury: a single case study. Spinal Cord Ser Cases. 2016;2:15025. doi:
10.1038/scsandc.2015.25
14. Calabrò RS, Cacciola A, Bertè F, Manuli A, Leo A, Bramanti A, et
al. Robotic gait rehabilitation and substitution devices in neurolog-
ical disorders: where are we now? Neurol Sci. 2016;37:503–514. doi:
10.1007/s10072-016-2474-4
15. Mehrholz J, Wagner K, Rutte K, Meissner D, Pohl M. Predictive validity
and responsiveness of the functional ambulation category in hemiparetic
patients after stroke. Arch Phys Med Rehabil. 2007;88:1314–1319. doi:
10.1016/j.apmr.2007.06.764
 3552  Stroke  December 2019
16. Collins TD, Ghoussayni SN, Ewins DJ, Kent JA. A six degrees-of-
freedom marker set for gait analysis: repeatability and comparison
with a modified Helen Hayes set. Gait Posture. 2009;30:173–180. doi:
10.1016/j.gaitpost.2009.04.004
17. Alexander LD, Black SE, Patterson KK, Gao F, Danells CJ,
McIlroy WE. Association between gait asymmetry and brain lesion
location in stroke patients. Stroke. 2009;40:537–544. doi: 10.1161/
STROKEAHA.108.527374
18. Patterson KK, Parafianowicz I, Danells CJ, Closson V, Verrier MC,
Staines WR, et al. Gait asymmetry in community-ambulating stroke
survivors. Arch Phys Med Rehabil. 2008;89:304–310. doi: 10.1016/j.
apmr.2007.08.142
19. Merletti R, Hermens H. Introduction to the special issue on the
SENIAM European concerted action. J Electromyogr Kinesiol. 2000;10:
283–286.
20. Bruno Garza JL, Eijckelhof BH, Johnson PW, Raina SM, Rynell PW,
Huysmans MA, et al. Observed differences in upper extremity forces,
muscle efforts, postures, velocities and accelerations across computer ac-
tivities in a field study of office workers. Ergonomics. 2012;55:670–681.
doi: 10.1080/00140139.2012.657692
21. Balasubramanian CK, Bowden MG, Neptune RR, Kautz SA. Relationship
between step length asymmetry and walking performance in subjects
with chronic hemiparesis. Arch Phys Med Rehabil. 2007;88:43–49. doi:
10.1016/j.apmr.2006.10.004
22. Galle S, Malcolm P, Collins SH, De Clercq D. Reducing the meta-
bolic cost of walking with an ankle exoskeleton: interaction between
actuation timing and power. J Neuroeng Rehabil. 2017;14:35. doi:
10.1186/s12984-017-0235-0
Downloaded from http://ahajournals.org by on November 1, 2020
23. Israel JF, Campbell DD, Kahn JH, Hornby TG. Metabolic costs and
muscle activity patterns during robotic- and therapist-assisted treadmill
walking in individuals with incomplete spinal cord injury. Phys Ther.
2006;86:1466–1478. doi: 10.2522/ptj.20050266
24. Awad LN, Bae J, Kudzia P, Long A, Hendron K, Holt KG, et al.
Reducing circumduction and hip hiking during hemiparetic walking
through targeted assistance of the paretic limb using a soft robotic exo-
suit. Am J Phys Med Rehabil. 2017;96(10 suppl 1):S157–S164. doi:
10.1097/PHM.0000000000000800
25. Lee SY, Han EY, Kim BR, Chun MH, Lee YK. Can lowering the guid-
ance force of robot-assisted gait training induce a sufficient metabolic
demand in subacute dependent ambulatory patients with stroke? Arch
Phys Med Rehabil. 2017;98:695–700. doi: 10.1016/j.apmr.2016.10.021
26. Kawamoto H, Kadone H, Sakurai T, Sankai Y. Modification of hemi-
plegic compensatory gait pattern by symmetry-based motion controller
of HAL. Conf Proc IEEE Eng Med Biol Soc. 2015;2015:4803–4807. doi:
10.1109/EMBC.2015.7319468
27. Chen B, Ma H, Qin LY, Gao F, Chan KM, Law SW, et al. Recent devel-
opments and challenges of lower extremity exoskeletons. J Orthop
Translat. 2016;5:26–37. doi: 10.1016/j.jot.2015.09.007
28. Kawamoto H, Kamibayashi K, Nakata Y, Yamawaki K, Ariyasu R,
Sankai Y, et al. Pilot study of locomotion improvement using hybrid as-
sistive limb in chronic stroke patients. BMC Neurol. 2013;13:141. doi:
10.1186/1471-2377-13-141
29. Li L, Ding L, Chen N, Mao Y, Huang D, Li L. Improved walking
ability with wearable robot-assisted training in patients suffering
chronic stroke. Biomed Mater Eng. 2015;26(suppl 1):S329–S340. doi:
10.3233/BME-151320