Archives of Physical Medicine and Rehabilitation

journal homepage: www.archives-pmr.org

Archives of Physical Medicine and Rehabilitation 2014;95:1878-87

ORIGINAL ARTICLE

Understanding Therapeutic Benefits of Overground

Bionic Ambulation: Exploratory Case Series in Persons
With Chronic, Complete Spinal Cord Injury
Jochen Kressler, PhD,a Christine K. Thomas, PhD,a,b,c Edelle C. Field-Fote, PT, PhD, FAPTA,a,b,d,e
Justin Sanchez, PhD,a,f Eva Widerstrom-Noga, DDS, PhD,a,b,d Deena C. Cilien, BS,a
Katie Gant, MS,a Kelly Ginnety, MS,a Hernan Gonzalez, BS,f Adriana Martinez, BS,a
Kimberley D. Anderson, PhD,a Mark S. Nash, PhD, FACSMa,b,d,g
From aThe Miami Project to Cure Paralysis, and Departments of bNeurological Surgery, cPhysiology and Biophysics, dRehabilitation Medicine,
e
Physical Therapy, fBiomedical Engineering, and gMedicine, Miller School of Medicine, University of Miami, Miami, FL.

Abstract
Objective: To explore responses to overground bionic ambulation (OBA) training from an interdisciplinary perspective including key components
of neuromuscular activation, exercise conditioning, mobility capacity, and neuropathic pain.
Design: Case series.
Setting: Academic research center.
Participants: Persons (NZ3; 2 men, 1 woman) aged 26 to 38 years with complete spinal cord injury (SCI) (American Spinal Injury Association
Impairment Scale grade A) between the levels of T1 and T10 for !1 year.
Intervention: OBA 3d/wk for 6 weeks.
Main Outcome Measures: To obtain a comprehensive understanding of responses to OBA, an array of measures were obtained while walking in
the device, including walking speeds and distances, energy expenditure, exercise conditioning effects, and neuromuscular and cortical activity
patterns. Changes in spasticity and pain severity related to OBA use were also assessed.
Results: With training, participants were able to achieve walking speeds and distances in the OBA device similar to those observed in persons
with motor-incomplete SCI (10-m walk speed, .11e.33m/s; 2-min walk distance, 11e33m). The energy expenditure required for OBA was similar
to walking in persons without disability (ie, 25%e41% of peak oxygen consumption). Subjects with lower soleus reflex excitability walked longer
during training, but there was no change in the level or amount of muscle activity with training. There was no change in cortical activity patterns.
Exercise conditioning effects were small or nonexistent. However, all participants reported an average reduction in pain severity over the study
period ranging between "1.3 and 1.7 on a 0-to-6 numeric rating scale.
Conclusions: OBA training improved mobility in the OBA device without significant changes in exercise conditioning or in neuromuscular or
cortical activity. However, pain severity was reduced and no severe adverse events were encountered during training. OBA therefore opens the
possibility to reduce the common consequences of chronic, complete SCI such as reduced functional mobility and neuropathic pain.
Archives of Physical Medicine and Rehabilitation 2014;95:1878-87
2014 by the American Congress of Rehabilitation Medicine

The desire to walk ranks among the most prevalent concerns related
to mobility identified by people with paraplegia caused by spinal
cord injury (SCI).1-4 Other complications related to the injury
include muscle atrophy, reductions in bone mineral density, skin
breakdown, urinary tract infections, spasticity, impaired lymphatic,
Disclosures: none.

vascular, and digestive function, and reduced cardiorespiratory capacity.5 Beyond these complications, persistent pain may negatively
affect mobility and activity levels.6,7
Approaches to mitigate these issues now include so-called
bionic devices such as robotic exoskeletons. Fully powered lower
extremity exoskeletons that facilitate movement with electric
motors are being developed by several groups.8,9 Different aspects

0003-9993/14/$36 - see front matter 2014 by the American Congress of Rehabilitation Medicine
http://dx.doi.org/10.1016/j.apmr.2014.04.026

Bionic walking: exploratory case series

of structural stability (ie, supporting persons with various neurologic levels of injury), device weight, and functional capacity (eg,
feedback, stair stepping, incline walking) are emphasized by the
various systems.9,10 Commercially available exoskeletal walking
devices must be used in conjunction with assistive devices for
balance, such as a walker or crutches.
The structure and mechanics of bionic exoskeletal walking
devices have been described from an engineering perspective, and
potential applications for civilian and military use have been
proposed.11,12 However, clinical data on user performance and
physiological responses to walking in these devices are limited.
Only a few studies13-18 have been published in the peer-reviewed
literature to date, none of which reported training that had been
standardized with regard to walking time, walking distance, or
pretraining values to assess potential pre-post changes with
training. While these studies represent a reasonable first attempt at
evaluating robotic exoskeletons, the available data are limited to
device use for mobility. Unexplored issues include whether
secondary complications of SCI are influenced by overground
bionic ambulation (OBA), specifically, whether these devices will
be most effective for improving neuromuscular activation,
exercise conditioning, or as assistive technology to improve
mobility. We therefore adopted a cross-disciplinary approach with
the objective to explore multifaceted responses to OBA training
with a bionic exoskeleton in a convenience sample of 3 participants with chronic, complete SCI. Key components focused on
were brain and muscle electrical activity, exercise conditioning as
measured by peak oxygen consumption (VO2peak), and the potential effects on markers of cardiometabolic risk, mobility capacity (ie, walking speeds as assessed by the 10-m walk test and
distances as assessed by the 2-minute walk test achieved while
using the device), and neuropathic pain.

Methods
Inclusion/exclusion criteria and informed consent
To be included in the study, subjects had to have motor-complete
SCI for !1 year. To fit the exoskeleton, subjects were required to
have a height of 1.56 to 1.9m and a weight of #100kg, with a hip
width of #.45m as measured across the frontal plane. Further
details are provided in supplemental appendix S1 (available online
only at http://www.archives-pmr.org/). The exclusion criteria were
as follows: (1) any surgery within the preceding 3 months
(assessed by participant self-report); (2) recent lower extremity
fracture (assessed by radiographs of the lower extremity bones and
joints in anteroposterior and lateral views); (3) participation in
lower limb exercise conditioning within past 3 months; (4)
occurrence of a pressure ulcer within past 3 months; (5) upper
limb pain that limited weight-bearing on forearm crutches; (6)
pregnancy; (7) exercise contraindications of the American College

List of abbreviations:
EEG
MG
MVC
NRS
OBA
SCI
TEE
VO2peak

electroencephalography
medial gastrocnemius
maximal voluntary contraction
numeric rating scale
overground bionic ambulation
spinal cord injury
total energy expenditure
peak oxygen consumption

www.archives-pmr.org

1879
of Sports Medicine; (8) type I or II diabetes defined by American
Diabetes Association guidelines (assessed by fasting blood
glucose analysis described below); (9) lower extremity spasticity
score in any joint exceeding 3 out of 4 (Modified Ashworth
Scale19); (10) unresolved deep vein thrombosis; (11) uncontrolled
autonomic dysreflexia (by self-report); or (12) significant leg
length discrepancies (>.13m of the upper leg or >.19m of the
lower leg.
Written and verbal informed consent was obtained from all
participants. The protocol was approved by the Human Subjects
Research Office, Miller School of Medicine, University of Miami.

Participants
Participants were 2 men and 1 woman aged 26 to 38 years with
complete SCI (American Spinal Injury Association Impairment
Scale grade A) between the levels of T1 and T10. Participant
details are summarized in supplemental table S1 (found in
supplemental appendix S1).

Device description
Detailed descriptions of the bionic device (Eskoa) have been
published elsewhere.20,21 A brief description is provided in
supplemental appendix S1, and the device is shown in supplemental
figure S1 (found in supplemental appendix S1).

Protocol
Subjects participated in 18 sessions of OBA, walking around a
24.4-m oblong track (walking direction reversed for each session)
for 1 hour per session, 3 times per week using a walker for
balance. All subjects walked with close contact guard provided by
a trainer, while tethered to an overhead track that would provide
support in the event of a fall. Missed sessions because of schedule
conflicts or equipment issues were made up during the next week
or extending the protocol as needed (all subjects completed the
protocol within 47e49d). Subject 1 missed sessions 14, 15, and
18; subject 2 missed sessions 13 through 16; and subject 4 missed
sessions 1 and 2. Subjects were encouraged to walk the full hour
but were allowed to rest (seated or standing while balanced by
investigators) as desired.
The Ekso allows for 3 different modes of walking: (1) FirstStep, wherein the investigator actuates steps; (2) ActiveStep,
wherein participants actuate their own steps via buttons on the
walker; and (3) ProStep, wherein participants actuate the subsequent step by moving their hips forward and shifting them laterally, upon which the device triggers the step. Participants were
progressed through each mode based on their walking quality as
assessed by the investigator and the subjects own feedback (ie,
expressed desire to progress to the next mode).

Outcome measures
Assessments were performed at baseline, at midpoint, and in
the last week to assess functional walking capacity, spasticity, and
VO2peak, and to analyze blood and diet. Some assessments were
performed during the training sessions to evaluate concurrent
responses or immediate training-related change. Subjects arrived
at the laboratory in the fed state and underwent a series of assessments for pain before and after OBA, and spinal reflex
excitability before OBA (see details below). During OBA, muscle
and brain activity, as well as metabolic measures, was acquired
(details below).

1880
To allow comparison with previously published studies, clinical measures of functional walking capacity were obtained using
the 10-m walk test and the 2-minute walk test22 while in the
exoskeleton. These measures were obtained during a nonassessment OBA session (ie, not 1st, 9th, or 18th session) within the
week of the baseline, midpoint, and final assessments.
The effects of OBA on clinical measures of spasticity were
assessed with the Spinal Cord Assessment Tool for Spastic Reflexes23 on a non-OBA training day at baseline, at midpoint, and in
the last week. Spinal reflex excitability was assessed by normalizing the maximal soleus H-reflex/M-wave. Larger H-reflex/
M-wave ratios are associated with greater excitability of the
stretch reflex24 and are often used as a measure of spasticity.25
Muscle activity during OBA was assessed from continuous
unilateral recordings of electromyographic activity from 5 paralyzed leg muscles (under no voluntary control; medial gastrocnemius [MG], soleus, vastus lateralis, tibialis anterior, biceps
femoris) and 2 arm muscles that could be activated voluntarily
(triceps brachii, wrist flexors). Details are provided in
supplemental appendix S1.
Brain activity was assessed by electroencephalography
(EEG) for its relation to control of the OBA.26,27 A hook-andloop fastener headband anchoring straps of surface electrodes
was fitted to the subjects head and worn throughout the OBA
sessions. Brain control commands were collected from a 10channel wireless EEG systemb (sample rate 256Hz, 16 bits of
resolution), with electrodes placed on the premotor, motor,
sensorimotor, and frontal areas using the international 10- to 20electrode placement system and linked mastoid reference. Details on signal quality assessment and artifact correction are
described in supplemental appendix S1. After impedance and
artifact rejection analysis, it was determined that only subjects 2
and 4 produced high-quality recordings suitable for follow-up
analysis. For the remaining subject, spectral modulation in the
ongoing EEG was derived using a fifth-order Butterworth filter
between 1 and 35Hz and computing the power (1-s windows) in
the delta, theta, alpha, low-beta (13e18Hz) and high-beta (18e
25Hz) bands.28 To analyze the relationship between OBA
training and cortical activation, the power in each band was then
plotted topographically to show cortical activation during robotassisted walking.29
The VO2peak was determined via a maximal, continuous
graded exercise test on an arm crank ergometerc on a non-OBA
day at baseline and in the last week, as previously described,30
with the exception that stages were increased in increments of
30W. The same system as for the graded exercise test was used to
analyze gases during OBA, and heart rate was assessed by a
simple, integrated heart rate monitor (Polar WearLinkd). Substrate
utilization parameters and caloric expenditure were calculated as
previously described.31 In addition, blood markers for cardiometabolic risk were assessed immediately before and after a
nonassessment OBA session at baseline and in the last week,
respectively, by standard protocols described in supplemental
appendix S1.
Participants were interviewed regarding pain and pain interference with sleep using the International SCI Basic Pain
Dataset32 and the Multidimensional Pain Inventory (SCI version)
subscale for pain severity.7 A brief description of these measures
is provided in supplemental appendix S1. The Neuropathic Pain
Symptom Inventory33 evoked pain subscale was used to quantify
any self-reported static and mechanical allodynia, and cold
allodynia.

J. Kressler et al
Dynamic mechanical allodynia was investigated by quantitative sensory testing34 using a soft brush and lightly brushing the
skin of the dermatomes at and 4 segments below the level of
injury. Assessment of thermal allodynia was made using 2 thermorollers, 1 set at 40$ C (7$ C above normal skin temperature,
which is about 33$ C) and the other at 25$ C. If pain was evoked in
the test area, the participant was asked to rate the pain intensity by
using a numeric rating scale (NRS).

Results
OBA mobility measures
All participants increased the number of steps taken and distance
walked per session in the device (fig 1). For subjects 1 and 4, these
increases were close to linear throughout the intervention period,
while subject 2 plateaued after session 11. Further details on
subjects OBA performance are provided in supplemental
appendix S1.

Clinical outcome measures

Functional walking capacity increased 2- to 3-fold for subjects 1,
2, and 4 over the training period (fig 2).
Spasticity
No changes in clinical measures of spasticity (Spinal Cord
Assessment Tool for Spastic Reflexes) beyond what could be
attributable to typical variability were observed (table 1). The
exception was subject 2, who demonstrated moderate ankle clonus
at baseline, which was reduced to mild clonus at midpoint and
final testing. All other measures indicated no spasticity or only
mild spasticity.

Electrophysiological measures
During walking, both MG and biceps femoris were activated
involuntarily near the end of each swing phase, but other leg
muscles were largely quiet. Some activation of MG also continued
throughout stance. In contrast, elbow extensors and wrist flexors
were voluntarily activated throughout the step cycle (see
supplemental fig S2, found in supplemental appendix S1). No
consistent changes were observed in the activation patterns for
either the leg or arm muscles during the test sessions or as a result
of the OBA training. Muscle activity during OBA was limited in
leg muscles in terms of both duration of activation (<5min/session
for all muscles and subjects except in 2 cases; fig 3A) and activation intensity (<2% of the maximal M-wave; fig 3B). Arm
muscles were activated almost continuously during walking with
intensities ranging from approximately 1% to 7% and from 1% to
3% of the maximal M-wave for the elbow extensors and wrist
flexors, respectively (see fig 3B). Maximal H-reflex/M-wave soleus ratios varied from 0 to .85 (session 1), indicating a wide range
of reflex excitability across subjects by physiological measures,
but there was no measurable effect from OBA training (H-reflex/
M-wave ratio range: 0e.65, session 18). Soleus H-reflex/M-wave
ratios were related to walking duration, however. Subjects with
lower reflex excitability walked for a longer time during test
training sessions (fig 3C).
Spectral EEG analysis (1e35Hz) of subjects 2 and 4 revealed
a temporal evolution in the spatial localization of activation
during OBA. These changes occurred both within a session
and across sessions (fig 4), wherein the normalized low-beta
www.archives-pmr.org

Bionic walking: exploratory case series

Fig 1

1881

Mobility measures for OBA sessions for each participant. Up Time, time in upright posture in the exoskeleton. # Z unit (number).

(13e18Hz) magnitude is presented for subject 2. Beta EEG activity is associated with active concentration and motor movement changes.
For subject 2, cortical activation across 3 sessions is presented
in the bottom row of figure 4, while row 1 represents the activation
during the first session, and row 2 represents the activation during
session 9. Initially during the first session, beta activity was
focused bilaterally in the parietal regions extending forward into
the motor cortex, indicating strong sensorimotor activation during
interaction with the robotic exoskeleton. In contrast during session
9, activation moved to the frontal cortex where a large cognitive
component was likely active. Lastly in session 18, there was
still activation in the central frontal region; however, it was
complemented with large activation over the electrode position
C3, indicating hand activation. This activity is understandable as
www.archives-pmr.org

subjects use the ActiveStep mode of the robot to trigger steps with
their hands. Observation of the evolution of the activity in the first
session shows that initially strong parietal activation momentarily
moves frontal-central. By the end of the first session, the activity
was strongest in the parietal region but was less focal. The spatial
location of the largest magnitude of the EEG moved to the C3/C4
and F3/F4 regions of the head (standard electrode positions for
EEG recordings).36 During session 9, very minimal changes in the
spread of cortical activation occur over the session. It was not
possible to create a within-session analysis of session 18 because
there were insufficient clean data from this recording. To quantify
the relative changes in cortical activation, the magnitudes of the
beta activity for all electrodes and sessions are presented in
table 2. Over sessions or time within a session, activation moves
from the parietal to hand regions of the cortex.

1882

J. Kressler et al
ratings of their worst pain, whereas subject 4 reported a slight
increase in pain intensity for the worst pain at the last session
compared with baseline. Reductions in pain intensity were
consistent, with declines in sleep interference for subjects 1 and 2.
Not all participants reported evoked pain in the painful area on the
Neuropathic Pain Symptom Inventory (0e10) or experienced pain
in response to mechanical and thermal stimulation at their level of
injury. Further details on these aspects of pain are provided in
supplemental appendix S1.

Discussion
In this case series, learning to use the exoskeletal device was
associated with improved walking speed and distance; however,
subjects remained under contact guard and tether. Changes in
exercise conditioning and muscle activation were minimal.
Cortical activity indicated no appreciable recovery of motor
control but rather processes associated with learning how to use
the device. Pain severity was reduced for all participants, and no
severe adverse events were encountered.

Fig 2 OBA walking capacity assessments of maximal 10-m walking

speed and 2-minute walking distance. *Minimal clinically important
differences for pre-post changes (!.06m/s35). Abbreviation: S,
subject.

Exercise conditioning measures

OBA mobility

For VO2peak, changes from baseline to the end of the OBA

intervention were inconsistent across subjects ("10%, 5%, and
7% for subjects 1, 2, and 4, respectively; table 3). Similarly,
changes from the 1st to 9th and 18th sessions for measures of
exercise intensity, substrate oxidation rates/ratios, and total energy
expenditure (TEE) during OBA were inconsistent across subjects
and of small magnitude (see table 3). Nevertheless, OBA was a
sufficient stressor to substantially elevate heart rate, VO2, TEE,
and fat oxidation above seated rest (see table 3). Combined with
the increased speed, the unchanged energy cost resulted in
improved walking economy for all participants (fig 5).
Cardiometabolic risk markers were mostly unchanged except
HOMA-IR, which declined consistently by 1e2 points from
baseline to the last week (median, 48; minimumemaximum, 25%e
60%). Details are reported in supplemental appendix S1 (see
supplemental tables S2 and S3 in supplemental appendix S1).

OBA speeds improved for all subjects and were largely within the
range of those reported by others who have assessed the use of
OBA devices.14,15 The fastest walker in our study (subject 1)
showed that it is possible to walk approximately 1km with the
device and achieve stepping totals close to 2500 steps within 1
hour (the threshold for healthy people considered to be at least
low active vs sedentary is 5000 steps/d37). No other study has
reported data on sustained OBA performance beyond 6 minutes or
100m. Given the trend toward a continued increase in walking
speed in this individual (see figs 1 and 2), there appears to be room
for even further improvement. However, not all subjects were able
to achieve such high levels of OBA performance.
Despite the differences in OBA proficiency, even the slowest
walker was able (after training) to achieve walking speeds that
are comparable to averages achieved by those with motorincomplete injuries walking at a slow self-selected pace.38
Whether these results translate into mitigating losses in functional mobility that are associated with the many secondary issues
that arise after SCI9,39-41 remains to be determined in future
studies.

Neuropathic pain
All participants experienced at-level neuropathic pain (2 participants in the severe category, ie, NRSZ7; 1 participant in the
moderate category, NRSZ4). All participants rated this pain as
the worst pain type experienced. The pain severity scores were
lower at the last session compared with baseline for all subjects
(see supplemental fig S3, found in supplemental appendix S1).
This was also the case for subjects 1 and 2 with respect to their

Table 1

Neuromuscular activation
The extent (duration, intensity) to which paralyzed leg muscles are
excited reflexly during OBA will determine the potential for leg

Spasticity as measured by Spinal Cord Assessment Tool for Spastic Reflexes

Ankle Clonus
Baseline

Midpoint

Flexor Spasm
Last Week

Baseline

Midpoint

Extensor Spasm
Last Week

Baseline

Midpoint

Last Week

Subject No.

1
2
3
4

0
2
1
0

0
2
2
0

0
1
NA
0

0
1
NA
0

0
1
NA
1

0
1
NA
0

0
0
2
1

0
0
1
1

0
0
NA
1

0
1
NA
1

0
0
NA
1

0
1
NA
1

0
0
2
0

0
0
2
0

0
0
NA
0

0
0
NA
0

0
0
NA
0

0
0
NA
0

NOTE. The Spinal Cord Assessment Tool for Spastic Reflexes is scored on scales from 0e3 as detailed in Benz et al.23 Generally, lower ratings indicate less
spasticity.
Abbreviations: L, left; NA, not applicable; R, right.

www.archives-pmr.org

Bionic walking: exploratory case series

Fig 3 Duration (A) and intensity (B) of muscle activity across

muscles, subjects (for each muscle, data are presented, left to right, for
subjects 1, 2, and 4, respectively), and sessions (symbols as in C). (C)
Soleus H-reflex/M-wave ratio across subjects and sessions in relation to
walking duration (yZ"29.167& 55.272, R2Z.81). Data from each
subject are circled and numbered. Abbreviations: BF, biceps femoris;

www.archives-pmr.org

1883
muscle conditioning, amelioration of spasticity, or both. Activation of leg muscles during OBA was limited in all subjects. Activity in most leg muscles lasted <5 minutes and was at an
intensity <2% of the maximal M-wave during a 1-hour training
session (see fig 3A, B). In comparison, able-bodied individuals
engage leg muscles during approximately 30% to 55% of the step
cycle during walking (w18e33min in 1h) at intensities w6% of
maximal voluntary contraction (MVC).42,43 Nevertheless, involuntary muscle activation may have impeded mobility in the device, as physiological assessments of reflex excitability showed
that subjects with a lower soleus H-reflex/M-wave ratio walked for
a greater time during training (see fig 3C). Arm muscles were
active for almost the entire duration of OBA, and intensities
reached approximately 7% of maximal M-wave (w45% MVC).
The amount of triceps brachii muscle activation necessary to
correctly position oneself during OBA varied widely across subjects (from 1.3% to 6.6% maximal M-wave or from 11% to 45%
MVC; see fig 3B), but their engagement of wrist flexors was
similar. Subjects 2 and 4 generally had higher levels of triceps
brachii muscle activation during walking compared with subject 1,
and they had to repeatedly interrupt the session because of upper
extremity fatigue. These subjects constantly tried to compensate
for either poor balance (subject 2) or fear of falling, despite the
tether and guarding (subject 4). It is therefore conceivable that the
limiting muscular factors of OBA are upper extremity muscle
strength and ultimately muscle endurance, particularly for the
triceps brachii,44 as well as leg muscle reflex excitability (see fig
3C). No training effect was observed in any of the electromyographic activity measures, and walking in this exoskeleton did not
result in any appreciable change in clinical (see table 1) or
physiological measures of reflex excitability (see fig 3C).
Repeated exposure to passive movements (ie, cycling) has been
reported to reduce spasticity,45,46 and some inconsistent data
indicating the potential for a reduction in spasticity have been
published for OBA.15 It is possible that these effects may occur in
a subpopulation of persons with SCI, that the reflex muscle activation measured in this study was insufficient to induce changes,
particularly when movement of the ankle joint is restricted, or that
18 training sessions were an insufficient dose to observe
these effects.
In general, the EEG activity related to OBA was associated
with the processes of learning to control the device rather than a
change in motor function related to control of the lower extremities. By assessing the changes in the spatial arrangement of EEG
activity over time, it was observed that the subject begins with
widespread activity in the sensorimotor and parietal regions. This
activation is likely due to the initial sensory feedback of device
use. Over time (both within a session and across sessions), the
focus of EEG activity changes to the frontal regions of the cortex,
which may indicate the more cognitive processes associated with
active engagement of learning how to appropriately ambulate
within the device constraints. Over time and especially in the later
sessions, this activity localized to the hand regions of the primary
motor cortex. The activity in session 18 is maximal in the area of
the C3 electrode, which corresponds to the area representing right
hand movement and could be indicative of a beta rebound

EMG, electromyographic activity; H/M, H-reflex/M-wave; %Mmax,

percent maximal direct motor response; SL, soleus; TA, tibialis anterior;
TB, triceps brachii; VL, vastus lateralis; WF, wrist flexors.

1884

J. Kressler et al

Fig 4 Normalized power in (mV2) for subject 2 over 3 sessions and over time in each session at the beginning, middle, and end of each session.
The bottom row compares over the 3 sessions.

oscillations of event-related synchronization activity. These

changes suggest that the learning process of OBA use is one in
which the subject is learning to primarily activate the motor circuits necessary to press the button to step in ActiveStep mode. In
line with unchanged electromyographic activity patterns, these
results would indicate that subjects are likely learning to operate
the device as a mobility tool rather than eliciting a rehabilitative effect.

Exercise conditioning
No previous study has reported data on work intensities, but others
have speculated that subjects work harder during OBA than do
able-bodied individuals during walking,15 and arm muscles were
activated continuously at high intensities during OBA in this study
(see fig 3B). In contrast, our data showed energy expenditures
(1.4e2.6kcal/min after training) and intensities (25%e41% of

www.archives-pmr.org

Bionic walking: exploratory case series

Table 2

1885

Power measurements in microvolts squared for subject 2 over 3 sessions at beginning, middle, and end of each session

Session

POz

Fz

Cz

C3

C4

F3

F4

P3

P4

Sum

1st early
1st middle
1st end

.0215
.0116
.0178

.0049
.0299
.0075

.022
.0127
.014

.0145
.0122
.0164

.005
.0117
.0161

.011
.0124
.0121

.0148
.0128
.0139

.0187
.0095
.0177

.0219
.0109
.0178

.1343
.1237
.1333

9th early
9th middle
9th end

.0007
.0015
.0016

.0028
.0069
.0066

.0013
.0034
.0032

.0018
.0035
.0033

.0026
.0033
.003

.0037
.0079
.0075

.009
.0082
.0074

.0012
.0015
.0016

.0008
.0016
.0016

.0239
.0378
.0358

18th

.0036

.0063

.0046

.0066

.0044

.0022

.0021

.0051

.0006

.0355

Abbreviations: C, cental; F, frontal; P, parietal; O, occipital; z, zero.

VO2peak) similar to those of walking in persons with motorincomplete SCI37 or without disability.47,48 These intensities fall
within the light to very light categories as defined by the American
College of Sports Medicine49 and are therefore likely too low to
Table 3 Measures of maximal exercise capacity before and after
6 weeks of OBA, as well as intensity, energy cost, and substrate
utilization during OBA
Variable
HRpeak (beats/min)

9th
18th
Subject Baseline Session Session

1
2
4
VO2peak (mL$kg"1$min"1) 1
2
4
% HRpeak
1
2
4
1
% VO2peak
2
4
TEE (kcal/min)
1
2
4
% CHO/Fat Ox
1
2
4
1
% DHROBA-seated
2
4
1
% DVo2 OBA-seated
2
4
1
% DTEEOBA-seated
2
4
1
% DFatOxOBA-seated
2
4

182
179
139
38.9
14.4
11.4
45
51
63
22
33
45
2.26
1.73
1.53
74/36
56/44
15/85
9
27
51
58
106
46
22
63
14
338
97
242

NA
NA
NA
NA
NA
NA
52
50
52
26
35
43
2.75
1.80
1.38
29/71
46/54
23/77
37
25
37
112
114
97
68
60
37
609
83
240

177
174
167
34.9
15.2
12.2
55
44
54
25
31
41
2.56
1.59
1.39
37/63
54/46
19/81
56
10
58
127
92
224
80
34
103
345
10
303

Abbreviations: CHO, carbohydrate; HR, heart rate; HRpeak, peak heart

rate obtained during graded arm geometry test; NA, not applicable; Ox,
oxidation; VO, volume oxygen; VO2peak, peak oxygen uptake obtained
during graded arm geometry test; % D, percent change.

www.archives-pmr.org

elicit fitness improvements. Consequently, measures of cardiorespiratory fitness did not change consistently with OBA training.
However, OBA was still a sufficient stressor to substantially
elevate heart rate, VO2, TEE, and fat oxidation above seated rest
(see table 3). In addition, the homeostatic model assessment of
insulin resistance improved both pre-post single 1-hour session of
OBA and from baseline to the last week of OBA training.
Follow-up studies will have to determine whether these changes
can translate into reduced cardiometabolic disease risk, perhaps
with more sustained OBA.

Neuropathic pain
Effects on pain related to OBA have not been well described and
appear inconsistent to date. Two studies14,15 have reported on the
effects on pain using visual analog scale ratings but without
defining the specific type of pain. Zeilig et al14 reported no
reduction in low levels of pain from before to after a single session
of OBA but consistent reduction in 1 person with high pain levels
before OBA. Esquenazi et al15 reported reduced pain for 5 of 12
subjects during approximately 22% to 27% of their OBA sessions,
while 1 subject reported increased pain. Our data suggest consistent
reductions in overall pain severity (see supplemental fig S3, found
in supplemental appendix S1) at session 18 compared with the first
session for all subjects. However, we also found consistent reductions in intensity of at-level neuropathic pain and sleep interference in subjects 1 and 2. Furthermore, both self-reported and
evoked allodynia in response to mechanical and thermal stimuli,
commonly thought of as a sign of neuronal hyperexcitability, were
consistently reduced from baseline and before and after OBA.

Fig 5

Walking economy during OBA for each participant.

1886

Study limitations
The data obtained from the cases in this series cannot be generalized to the SCI population and are limited to the clinical environment (under tether and contact guard supervision) and protocol
specifics (ie, training 3d/wk) under which they were obtained. Any
changes or lack thereof may not be the same in an everyday home
or community environment.

Conclusions
Persons with chronic, complete SCI who use a bionic exoskeleton
can achieve walking speeds and distances similar to the averages of
those with motor-incomplete injuries without exaggerated energy
cost. Little or no improvement in either activation of leg muscles or
exercise conditioning could be demonstrated with training for 1
hour, 3 times a week for 6 weeks. Nevertheless, the low energy cost
associated with OBA and the fact that 18 training sessions reduced
overall pain severity, intensity of at-level neuropathic pain, and
sleep interference present an interesting opportunity to explore
effects resulting from prolonged use of the exoskeleton. Whether
extended use would result in more favorable changes in common
consequences of reduced mobility in those with chronic, complete
SCI remains to be determined in follow-up studies.

Suppliers
a. Ekso Bionic, 1414 Harbour Way South, Ste 1201, Richmond,
CA 94804.
b. ABM, Advanced Brain Monitoring Inc, 2237 Faraday Ave, Ste
100, Carlsbad, CA, 92008.
c. Monark Exercise AB, Kroons vag 1 Box 6, 780 50 Vansbro,
Sweden.
d. Polar Electro Inc, 1111 Marcus Ave, Ste M15, Lake Success,
NY 11042.

Keywords
Ambulation; Bionics; Evaluation; Rehabilitation; Spinal cord
injuries

Corresponding author
Jochen Kressler, PhD, Lois Pope Life Center, R-48, 1095 NW
14th Terrace, Miami, FL 33136. E-mail address: JKressler@med.
miami.edu.

Acknowledgments
We thank Maydelis Escalona, BBA, for assisting and conducting
pain assessments; Daniele Cileen, BS, Meagan Mayo, and Isela
Salazar-Martinez, PhD, for assisting with electromyographic data
collection; and Gizelda Casella, PhD, Meagan Mayo, and Isela
Salazar-Martinez, PhD, for analysis of electromyographic data.

References
1. Anderson KD. Consideration of user priorities when developing neural
prosthetics. J Neural Eng 2009;6:055003.

J. Kressler et al
2. Brown-Triolo DL, Roach MJ, Nelson K, Triolo RJ. Consumer perspectives on mobility: implications for neuroprosthesis design. J
Rehabil Res Dev 2002;39:659-70.
3. Widerstrom-Noga EG, Felipe-Cuervo E, Broton JG, Duncan RC,
Yezierski RP. Perceived difficulty in dealing with consequences of
spinal cord injury. Arch Phys Med Rehabil 1999;80:580-6.
4. Ditunno P, Patrick M, Stineman M, Ditunno J. Who wants to walk?
Preferences for recovery after SCI: a longitudinal and cross-sectional
study. Spinal Cord 2008;46:500-6.
5. Farris RJ. Design of a powered lower-limb exoskeleton and control for
gait assistance in paraplegics [dissertation]. Nashville: Vanderbilt
University; 2012.
6. Burchiel KJ, Hsu FPK. Pain and spasticity after spinal cord injury:
mechanisms and treatment. Spine (Phila Pa 1976) 2001;26(24 Suppl):
S146-60.
7. Widerstrom-Noga EG, Duncan R, Felipe-Cuervo E, Turk DC.
Assessment of the impact of pain and impairments associated with
spinal cord injuries. Arch Phys Med Rehabil 2002;83:395-404.
8. Farris RJ, Quintero HA, Goldfarb M. In: Performance evaluation of a
lower limb exoskeleton for stair ascent and descent with paraplegia.
Conf Proc IEEE Eng Med Biol Soc 2012;2012:1908-11.
9. Gancet J, Ilzkovitz M, Cheron G, et al. MINDWALKER: a brain
controlled lower limbs exoskeleton for rehabilitation. Potential applications to space. 11th Symposium on Advanced Space Technologies in Robotics and Automation; European Space Agencys European
Space Research and Technology Centre. Noordwijk, The Netherlands;
April 12-14, 2011.
10. Contreras-Vidal JL, Presacco A, Agashe H, Paek A. Restoration of
whole body movement: toward a noninvasive brain-machine interface
system. IEEE Pulse 2012;3:34-7.
11. Mohammed S, Amirat Y. In: Towards intelligent lower limb wearable
robots: Challenges and perspectives-state of the art. IEEE International Conference on Robotics and Biomimetics (ROBIO 2008); 2008;
Bangkok: IEEE; Feb 22-15, 2009. p 312e317.
12. Dollar AM, Herr H. Lower extremity exoskeletons and active orthoses:
challenges and state-of-the-art. IEEE Trans Robot 2008;24:144-58.
13. Tsukahara A, Hasegawa Y, Sankai Y. In: Gait support for complete spinal cord injury patient by synchronized leg-swing with
HAL. 2011 IEEE/RSJ International Conference on Intelligent
Robots and Systems (IROS); San Francisco: IEEE; Sep 25-30,
2011. p 1737e42.
14. Zeilig G, Weingarden H, Zwecker M, Dudkiewicz I, Bloch A,
Esquenazi A. Safety and tolerance of the ReWalk exoskeleton suit for
ambulation by people with complete spinal cord injury: a pilot study. J
Spinal Cord Med 2012;35:96-101.
15. Esquenazi A, Talaty M, Packel A, Saulino M. The ReWalk powered
exoskeleton to restore ambulatory function to individuals with
thoracic-level motor-complete spinal cord injury. Am J Phys Med
Rehabil 2012;91:911-21.
16. Fineberg DB, Asselin P, Harel NY, et al. Vertical ground reaction forcebased analysis of powered exoskeleton-assisted walking in persons with
motor-complete paraplegia. J Spinal Cord Med 2013;36:313-21.
17. Talaty M, Esquenazi A, Briceno JE. Differentiating ability in users of
the ReWalk! powered exoskeleton: an analysis of walking kinematics. IEEE Int Conf Rehabil Robot 2013;2013:6650469.
18. Jung JY, Park H, Yang HD, Chae M. Brief biomechanical analysis
on the walking of spinal cord injury patients with a lower limb
exoskeleton robot. IEEE Int Conf Rehabil Robot 2013;2013:
6650351.
19. Bohannon RW, Smith MB. Interrater reliability of a Modified Ashworth Scale of muscle spasticity. Phys Ther 1987;67:206-7.
20. Strausser KA, Kazerooni H. In: The development and testing of a
human machine interface for a mobile medical exoskeleton. 2011
IEEE/RSJ International Conference on Intelligent Robots and Systems
(IROS); San Francisco: IEEE; Sep 25-30, 2011. p 4911e6.
21. Strausser KA, Swift TA, Zoss AB, Kazerooni H, Bennett BC. In: Mobile
exoskeleton for spinal cord injury: development and testing. Arlington:
ASME; Oct 31-Nov 2, 2011. p 419e25.

www.archives-pmr.org

Bionic walking: exploratory case series

22. Field-Fote EC, Roach KE. Influence of a locomotor training approach
on walking speed and distance in people with chronic spinal cord
injury: a randomized clinical trial. Phys Ther 2011;91:48-60.
23. Benz EN, Hornby TG, Bode RK, Scheidt RA, Schmit BD. A physiologically based clinical measure for spastic reflexes in spinal cord
injury. Arch Phys Med Rehabil 2005;86:52-9.
24. Nakazawa K, Kawashima N, Akai M. Enhanced stretch reflex excitability of the soleus muscle in persons with incomplete rather than
complete chronic spinal cord injury. Arch Phys Med Rehabil 2006;87:
71-5.
25. Pierrot-Deseilligny E, Burke D. The circuitry of the human spinal
cord: its role in motor control and movement disorders. New York:
Cambridge University Pr; 2005.
26. Pfurtscheller G, Brunner C, Schlogl A, Lopes SFH. Mu rhythm (de)
synchronization and EEG single-trial classification of different motor
imagery tasks. Neuroimage 2006;31:153.
27. Pfurtscheller G, Leeb R, Keinrath C, et al. Walking from thought.
Brain Res 2006;1071:145-52.
28. Barlow JS. The electroencephalogram: its patterns and origins.
Cambridge: MIT Pr; 1993.
29. Wagner J, Solis-Escalante T, Grieshofer P, Neuper C, Muller-Putz G,
Scherer R. Level of participation in robotic-assisted treadmill walking
modulates midline sensorimotor EEG rhythms in able-bodied subjects.
Neuroimage 2012;63:1203-11.
30. Jacobs KA, Burns P, Kressler J, Nash MS. Heavy reliance on carbohydrate across a wide range of exercise intensities during voluntary
arm ergometry in persons with paraplegia. J Spinal Cord Med 2013;
36:427-35.
31. Kressler J, Jacobs K, Burns P, Betancourt L, Nash M. Effects of circuit
resistance training and timely protein supplementation on exerciseinduced fat oxidation in tetraplegic adults. Top Spinal Cord Inj
Rehabil 2014;20:113-22.
32. Widerstrom-Noga E, Biering-Sorensen F, Bryce T, et al. The International Spinal Cord Injury Pain Basic Data Set. Spinal Cord 2008;46:
818-23.
33. Bouhassira D, Attal N, Fermanian J, et al. Development and validation
of the neuropathic pain symptom inventory. Pain 2004;108:248-57.
34. Finnerup NB, Johannesen IL, Fuglsang-Frederiksen A, Bach FW,
Jensen TS. Sensory function in spinal cord injury patients with and
without central pain. Brain 2003;126(Pt 1):57-70.
35. Musselman KE. Clinical significance testing in rehabilitation research:
what, why, and how? Phys Ther Rev 2007;12:287-96.

www.archives-pmr.org

1887
36. American Electroencephalographic Society guidelines for standard
electrode position nomenclature. J Clin Neurophysiol 1991;8:
200-2.
37. Tudor-Locke C, Bassett JDR. How many steps/day are enough? Preliminary pedometer indices for public health. Sports Med 2004;34:1-8.
38. Kressler J, Nash, Burns PA, Field-Fote EC. Metabolic responses to
four different body weight supported locomotor training approaches in
persons with incomplete spinal cord injury. Arch Phys Med Rehabil
2013;94:1436-42.
39. Putzke JD, Richards JS, Hicken BL, DeVivo MJ. Predictors of life
satisfaction: a spinal cord injury cohort study. Arch Phys Med Rehabil
2002;83:555-61.
40. Noreau L, Shephard RJ. Spinal cord injury, exercise and quality of life.
Sports Med 1995;20:226-50.
41. Swinnen E, Duerinck S, Baeyens JP, Meeusen R, Kerckhofs E.
Effectiveness of robot-assisted gait training in persons with spinal cord
injury: a systematic review. J Rehabil Med 2010;42:520-6.
42. Thomas CK, Peterson L, Ferrell S, Cohan C, Klein CS. Daily use
of leg muscles in healthy subjects. Med Sci Sports Exerc 2010;
42:25.
43. Stephens MJ, Yang JF. Loading during the stance phase of walking in
humans increases the extensor EMG amplitude but does not change
the duration of the step cycle. Exp Brain Res 1999;124:363-70.
44. Thomas CK, Zijdewind I. Fatigue of muscles weakened by death of
motoneurons. Muscle Nerve 2006;33:21-41.
45. Rayegani SM, Shojaee H, Sedighipour L, Soroush MR, Baghbani M,
Amirani OB. The effect of electrical passive cycling on spasticity in
war veterans with spinal cord injury. Front Neurol 2011;2:39.
46. Kiser TS, Reese NB, Maresh T, et al. Use of a motorized bicycle
exercise trainer to normalize frequency-dependent habituation of
the H-reflex in spinal cord injury. J Spinal Cord Med 2005;28:
241-5.
47. Browning RC, Kram R. Energetic cost and preferred speed of walking
in obese vs. normal weight women. Obes Res 2005;13:891-9.
48. Schrack JA, Simonsick EM, Ferrucci L. The relationship of the energetic cost of slow walking and peak energy expenditure to gait speed
in mid-to-late life. Am J Phys Med Rehabil 2013;92:28-35.
49. Garber CE, Blissmer B, Deschenes MR, et al. American College of
Sports Medicine position stand. Quantity and quality of exercise for
developing and maintaining cardiorespiratory, musculoskeletal, and
neuromotor fitness in apparently healthy adults: guidance for prescribing exercise. Med Sci Sports Exerc 2011;43:1334-59.

1887.e1

J. Kressler et al

Supplemental Appendix S1

Supplemental Table S2
disease risk

Methods
Characteristics of the subjects are shown in supplemental table S1.

Variable

Inclusion criteria

Total cholesterol
(mg/dL)

Subjects were evaluated by a physical therapist to assess joint

range of motion, strength, transfer status, sitting balance, and leg
lengths. Subjects were required to have minimal limitations that
could reasonably be accommodated by adjusting the device settings. Subjects were required to have sufficient upper extremity
strength to manage a walker and to transfer to and from their
wheelchair with no greater than minimal assistance (ie, 25% effort
from another person), and sufficient balance to sit unsupported.
Subjects also had to have been involved in a standing program on
a regular basis (ie, at least once per week as per participant selfreport) in order to ensure sufficient orthostatic tolerance for OBA.

Brief description of bionic exoskeleton

In brief, the device is a lower extremity robotic exoskeleton with 2
legs connected to a torso structure (containing the computer and
batteries). The torso is aligned to the users lower back, while the
legs of the exoskeleton are fastened to the users legs by hook-andloop fastener straps that align the users lower back and joints with
those of the robot (supplemental fig S1). In addition, 2 straps are
tightened over the users shoulders similar to wearing a backpack.
The device has 4 motors that actuate the hip and knee joints in the
sagittal plane. Movements of these joints in other directions are
restricted. The ankle joints of the exoskeleton allow passive
(spring) movement in the sagittal plane only.
The locomotor settings were set at factory defaults (step
length, 12in; step height, 1in; swing time, 1.3s). For subject 1
(only), step length was increased to 15in and swing time reduced
to 1.0 seconds to accommodate his wish to walk faster. For all
sessions, participants were tethered to an overhead track and lift
system (without weight support), and contact guarding was provided by trained investigators.

Electromyographic activity assessment

To record electromyographic activity, pairs of electrodes (Superior
Silver Electrodes,a trimmed to 1&2.5cm) were placed on the skin
over each test muscle belly, 4cm apart. To time-lock the electromyographic activity to the step cycle, the knee angle was recorded
simultaneously by strapping a goniometer (SG150b) to the leg of
the exoskeleton on the test side. All data were amplified (w300&;
Z03b), filtered (15e500Hz), and sampled continuously at 1000Hz
to a custom-made, battery-powered data logger.1 Offline, custom
software written in Matlabc and DADiSPd was used to calibrate the

Supplemental Table S1

HDL (mg/dL)

LDL (mg/dL)

VLDL (mg/dL)

Total cholesterolto-HDL ratio

TG (mg/dL)

Lipid markers of cardiometabolic

Baseline Week

Final Week

Subject
No.

Pre

Post

Pre

Post

1
2
4
1
2
4
1
2
4
1
2
4
1
2
4
1
2
4

155
166
176
63
32
59
80
120
98
12
14
19
2.5
5.19
2.98
58
72
97

135
155
169
52
30
57
74
113
93
9
12
19
2.6
5.17
2.96
44
58
96

157
169
155
66
32
52
82
122
82
9
15
21
2.4
5.28
2.98
47
77
96

137
158
149
53
31
49
75
113
81
9
14
19
2.6
5.1
3.04
47
71
95

Abbreviations: HDL, high-density lipoprotein; LDL, low-density lipoprotein; TG, triglycerides; VLDL, very low density lipoprotein.

data using a 1-mV sine wave, rectify then integrate the electromyographic activity every 10 milliseconds.1 Electromyographic
activity was separated from baseline noise by calculating a
threshold (mean 3 SDs of the highest 10% of integrals) from five
30-second records of quiet time (no electromyographic activity1).
The duration and mean intensity of electromyographic activity
were calculated every 10 minutes during walking. All electromyographic activity data were normalized to the respective mean
maximal muscle compound action potential (M-wave), which was
evoked by delivering 3 single supramaximal stimuli at 1Hz to the
appropriate peripheral nerve (data from biceps femoris were
normalized to vastus lateralis values since the sciatic nerve is too
deep to stimulate reliably from the skin surface). This normalization is reasonable given that atrophy in hamstrings and quadriceps
muscles is similar after SCI.2 The maximal soleus H-reflex (H) was
evoked by delivering single pulses of increasing intensity (subthreshold to maximal) to the tibial nerve. Subjects were then asked
to perform three 5-second MVCs of each muscle while seated in
their wheelchair, with 1 minute of rest between contractions. The
presence or absence of electromyographic activity confirmed
whether the muscle was controlled voluntarily or was paralyzed,
respectively. Electromyographic activity recorded from arm muscles during walking was also normalized to MVC values.

Subject characteristics

Subject No.

Age (y)

Sex

LOI

AIS

Height (m)

Weight (kg)

VO2peak
(mL$kg"1$min"1)

HRpeak
(beats/min)

1
2
4

26
27
38

M
M
F

T9/10
T7
T1/2

A
A
A

1.80
1.88
1.63

68
87
70

39
15
12

182
179
167

Abbreviations: AIS, American Spinal Injury Association Impairment Scale; F, female; HRpeak, peak heart rate; LOI, level of injury; M, male.

www.archives-pmr.org

Bionic walking: exploratory case series

Supplemental Table S3
disease risk

Variable
Glucose (mg/dL)

Insulin (mIU/mL)

Hb A1C (%)

HOMA-IR

1887.e2

Glycemic markers of cardiometabolic

Baseline Week

Final Week

Subject
No.

Pre

Post

Pre

Post

1
2
4
1
2
4
1
2
4
1
2
4

84
89
79
11
15
11
4.6
5.6
5.2
2.34
3.22
2.18

84
73
70
7
1
12
4.6
N/A
5.2
1.47
0.19
2.10

81
90
69
6
6
7
4.6
5.6
5.1
1.21
1.28
1.14

84
47
78
6
9
8
4.7
N/A
5.1
1.17
1.07
1.57

Abbreviations: Hb A1C, glycosylated hemoglobin; HOMA-IR, homeostatic model assessment of insulin resistance; N/A, not available.

Supplemental Fig S2 Electromyogram from leg and arm muscles

and knee angle during 4 typical steps (subject 2, session 1). Dashed
and dotted vertical lines show the swing phase of the test leg.

EEG quality assessment and artifact correction

All datasets were checked for signal quality using impedance
measurements before and after each testing session. Data recorded
with impedances above 90kU were not used in the study. All
analysis of neural intent and motor function was performed offline
with Matlab software. Because of the noisy nature of ambulatory
EEG, additional offline artifact rejection was used to ensure signal
quality. Large-amplitude artifacts (>100mV) were removed from
the recordings. After time domain rejection of noise recordings,
independent component analysis (ICA) was applied using EEG
lab Infomax ICA formula.3 ICA allowed the removal of eye blink
artifacts from the data as described previously.4

Cardiometabolic risk marker assessment

Supplemental Fig S1

www.archives-pmr.org

Subject during OBA assessment session.

Participants abstained from caffeine/alcohol for 24 hours before

testing, which was conducted after an overnight (10h) fast and
>24 hours after the last OBA bout. Lipid profiles and glucose

1887.e3

J. Kressler et al

Supplemental Fig S3 (A) Average pain severity subscale scores (0e6) of the Multidimensional Pain InventoryeSCI version. (B) Average pain
intensity scores (NRS) averaged from morning and afternoon values. (C) Sleep interference item (0e6) of the International SCI Basic Pain Dataset.
*Minimal clinically important differences for pre-post changes (!30%).

levels were determined as described by Nash and Mendez.5 Insulin was measured by a solid-phase radioimmunoassay (Coat-ACounte). The homeostatic model assessment of insulin resistance
was calculated by the original model as glucose (mg/dL) & insulin
(mIU/L)/405.6

Pain scales
While the International SCI Basic Pain Dataset assesses the intensity of a specific pain type on an NRS ranging from 0 to 10 (ie,
0 denoting no pain; 10 denoting pain as intense as you could
imagine), the pain severity subscale of the Multidimensional Pain
InventoryeSCI version assesses overall pain severity. The latter
scale consists of 3 items rated on an NRS from 0 to 6. One item
concerns current pain intensity; one, severity over the past week;
and one, suffering caused by pain. The sum score is presented as
an average of these 3 items.

Results
OBA performance
Subject 1 was able to walk in the device continuously for each
entire session from session 4 onward (see fig 1A). From sessions 8
and 14 onward, swing time and step length were adjusted to 1
second and 15in, respectively, because of his high proficiency.
Subjects 2 and 4 were not able to walk in the device throughout
entire sessions without breaks, usually because of discomfort/fatigue in the arm musculature (see Neuromuscular section below),
and they remained at the default device settings throughout the
study. Progression through walking modes also varied across subjects. Subject 1 progressed from FirstStep to ActiveStep at session
4, and to ProStep at session 6. Subject 2 progressed to ActiveStep at
session 6, tried ProStep at session 10 but preferred to walk in

www.archives-pmr.org

Bionic walking: exploratory case series

ActiveStep throughout the remainder of the study. Subject 4 progressed to ActiveStep at session 4 and to ProStep at session 9.

Cardiometabolic risk markers

Lipid markers showed no consistent tendencies for changes from
baseline to the last week of OBA intervention (median, "1%;
minimum-maximum, "22% to 25%; supplemental table S2).
Glucose did not show consistent changes across subjects (median,
2%; minimum-maximum, "11% to 36%), while the homeostatic
model assessment of insulin resistance declined consistently by 1
to 2 points from baseline to the last week (median, 48%;
minimum-maximum, 25%e60%; supplemental table S3).

Evoked pain and response to thermal stimulation

Subjects 1 and 4 reported minimal levels of evoked pain (average
rating of 1 on the Neuropathic Pain Symptom Inventory) only on 1
occasion, and thermal and mechanical stimulation evoked no pain
throughout the study. Subject 2 reported evoked pain in all sessions
(2.33, 1st session; 3.67, 9th session; 1.00, 18th session). However,
the average rating was much lower at the last session compared with
the first. Consistent with this participants self-reported evoked pain
(allodynia), a response to thermal and mechanical stimuli was
evoked at all sessions. Interestingly, this subject reported allodynia
in the moderate range (NRSZ5) before training at sessions 9 and
18, but no evoked pain posttraining on the same day, suggesting
some beneficial effects on evoked pain.

www.archives-pmr.org

1887.e4

Suppliers
a. Uni-Patch, 1313 W Grant Blvd, Wabasha, MN 55981.
b. Motion Lab Systems, 15045 Old Hammond Hwy, Baton
Rouge, LA 70816.
c. The Mathworks, 3 Apple Hill Dr, Natick, MA 01760.
d. DSP Development Corp, 1 Bridge St, Newton, MA 02458.
e. Siemens AG, Wittelsbacherplatz 2, 80333 Munich, Germany.

References
1. Klein CS, Peterson LB, Ferrell S, Thomas CK. Sensitivity of 24-h
EMG duration and intensity in the human vastus lateralis muscle to
threshold changes. J Appl Physiol 2010;108:655-61.
2. Shah PK, Stevens JE, Gregory CM, et al. Lower-extremity muscle
cross-sectional area after incomplete spinal cord injury. Arch Phys
Med Rehabil 2006;87:772-8.
3. Makeig S, Bell AJ, Jung TP, Sejnowski TJ. Independent component
analysis of electroencephalographic data. Adv Neural Inf Process Syst
1996;8:145-51.
4. Delorme A, Sejnowski T, Makeig S. Enhanced detection of artifacts in
EEG data using higher-order statistics and independent component
analysis. Neuroimage 2007;34:1443.
5. Nash MS, Mendez AJ. A guideline-driven assessment of need for
cardiovascular disease risk intervention in persons with chronic
paraplegia. Arch Phys Med Rehabil 2007;88:751-7.
6. Wallace TM, Levy JC, Matthews DR. Use and abuse of HOMA
modeling. Diabetes Care 2004;27:1487-95.