PM R 10 (2018) S174-S188

www.pmrjournal.org

Innovations Influencing Physical Medicine and Rehabilitation

Exoskeleton and End-Effector Robots for Upper and

Lower Limbs Rehabilitation: Narrative Review
Franco Molteni, MD, Giulio Gasperini, MD, Giovanni Cannaviello, MD,
Eleonora Guanziroli, Eng, PhD

Abstract

Recovery of upper and lower limbs function is essential to reach independence in daily activities in patients with upper motor
neuron syndrome (UMNS). Rehabilitation can provide a guide for motor recovery influencing the neurobiology of neuronal plasticity
providing controlled, repetitive, and variable patterns. Increasing therapy dosage, intensity, number of repetition, execution of
task-oriented exercises, and combining top-down and bottom-up approaches can promote plasticity and functional recovery. Robotic
exoskeletons for upper and lower limbs, based on the principle of motor learning, have been introduced in neurorehabilitation. In this
narrative review, we provide an overview of literature published on exoskeleton devices for upper and lower limb rehabilitation
in patients with UMNS; we summarized the available current research evidence and outlined the new challenges that neuro-
rehabilitation and bioengineering will have to face in the upcoming years. Robotic treatment should be considered a rehabilitation
tool useful to generate a more complex, controlled multisensory stimulation of the patient and useful to modify the plasticity of
neural connections through the experience of movement. Efficacy and efficiency of robotic treatment should be defined starting
from intensity, complexity, and specificity of the robotic exercise, that are related to human-robot interaction in terms of
motion, emotion, motivation, meaning of the task, feedback from the exoskeleton, and fine motion assistance. Duration of a single
session, global period of the treatment, and the timing for beginning of robotic treatment are still open questions. There is the need to
evaluate and individualize the treatment according to patient’s characteristics. Robotic devices for upper and lower limbs open
a window to define therapeutic modalities as possible beneficial drug, able to boost biological, neurobiological, and epigenetic
changes in central nervous system. We need to implement large and innovative research programs to answer these issues in the
near future.

Introduction If walking is a cyclical movement that involves

cortical and subcortical areas, upper limb movements
Upper motor neuron syndrome is associated with an are multijoint and complex without any cyclical/ste-
impairment of upper and lower limb functions; facili- reotypical pattern. The shoulder is deputized to the
tation of walking and of upper extremity function is orientation of the upper limb into the space; the elbow
essential to reach independence in daily activities. and the wrist are important for turning away/
Walking is “the repeated performance by the lower approaching objects from/to the subject’s body, and
limbs of a sequence of motions that simultaneously the hand is necessary to grasp and manipulate objects.
advances the body along the desired line of progression Hundreds of different strategies can be used to perform
while also maintaining a stable weight-bearing posture the same goal-directed movement because the upper
[1]”. The elements characterizing “normal” walking are limb is a kinematically redundant multijoint system with
progression, standing stability, and energy conserva- multiple degrees of freedom [2]. The choice of the di-
tion. Pathologic conditions alter the mode and the rection of the movement to be performed should be
efficiency of walking; if forward progression and stance taken into the highest consideration, approaching or
stability are preserved, the loss of control capacity moving away from the body, toward an object, or to-
necessitates substitution. ward a definite part of patient’s own body. Targeting

1934-1482/$ - see front matter ª 2018 by the American Academy of Physical Medicine and Rehabilitation
https://doi.org/10.1016/j.pmrj.2018.06.005
 F. Molteni et al. / PM R 10 (2018) S174-S188 S175

the patient movements toward a precise part of the

body could improve body perception and the recovery of
the body schema [3].
In this complexity of movements and motor control,
rehabilitation can provide a guide of motor recovery
influencing the neurobiology of neuronal plasticity
providing controlled, repetitive, and variable patterns.
Plasticity is the capacity to modify the organization of
neural components during human life [4]; adaptive modi-
fications are a neural plastic phenomenon in response to
anatomical or functional changes in the brain. Plastic
phenomena can occur after brain damage [5,6]. with
reorganization of the connections that is modulated/
influenced by experiences [7,8]. Plastic rewiring involve
the perilesional tissue and injured hemisphere, the
contralateral brain, subcortical, and spinal regions [9].
Increasing therapy dosage, intensity, number of repeti-
tion, execution of task-oriented exercises, and combining
top-down and bottom-up approaches, can promote plas-
ticity and functional recovery [10-12].
According to the principles of neuroplasticity and
motor learning, robotic devices have been introduced in
the rehabilitation process because they can maximize
afferent input from peripheral joints and provide task-
specific stimulation to the central nervous system to
promote functional recovery. Figure 2. End-effector for lower limb rehabilitation.
Robotic devices for upper and lower limb rehabilita-
tion can be divided into end-effector (Figures 1 and 2)
and exoskeleton devices (Figures 3-5). The end-effector
systems, in which movements are generated from the
most distal segment of the extremity and no alignment
between patient-robot joints is required, use footplates
or handles to generate a motion of the limb in space
[13-15]. In the exoskeleton system, there is a one-to-
one correspondence between robots and human joints,
and each single joint is guided along a preprogramed
trajectory. The exoskeletons can be classified according
to the number of joints controlled and can be divided in
unilateral and bilateral robots.
Electromechanical devices can be also divided in
active, passive, active-assisted, resistive and interac-
tive devices. Active devices provide a patients’ active
assistance to the movement; passive devices ensure a
nonpowered support to patients’ upper and lower
limbs; active-assisted devices complete the movement
after the user initiates a movement; resistive devices
offer an opposition to the subject’s movement execu-
tion; and interactive devices allow the correction of
movements with a combination of actuators and con-
trol strategies [16].
Powered robotic devices for the lower limb can be
divided into “static robots” and overground robots [17].
Static robots allow patients to perform gait training in a
fixed and confined area; examples of these devices are
treadmill-based body weight support (BWS), in which
patients only receive body weight assistance without
Figure 1. End-effector for upper limb rehabilitation. lower limb support; treadmill-based exoskeletons,
S176 Exoskeleton and End-Effector Robots

Figure 3. Exoskeleton for upper limb rehabilitation.

where there is the combination of BWS and lower limb

assistance by powered exoskeleton during gait
(Figure 4); and BWS end-effector (BWS foot-plate Figure 5. Overground exoskeleton for lower limb rehabilitation.
trainers), where there is a combination of BWS and

only the patients’ feet are actuated and controlled.

Overground robots allow the patients to walk over-
ground and to explore the environment; patients are not
confined to a fixed area. Overground exoskeleton
(Figure 5) can be divided also in assistive and rehabili-
tation devices; the aim of assistive devices is to ensure
mobility in a community environment whereas rehabil-
itation devices can be used to address the recovery of
gait function in patients with neurologic injuries. In
these devices, at least 2 joints are actuated (hip and
knee) and usually 1 joint (ankle) is controlled with a
mechanical joint. Overground exoskeletons are bilateral
robots and usually their involvement requires upper
limb aids to maintain balance.
Robotic exoskeletons for gait rehabilitation guide the
legs through preprogramed physiological gait patterns
while the subject experiences near-normal propriocep-
tive input during limb loading. The preprogramed
walking pattern is quite similar to a normal gait, and it
includes gait cycle timing, interlimb and interjoint co-
Figure 4. Body weight support exoskeleton for lower limb ordination, appropriate limb loading, and afferent
rehabilitation. signals.
 F. Molteni et al. / PM R 10 (2018) S174-S188 S177

Robotic exoskeletons are built in accordance with our research included only patients with spinal cord
modern principles of overground gait rehabilitation. In injury (SCI) or stroke. Pediatric patients were excluded
particular, these devices should maximize loading of the from this review.
lower limbs instead of the upper extremities, promote
hip extension and limb unloading synchronization with Results
simultaneous loading of the contralateral lower limb to
promote swing initiation, and promote step initiation Robotics for Lower Limbs Following Stroke
from a stride position to allow weight transfer from an
extended and loaded limb forward to the unloaded The results of the clinical trials in which lower limb
limb. robotic devices were used as a rehabilitative tool in
The aim of this article is to provide an overview of stroke patients are reported in Table 1.
recent reviews published about exoskeleton devices for A Cochrane review by Mehrholz et al [18] showed that
the rehabilitation of upper and lower limbs in patients robotic gait training combined with physiotherapy might
with upper motor neuron syndrome, to summarize the improve recovery of independent walking in poststroke
available current research evidence, and to underline patients. There was no difference in velocity and
limitations and future directions for exoskeleton use in endurance in stroke patients when treated with robots
clinical practice. compared with conventional treatments at the end of
the training period and at follow-up. People in the first 3
Literature Search Strategy months after stroke and those who were not able to
walk seemed to benefit most from robotic intervention.
We focused on Medline, PubMed, PubMed Central, Walking velocity at the end of the training period was
PEDro, and Scopus databases for our research, and we higher when end-effector devices were used compared
included reviews with higher levels of evidence (I or II). with exoskeleton ones. This review also highlighted that
Various criteria were selected to try to include all there was no unique definition of frequency, duration,
relevant reviews. We used the key words robots, reha- and timing of the robotic gait training.
bilitation, upper limb, lower limb, review; articles Bruni et al [19] in 2018 found that end-effector robots
since January 2010 were included, because previously were significantly more effective in improving walking
exoskeleton reviews included only small randomized speed when compared to a control group and they found
controlled trials and the clinical use of exoskeletons in no evidence that exoskeleton robots were more effec-
rehabilitation was limited. Only English-language arti- tive than conventional therapy. Subacute stroke pa-
cles were included in our work. tients treated with end-effector devices seem to best
In PubMed Central, the search phrases for lower limb benefit in terms of walking velocity whereas no statis-
were as follows: “robotics” OR “robotic” AND “gait” OR tical difference between groups was found in chronic
“gait” AND “walking” OR “walking” AND “rehabilitation” stroke patients.
OR “rehabilitation” OR “rehabilitation” AND “review” Lo et al [20] performed a subgroup analysis of pa-
OR “review literature as topic” OR “review.” The search tients with mild and severe impairments; they showed
phrases for upper limb were as follows: “robotics” OR that patients with severe impairments treated with
“robotics” AND “upper extremity” OR “upper” AND robotic devices reached better improvements than
“extremity” OR “upper extremity” OR “upper” AND patients treated with conventional therapy [21].
“limb” OR “upper limb” AND “review” OR “review Reviews about robotic rehabilitation for gait recovery
literature as topic” OR “review.” The inclusion of in stroke patients mainly involved static exoskeletons
“exoskeleton” and “review” terminologies was to focus [22]; reviews about wearable powered exoskeletons are
the search on clinical evidences on the use of robotic few, with small populations, and are mainly technical,
devices in clinical practice. The papers found were scoping, or narrative [23,24].
supplemented by the authors’ personal experience from
previous and ongoing work with exoskeletons. Robotics for Lower Limbs Following SCI
We identified 228 reviews, 96 about lower limb
robotic devices and exoskeletons and 132 about upper Clinical reviews on robotic devices for lower limbs in
limb robotic devices and exoskeletons. For an initial SCI patients (Table 2) can be divided into 2 subgroups:
selection, it was decided to read the titles and the ab- robotics used as rehabilitative device and robotics used
stracts (if available) of the identified publications. We as assistive device, in which only exoskeleton and no
excluded all articles with unavailable full text. Then, end-effector are included. The term assistive refers to
we examined the full text for the remaining studies and the possibility to use exoskeleton as an orthosis. Our
included clinical reviews excluding technical, scoping, search found 2 reviews about the use of powered
and narrative ones. At the end of the selection process, overground exoskeleton as assistive device.
15 clinical reviews for the lower limb and 8 reviews for In 2015, Louie et al [25] analyzed the use of powered
the upper limb were included. Clinical reviews found in overground exoskeleton as an assistive tool to restore
 S178
Table 1
Main Results of Clinical Review About the Use of Robotic Devices for Lower Limb on Stroke Patients
Year of Number of Type of Control Outcome
Authors Publication Publications Pathology Studies Groups Type of Robots Aim Measures Results Limitation
Lo et al [20] 2017 21 Stroke (acute RCT Yes
End-effector Effectiveness FIM (Walking), No differences between control Heterogeneity
and chronic)
Exoskeleton both Fugl-Meyer and experimental group. of studies.
overground both (Lower Limb), MI Subgroup with severe No distinction
treadmill (Lower Limb), impairments treated with between
EFAP and mEFAP robotic devices showed robots.
better improvements than
conventional group.
Bruni et al [19] 2018 13 Stroke (acute RCT Yes
Exoskeleton Efficacy of 10mWT End-effector more effective Heterogeneity
and chronic)
End-effector robotic 6minWT than conventional treatment. of studies.

Body weight devices / TUG No differences between control No evaluation
support walking 5mWT group and exoskeletons. of virtual
speed (m/s) FAC Acute patients: better reality
improvement with end- effects.
effector or exoskeleton.
Chronic patients: no significant

Exoskeleton and End-Effector Robots

difference between control
and robot group.
Mehrholz et al 2017 36 Stroke (acute RCT Yes
End-effector Efficacy Independence in People in the acute and Heterogeneity
[18] and chronic)
Body weight walking subacute phases after stroke in outcomes.
support 10mWT profit more than people Different types
exoskeleton 6minWT treated after 3 months of robots.

Exoskeleton poststroke.
Ambulatory people do not
benefit from
electromechanical and robot-
assisted gait training.
Walking velocities at the end of
the intervention phase were
higher when end-effector
devices were used compared
with participants who
received training by an
exoskeleton device.
Swinnen et al 2014 9 Stroke patients RCT Yes
Body weight Effect on Tinetti-Berg No significant difference in Limited
support balance Balance Scale balance between number of
exoskeleton Postural sway test experimental and control studies,
TUG group. small
population.
Ada et al 2010 6 Nonambulatory RCT Yes
Body weight Efficacy Independence in Patients treated with body No follow-up.
acute stroke support walking weight support exoskeleton Heterogeneity
exoskeleton 10mWT achieved more independent in outcomes.
6minWT walking than patients in
control group.
RCT ¼ Randomized controlled trials; FIM ¼ Functional Independence Measure; MI ¼ Motricity Index; EFAP ¼ Emory Functional Ambulation Profile; mEFAP ¼ modified Emory
Functional Ambulation Profile; 10mWT ¼ 10-meter walking test; 6minWT ¼ 6-minute walking test; TUG ¼ Timed Up and Go; 5mWT ¼ 5 meters walking test; FAC ¼ Functional Ambulation
Classification.
 F. Molteni et al. / PM R 10 (2018) S174-S188 S179

walking functions in complete SCI patients; they eval- of mobility and independence for robot-assisted gait
uated all studies in which walking outcomes using training if compared to conventional overground gait
powered exoskeleton were collected. They claimed that training. The data for participants included in the study
powered wearable exoskeletons allow individuals with more than 1 year postinjury showed improvements in
complete thoracic SCI to walk at modest speed, that is, gait speed and balance with robot-assisted gait training
influenced by lesion levels and training duration. In fact, when compared with no intervention. Authors underline
gait mean velocity using an exoskeleton was 0.26 m/s, also that in the literature there are not randomized
which is not considered suitable for community ambu- controlled trials on end-effector and powered wearable
lation. Forrest et al [26] identified 0.44 m/s as the overground exoskeleton.
threshold velocity for limited community ambulation
after incomplete SCI. These results are confirmed by Robotics for Upper Limbs Following Stroke
Miller et al [27], who published the first meta-analysis to
evaluate the clinical effectiveness and safety of pow- The main results of the clinical reviews in which
ered wearable overground gait training in complete SCI upper limbs robotic devices were used in clinical prac-
patients. They reported that powered exoskeletons tice are reported in Table 3.
allow patients with SCI to safely ambulate in real-world Peter et al [35] in 2011 evaluated the efficacy of ro-
settings; the achieved gait speed suggested the possi- botic upper limb rehabilitation in hemiplegic patients
bility to use exoskeletons for independent ambulation in after stroke concluding that robotic treatment was
home and community environments with a reduced more effective than traditional physiotherapy in terms
level of physical activity if compared to those obtained of motor control and functional outcome assessment.
during the use of knee-ankle-foot orthosis or reciprocal They used the Fugl-Meyer scale for measuring motor
gait orthosis. control of the upper limb, Modified Ashworth Scale for
Other relevant reviews on SCI patients focused on the spasticity, and Functional Independence Measure for
possibility to use powered robotic exoskeletons as a the disability. Studies with different treatment dura-
rehabilitative tool in patients with incomplete SCI. tions (from 3 to 12 weeks) were included; moreover, the
Initial reviews involving SCI patients [28,29], duration of a single session was not specified, and some
compared BWS treadmill or robotic BWS treadmill with trials did not report the subacute or chronic stage of the
conventional therapy and concluded that BWS treadmill stroke. The authors could not draw any general con-
is equivalent to overground gait training in restoring gait clusions because of these limitations despite the fact
function in subacute incomplete SCI patients. that several hundreds of patients were treated with
More recently, Aguirre et al [30] showed no differ- robotic devices.
ences between control group and exoskeletons in terms Mehrholz et al [36] in 2015 assessed the effectiveness
of velocity but a large effect on the Walking Index for of electromechanical and robot-assisted arm training
Spinal Cord Injury and the Functional Independence for improving activities of daily living, arm function and
Measure was observed favoring the exoskeleton group in arm muscle strength in patients after stroke. They
incomplete SCI patients. Mehrholz et al [31] observed no concluded that people who undergo robotic training for
difference in gait speed but clinically important effects the upper limb may improve arm and hand strength and
on distance in patients treated with BSW treadmill and function in activities of daily living but the quality of the
robotic devices (both end-effector and exoskeleton) evidence was from low to very low because of the lack
compared with overground training. of blinding and the randomization of studies included in
Fisahn et al [32] in 2016 in their review did not find the review. Furthermore, the trials included in this re-
changes in 10-meter walking test and Spinal Cord Inde- view differed in treatment intensity, duration, and in
pendence Measure score between baseline and the end patients’ characteristics.
of treatment on chronic patients, whereas 6-minute Sheng et al [37] in 2015 first reviewed the literature
walking test, Walking Index for Spinal Cord Injury, and about bilateral robotic treatment compared to unilat-
Functional Independence Measure showed different re- eral. They focused on the robotic treatment as a useful
sults, with some results in favor of exoskeletons and tool to provide long-term intensive and accurate reha-
some others in favor of traditional treatment. bilitation and to avoid the physical burden of therapists,
Another review published by Cheung and colleagues with improved efficiency as 1 therapist could treat
[33] in 2017 found significant improvement in walking multiple patients more frequently. However, despite
speed and endurance in patients treated with robotic promising results in some of the trials considered, the
devices when compared to traditional treatment, but reviewers concluded that the clinical data in the liter-
controversial results were found when different robotic ature was insufficient to justify the effectiveness of the
devices were compared. bilateral training when compared with other therapy.
Nam et al [34] in 2017 reported data for participants Experimental groups were small, and trials varied in
less than 6 months postinjury showing improvements in treatment duration, dose-matched training protocols,
walking distance, lower limb strength, functional level and effective measure methods.
Table 2

S180
Main Results of Clinical Review About the Use of Robotic Devices for Lower Limbs on Spinal Cord Injury Patients
Year of Number of Type of Control
Authors Publication Publications Pathology Studies Groups Type of Robots Aim Outcome Measures Results Limitation
Aguirre et al 2017 6 Incomplete SCI RCT Yes
Body weight Effectiveness 10mWT No differences in walking Different
[30] at different support 6minWT speed between control frequency and
levels exoskeleton WISCI and exoskeleton group duration of
MRC but large effect on robotic
SCIM WISCI and FIM treatment
FIM Different setting of
the device
Heterogeneity of
trial design,
intervention and
participants
Cheung et al 2017 11 SCI acute and RCT Yes
Body weight Effectiveness Walking speed Significant improvement Different study
[33] chronic support Walking endurance in walking speed and design
exoskeleton Walking index for SCI endurance in Different lesion
Lower limb strength experimental group if level

Exoskeleton and End-Effector Robots

compared to control
group
Mehrholz et al 2017 13 SCI acute and RCT Yes
Body weight Effectiveness Walking speed Walking Body weight support Different
[31] chronic support distance exoskeleton produced frequency and
exoskeleton no effect on velocity duration of

Overground but important clinical robotic
exoskeleton change on distance treatment
Robot-assisted gait No clustering
training induced no according to
effect on velocity but patient’s lesion
important clinical level
change on distance
Nam et al [34] 2017 10 Incomplete SCI RCT Yes
Body weight Effectiveness 10mWT The data for participants Different
acute support 6minWT <6 mo postinjury frequency and
chronic exoskeleton WISCI showed improvements duration of
MAS in walking distance, robotic
TUG lower limb strength, treatment
and functional level of No overground
mobility and exoskeleton and
independence for end-effector
robotic gait training
over conventional
overground gait
training
The data for participants
>1 year postinjury
showed improvements
in gait speed and
balance with robotic
gait training compared
with no intervention
Fisahn et al 2016 11 Chronic SCI RCT Yes
Body weight Effectiveness both 6minWT No consistent benefit Only Lokomat
[32] Complete support as assistive both 10mWT from exoskeleton No data to compare
incomplete exoskeleton as rehabilitative FIM rehabilitation versus locomotion
(10 studies with SCIM conventional assistance with
Lokomat) treatment in SCI exoskeleton
versus
conventional
knee-ankle-foot
orthosis
Miller et al 2016 14 Complete All Yes
Exoskeleton Clinical 6minWT Powered exoskeletons Different
[27] incomplete effectiveness allow patients with SCI frequency and
SCI to safely ambulate in duration of
real-world settings robotic
treatment
Louie et al 2015 15 Complete All No
Overground Ability of powered 10mWT Powered exoskeleton Different
[25] /incomplete exoskeleton exoskeleton to 2minWT allow individuals with frequency and
SCI provide gait (gait 6minWT complete SCI to walk duration of
Chronic speed) at 0.26 m/sdnot robotic

F. Molteni et al. / PM R 10 (2018) S174-S188

useful for community treatment
ambulation Small number of
participants
Different type of
exoskeleton
control
Morawietz 2013 8 Incomplete SCI RCT yes
Body weight Effectiveness Gait speed, walking No difference in acute Small sample
et al Acute/chronic support endurance, level of patients in gait speed Different
exoskeleton independence of between control group frequency and
walking, use of and exoskeleton group duration of
assistive devices and robotic
walking aids, step treatment
length, symmetry of
gait, or balance during
sit, stance, or
locomotion
Wessel et al 2010 18 Incomplete SCI All Yes
Body weight Effectiveness FIM Overground gait training Only 2 RCTs on
Chronic/acute support Walking velocity is equivalent to body acute patients
exoskeleton weight supported No standard
training to restore measures
walking function in
subacute incomplete
SCI
RCT ¼ Randomized controlled trials; 10mWT ¼ 10 meter walking test; 6minWT ¼ 6 minutes walking test; FIM ¼ Functional Independence Measure; WISCI ¼ Walking Index for Spinal Cord
Injury; MRC ¼ Medical Research Council; SCIM ¼ Spinal Cord Independence Measure; TUG ¼ Time Up and Go; MAS ¼ Modified Ashworth Scale; 2minWT ¼ 2 minutes walking test.

S181
Table 3

S182
Main Results of Clinical Review About the Use of Robotic Devices for Upper Limb in neurologic Patients
Year of Number of Type of Control
Authors Publication Publications Pathology studies Groups Aim Outcome Measures Results Limitation
Zhang et al 2017 13 Stroke (acute and RCT Yes Evaluate the Fugl-Meyer Upper Motor recovery Heterogeneity of study
[40] chronic) effectiveness of Limb (divided in (Fugl-Meyer score) designs and patients
robotic training (RT) proximal and in the RT group RT only in addiction to
and conventional distal) was significantly CT
training (CT) in greater than that
improving the motor in the CT group
recovery of paretic
upper limb
Singh et al 2018 12 Spinal Cord Injury RCTs non- Some Evaluate feasibility and ARAT Increased ROM, Methodologic quality
[41] (SCI) (cervical) RCTs clinic outcomes of JTHFT pinch and grip and risk of bias in
robot-assisted upper UEMS strength, UEMS non-RCTs, limited
extremity training in MAS and muscle quality of studies,
SCI patients AOU-MAL strength variability in study
designs, outcome
measurement tools,

Exoskeleton and End-Effector Robots

and training
protocols used
Bertani 2017 17 Stroke (acute and RCTs, 2 SR, Yes Effectiveness of Fugl-Meyer Robot-assisted Methodological
et al [39] chronic) 1 meta- different robotic MAS rehabilitation is limitation of trials,
analysis devices (end-effector FIM more effective in heterogeneity of
and exoskeleton MAL improving upper groups
robots) in comparison limb motor
with any other type of function recovery,
intervention especially in
chronic stroke
patients
Lo et al 2017 30 Stroke (acute and RCT Yes Effectiveness Fugl-Meyer No differences Heterogeneity of
[20] chronic) Scale Assessment between control studies
No differentiation (upper extremity) and exoskeleton No distinction between
group were found the 2 types of robots
No differentiation
between acute and
chronic stroke
Veerbeek 2017 44 Stroke (acute and RCTs Yes Effects of robot-assisted Fugl-Meyer Significant but small No distinction was
et al [38] chronic) therapy for the Assessment, improvements in made between
differentiation paretic upper limb muscle strength motor control (w2 different types of
(RT-UL) and tone, upper points FMA arm) control
limb capacity, and and muscle interventions,
basic ADL strength ignored findings
Negative effect on regarding movement
muscle tone kinematics
No effects for upper
limb capacity and
basic ADL
 F. Molteni et al. / PM R 10 (2018) S174-S188 S183

Veerbeek et al [38] considered a large number of

ARAT ¼ Action Research Arm Test; JTHFT ¼ Jebsen Taylor Hand Function Test; UEMS ¼ ASIA upper extremity motor scores; MAS ¼ Modified Ashworth Scale; AOU-MAL ¼ Amount of Use Scale of
Heterogeneity of study
designs and patients
Type of treatment and

Evidence quality from

training, intensity,
randomized controlled trials. They evaluated both the
Different amount of

Different amount of

No differentiation
low to very low
proximal and distal part of the upper limb, and showed

characteristics

subacute, and
chronic phase
among acute,
and duration results on the impact of robotic treatment with signifi-
participant cant but small improvements in motor control (w2
training

points Fugl-Meyer Assessment for the arm) and muscle

strength of the paretic arm, but a negative effect on
muscle tone. No effects were found for upper limb ca-
pacity and activities of daily living. Shoulder/elbow ro-
Improvement in ADL,

Improvement in Fugl-
hand functions and
bilateral training is

Meyer assessment
muscles strength

and FIM but no botics showed small but significant effects on motor
more effective

control and muscle strength, whereas elbow/wrist ro-

No evidence if

treatments
than other

change in
spasticity botics had small but significant effects only on motor
control. They found robotic therapy useful in increasing
the number of repetitions and hence intensity of prac-
tice post stroke, but studies were not homogeneous for
baseline time after the acute event and limited findings
Motor status score

could mainly be related to a poor understanding of

Range of Motion
Motricity Index

robot-induced motor learning.

Barthel Index

Barthel Index

Bertani et al [39] in 2017 focused not only on the

Fugl-Meyer

Fugl-Meyer
Fugl-Meyer

Wolf Test

the Motor Activity Log; FIM ¼ Functional independence measure; MAL ¼ Motor activity log; MRC ¼ Medical Research Council

number of repetitions but also on task-oriented robotic

activity and on the possibility to exactly quantify the
MRC

MAS
FIM

FIM

FIM

intensity of robotic treatment. Moreover, they

compared 2 different subgroups, end-effectors and
exoskeletons, and found a difference in arm function
(Fugl-Meyer) in the exoskeleton group and not in the
end-effector group. Authors stated that this effect
Effectiveness

Effectiveness

Effectiveness

could be due to a potentiation of cortical plasticity

within the affected hemisphere, besides a reduction of
the interhemispheric inhibition. They concluded that in
comparison with conventional therapy, the robot-
assisted rehabilitation treatment was more effective
in improving upper limb motor function recovery,
Yes

Yes

Yes

especially in chronic stroke patients. No significant im-

provements were observed in the reduction of muscle
tone or daily living activities.
Lo et al [20] in 2017 assessed the effectiveness of
RCT

RCT

RCT

robotic treatment compared to conventional treatment

and concluded that robotic training is as effective as
conventional training for upper limb motor movement
and for activities of daily living, when the number of
movement repetition was the same in a single session.
Recently, Zhang et al [40] evaluated the effective-
Stroke

Stroke

Stroke

ness of robotic training and conventional training in

improving the motor recovery of paretic upper limbs
in stroke patients. They found significant differences in
motor recovery between the conventional and robotic
groups in chronic stroke patients but not in the acute/
subacute stage. Moreover, they focused their attention
34

34

30

on how new robotic instruments can drive and enhance

the relearning process through multisensorial affer-
ences using sensorimotor feedbacks across a
2016

2015

2011

3-dimensional workspace, such as visual and auditory

feedback during training sessions, and reducing fatigue
by supporting upper limb in the early phase of rehabil-
et al [36]
Sheng et al

Peter et al
Mehrholz

itation after stroke. In the randomized controlled trials

[37]

[35]

that they considered, robotic treatment was always

delivered as supplementary therapy, but results of
S184 Exoskeleton and End-Effector Robots

combined therapy were promising in terms of motor the various robotic systems used (ie, end-effector,
recovery of the upper limb. exoskeletons, proximal, distal, unilateral, bilateral)
for the rehabilitation of the upper or lower limbs. Cli-
Robotics for Upper Limbs Following SCI nicians should understand the heterogeneity of the ro-
botic systems to properly adapt them to the
The only review focusing on exoskeletons for upper heterogeneity of the patients.
limb rehabilitation in SCI patients, published by Singh
et al [41] in 2018, included both assistive and resistive Exoskeletons and End-Effectors: Fine-Tuning
devices for rehabilitation training. They concluded that and Tailored Treatments
the use of robots for upper extremity in SCI appears to
be feasible in terms of safety and tolerance and has From a technical point of view, there are substantial
some beneficial utility in both inpatient and outpatient, differences between end-effector and exoskeletons in
but no evidence can be affirmed; authors in conclusions terms of operating methods, control of movements,
highlight the need for more rigorous studies with a degree of assistance, degrees of freedom, and move-
larger experimental group. ments performed. Both end-effectors and exoskeleton
robotic devices have strengths and weaknesses. It is,
Discussion therefore, important to consider the rationale of the 2
types of devices and the related benefits or disadvan-
Evidence and Limitations tages of each.
It is noteworthy that recent clinical research mainly
There is general consensus that rehabilitative in- focused on strength, speed, and endurance recovery
terventions are more effective if they ensure early, with robotics without considering movement quality.
intensive, task-specific, and multisensory stimulations Robotic exoskeletons offer software settings that
with both bottom-up and top-down integration. Robotic allow clinicians to adjust the level of motor assistance
systems are well suited to produce intensive, task- and grade the assistance level up or down as the pa-
oriented motor training for moving the patient’s limbs tient’s conditions progresses. It is possible to customize
under the supervision/help of a therapist. In this each training session according to the patient’s char-
context, robots would enhance conventional rehabili- acteristics. Robotic exoskeletons can be adjusted to
tation via intense and task-oriented training [42]. provide the minimum amount of motor power required,
Some reviews on robotic gait training highlighted allowing to maximize patients’ strength and motor
better outcomes in patients who used end-effectors if function while working within the programed trajectory
compared to stationary exoskeletons both for stroke and to control movement quality. Software advances have
SCI; some others found the superiority of robotic also allowed clinicians not only to use exoskeletons as
treatment (both exoskeleton and end-effector) if an assistance tool but also to offer resistance to the
compared to conventional treatment group. All the re- desired muscles and to modify training time to further
views outlined limitations in study design, population, challenge the patient and improve strength and endur-
and assessment analysis with varied treatment in- ance [43].
tensity, frequency, and small numbers of subjects. In literature reviews, it is never specified how the
The same conclusions could be drawn for the reviews setting of the robot is performed in relation to the pa-
on upper limb rehabilitation, with adjunctive issues tient’s characteristics; most reviews pointed out the
arising from the correct clinical evaluation of the necessity to define guidelines for standardized rehabil-
complexity of the tasks required to the upper limb in itation therapeutic protocols in order to optimize robots
activities of daily living. Moreover, these reviews use. We should change the idea of a standardized pro-
involved a multitude of different devices and rarely was tocol for all the patients and promote the personaliza-
there a clear differentiation between groups treated tion of treatment, adapting the robots to the patient’s
with end-effector or exoskeletons. The duration of the clinical condition. In this context, it might be useful to
single session and the number of repetitions were rarely use tools such as electroencephalography or electro-
specified, and the “conventional treatment” could myography to optimize the patient robot interaction.
consistently differ between various studies. The fine-tuning of the exoskeleton control system is
The present literature review highlights a huge crucial, and can be set according to the residual func-
heterogeneity of robotic systems used both for walking tional abilities of the patient. The interaction between
and for the upper limb, an extreme variability of pro- exoskeleton and the patient is both physical and
posed treatments, and clear limitations in the meth- cognitive [44]. Physical Human-Robot Interaction in-
odological aspects of the studies. These aspects have cludes the generation of supplementary forces to
contributed in making the results of clinical reviews on overcome human physical limits. Cognitive Human-
the clinical use of robotic devices inconclusive so far. In Robot Interaction highlights the possibility of maintain-
clinical studies, there is no clear difference between ing the control of the robot from the human. Given the
 F. Molteni et al. / PM R 10 (2018) S174-S188 S185

use of a commercial device, both aspects of Human- when it comes into contact with augmenting artificial
Robot Interaction is dependent on the robotic device sensors and effectors and, on the other hand, the
proper setting; the fine-tuning procedure is necessary to changes that the use of external augmenting devices
ensure the best power transfer between subject and produces in the brain). The neural correlates for
robot. Surface electromyography of key muscles con- augmentation-related plasticity are sensorimotor
trolling multijoint coordination of upper/lower limbs is training, cognitive enhancement, cross-modal plas-
an effective way to noninvasively define motor control ticity, sensorimotor functional substitution, and use and
during spontaneous movements. embodiment of tool [49].
For wearable robotic overground exoskeletons, it Within the motor domain, a substantial body of
would be very important to identify electromyographic literature is devoted to the motor system plasticity
paradigms that are able to define the optimal exoskel- induced by robotic training, whereas very little atten-
eton setting (ie, oscillation on the frontal plane, the tion is placed on how robotics could affect the sensory
step length, swing time, trunk oscillation and so on); domain and how the acquisition or improvement of
this could lead to optimize the interaction between the robot-mediated sensory functions could be accompa-
user and the robot. Therefore, patient-robot interaction nied by plastic changes in the brain.
is a crucial key point for exoskeleton use in It is necessary to keep in mind that the sensory
rehabilitation. components play a crucial role, maybe as important as
the motor components, in the enhancement of a
Neuroplasticity and Body Perception: New performance.
Perspectives Technologies such as exoskeletons can give percep-
tion of near-physiologic movements and redirect motor
Overground exoskeletons require that the patient experiences that could induce, through the emulation of
actively interfaces with the exoskeleton. With powered the physiologic motor control, a new organizational
wearable exoskeletons, the patient is responsible for model of the central nervous system [50].
maintaining trunk and balance control and for navi- Technology could become a way for inducing motor
gating over different surfaces. Contribution of visual- experiences not otherwise possible with the work of the
spatial and vestibular components, as well as patient physiotherapist alone [51]. If this is true, we need to
engagement, is required. move from a concept of repetitions or force generation
The mechanism of visual exploration and visual- to a concept of perception of the movement.
motor coordination are completely different during When a movement is made under the robotic control,
the training with a wearable overground exoskeleton or in addition to the typical motor areas, brain areas such
with a stationary treadmill exoskeleton. The integration as the insula, the amygdala, and other neural circuits
of vestibular and somatosensory visual inputs is crucial connected to the deepest centers are activated. These
for the restoration of motor control, the repair of the centers are important in determining movement mem-
body schema, and for cortical mechanisms involved in ory and motivation to the movement [52]. So, the use of
gait control [45]. robotic instruments could induce a series of sensory
Emerging and advancing robotic technologies can processes related to a complex internal representation
enhance clinical therapeutic techniques by allowing of the movements, and such processes, through stimu-
therapists to activate and/or modulate neural networks lation, emulation, and imagination, could find a way to
otherwise lost, to maximize recovery and functional be stored in the central nervous system. Exoskeleton-
outcomes [46]. Despite most studies claiming that robots driven movements could become not only an action
would increase rehabilitation intensity, repetition of per se but a feeling, an idea, and a new motor planning
tasks alone is not sufficient to guide neural plasticity modality that is structurally coupled with the reprog-
[47]. In this perspective, robotics should not be consid- ramming of the motion control experience according to
ered a simple substitution of physiotherapists to make the principles of motor imagery and action observation.
patients perform only intensive and repetitive stereo- Methods of perceptual-motor learning can make the
typed movement patterns, as in the case of most of the patients easily adapted and trained to these new con-
existing devices, maybe with little patient participation ditions, as well as implicit or explicit learning modal-
and emotional engagement. Robotic treatment should ities, according to the experiences that technology
be considered a rehabilitation tool useful to generate a allows to do [53]. Sensory and motor systems are syn-
more complex, controlled multisensory stimulation of chronously engaged during perceptual-motor perfor-
the patient and to modify the plasticity of neural con- mance induced by robotic interventions, and this could
nections through the experience of movement [48]. facilitate perceptual-motor skill learning and transfer
Today’s achievements in biomedical sciences and [54]. Dynamic technology-mediated interactions be-
engineering need to reconsider the concept of human- tween the subject and the environment could induce
robot interaction in the framework of human augmen- more adaptive plasticity mechanisms, avoiding mal-
tation (the plastic process that the brain undergoes adaptive plasticity.
S186 Exoskeleton and End-Effector Robots

Another key point is that robotic exoskeleton sys- one as a new way to perform sensory-motor and
tems, whether for the upper or lower limb, could help emotional rehabilitation as a part of a complex neural
an engaged patient to give intention to the intensity of network repair process for adaptive plasticity.
the exercise. Intention is as important as action. Psy-
chological states such as intention, motivation, and Conclusions
engagement are known to be critical for the success of
rehabilitation [55]. Exoskeletons are a neural interface for rehabilitation
Therefore, it is fundamental to move from the to exploit and promote neural plasticity. To define the
concept of empowerment, that is, a function of the efficacy and efficiency of the treatment using exo-
intensity of the training or the number of repetitions, to skeletons we have to analyze intensity, complexity and
the concept of perception of movement, followed by specificity of the therapeutic exercise performed, that
intention of movement, followed by motivational and are related to human-robot interaction in terms of
emotional intensity. The number of repetitions made by motion, emotion, motivation, meaning of the task,
a robotic system thus are a consequence and not the feedback from the exoskeleton, and fine motion assis-
focus of treatment. tance. We are moving to change completely the con-
Recent studies have pointed out that the strong ventional treatment in terms of intensity and
connection between the exoskeletons and body complexity of therapeutic exercises as well as of a
perception, often termed “embodiment [56],” could be precise tailored treatment, depending on the ability of
one of the most crucial factors affecting functional re- the patient to optimize modulation and enhancement
covery [57]. This area of research has great potential for neuroplasticity. Duration of single session and global
promoting humanmachine interactions and for poten- period of the treatment is important as well as when to
tiating adaptive plasticity mechanisms that can sustain start after the onset of the disease. It is possible to
recovery. conclude that exercise through wearable technological
Very recently, Pazzaglia et al [58] focused on the devices open a window to define exercise as a beneficial
aspects of cortical reorganization, functional adapta- drug that can be prescribed as a medicine to boost
tion and multisensory changes of bodily signals in SCI biological, neurobiological, and epigenetic changes.
patients and concluded that these aspects are currently There are open questions related to the definition of
the major neurobiological mechanisms underlying bodily complex adaptive fine-tuning of the control interactions
experience of an assistive device. Embodying a robotic between sensors, interfaces, actuators, and clinical
exoskeleton may enhance the competence and safety of management of the rehabilitation processes through
movements. Expressing physical body awareness of a exoskeletons. We need to implement large and innova-
tool is essential to experiencing a real sense of acting tive research programs able to provide answers in the
in/on the world. This is particularly true when exo- near future. It is time to change.
skeletons are used as assistive devices to reengage in
everyday activities. References
The authors concluded that to change and/or
improve the clinical condition of an exoskeleton wearer, 1. Perry J, Burnfield JM. Gait Analysis, Normal and Pathological
there is a need to endow the patients (especially those Function. 2nd ed. Thorofare, NJ: Slack Inc; 2010.
2. Caimmi M, Guanziroli E, Malosio M, et al. Normative data for an
with SCI) with a strong sense of identification with, and
instrumental assessment of the upper-limb functionality. Biomed
effective inclusion of, wearable robotic legs (embodi- Res Int 2015;2015:484131.
ment), to provide the user with better control over their 3. Morasso P, Casadio M, Mohan V, Rea F, Zenzeri J. Revisiting the
actions (agency), and to take full advantage of mecha- body-schema concept in the context of whole-body postural-focal
nisms that enhance multimodal processing and dynamics. Front Hum Neurosci 2015;9:83.
4. Sale A, Berardi N, Maffei L. Enrich the environment to empower
neuroplasticity.
the brain. Trends Neurosci 2009;32:233-239.
The embodiment of a wearable exoskeleton is an 5. Gerloff C, Bushara K, Sailer A, et al. Multimodal imaging of brain
important issue in people suffering from complete SCI. reorganization in motor areas of the contralesional hemisphere
An embodied exoskeleton could be utilized not only to of well recovered patients after capsular stroke. Brain 2006;
provide a walking experience for nonfunctional ambu- 129(pt 3):791-808.
6. Nudo RJ. Postinfarct cortical plasticity and behavioral recovery.
lators but could allow the patients to decide how,
Stroke 2007;38(2 suppl):840-845.
where, and when to move and how to interact with the 7. Stroemer RP, Kent TA, Hulsebosch CE. Acute increase in expression
environment. The position in which a patient decides to of growth associated protein GAP-43 following cortical ischemia in
move and the upright ability to explore the space using rat. Neurosci Lett 1993;162:51-54.
a wearable exoskeleton that provides all the possible 8. Li S, Carmichael ST. Growth-associated gene and protein expres-
sion in the region of axonal sprouting in the aged brain after
sensations from a sensory-motor point of view deeply
stroke. Neurobiol Dis 2006;23:362-373.
change the rehabilitation perspective. Exoskeleton- 9. Alia C, Spalletti C, Lai S, et al. Neuroplastic changes following
robotic treatment should be considered not only in the brain ischemia and their contribution to stroke recovery: Novel
empowerment framework but also in the embodiment approaches in neurorehabilitation. Front Cell Neurosci 2017;11:76.
 F. Molteni et al. / PM R 10 (2018) S174-S188 S187

10. Feys H, De Weerdt W, Verbeke G, et al. Early and repetitive 30. Aguirre-Güemez AV, Pérez-Sanpablo AI, Quinzaños-Fresnedo J,
stimulation of the arm can substantially improve the long-term Pérez-Zavala R, Barrera-Ortiz A. Walking speed is not the best
outcome after stroke: A 5-year follow-up study of a randomized outcome to evaluate the effect of robotic assisted gait
trial. Stroke 2004;35:924-929. training in people with motor incomplete spinal cord injury: A
11. Lo AC, Guarino PD, Richards LG, et al. Robot-assisted therapy for systematic review with meta-analysis. J Spinal Cord Med
long-term upper-limb impairment after stroke. N Engl J Med 2010; 2017:1-13.
362:1772-1783. 31. Mehrholz J, Harvey LA, Thomas S, Elsner B. Is body-weight-
12. Klamroth-Marganska V, Blanco J, Campen K, et al. Three-dimen- supported treadmill training or robotic-assisted gait training
sional, task-specific robot therapy of the arm after stroke: A superior to overground gait training and other forms of physio-
multicentre, parallel-group randomised trial. Lancet Neurol 2014; therapy in people with spinal cord injury? A systematic review.
13:159-166. Spinal Cord 2017;55:722-729.
13. Hesse S, Uhlenbrock D. A mechanized gait trainer for restoration of 32. Fisahn C, Aach M, Jansen O, et al. The effectiveness and safety
gait. J Rehabil Res Dev Clin Suppl 2000;37:701-708. of exoskeletons as assistive and rehabilitation devices in the
14. Hesse S, Waldner A, Tomelleri C. Innovative gait robot for the treatment of neurologic gait disorders in patients with spinal
repetitive practice of floor walking and stair climbing up and down cord injury: A systematic review. Global Spine J 2016;6:
in stroke patients. J Neuroeng Rehabil 2010;7:30. 822-841.
15. Freivogel S, Mehrholz J, Husak-Sotomayor T, Schmalohr D. Gait 33. Cheung EYY, Ng TKW, Yu KKK, Kwan RLC, Cheing GLY. Robot-
training with the newly developed “LokoHelp”dsystem is feasible assisted training for people with spinal cord injury: A meta-anal-
for non-ambulatory patients after stroke, spinal cord and brain ysis. Arch Phys Med Rehabil 2017;98:2320-2331.e12.
injury. A feasibility study. Brain Inj 2008;22:625-632. 34. Nam KY, Kim HJ, Kwon BS, Park JW, Lee HJ, Yoo A. Robot-assisted
16. Chang WH, Kim YH. Robot-assisted therapy in stroke rehabilita- gait training (Lokomat) improves walking function and activity in
tion. J Stroke 2013;15:174-181. people with spinal cord injury: A systematic review. J Neuroeng
17. Chen G, Chan CK, Guo Z, Yu H. A review of lower extremity Rehabil 2017;14:24.
assistive robotic exoskeletons in rehabilitation therapy. Crit Rev 35. Péter O, Fazekas G, Zsiga K, Dénes Z. Robot-mediated upper limb
Biomed Eng 2013;41:343-363. physiotherapy: Review and recommendations for future clinical
18. Mehrholz J, Thomas S, Werner C, Kugler J, Pohl M, Elsner B. trials. Int J Rehabil Res 2011;34:196-202.
Electromechanical-assisted training for walking after stroke. 36. Mehrholz J, Pohl M, Platz T, Kugler J, Elsner B. Electromechanical
Cochrane Database Syst Rev 2017;5:CD006185. and robot-assisted arm training for improving activities of daily
19. Bruni MF, Melegari C, De Cola MC, Bramanti A, Bramanti P, living, arm function, and arm muscle strength after stroke.
Calabrò RS. What does best evidence tell us about robotic gait Cochrane Database Syst Rev 2015:CD006876.
rehabilitation in stroke patients: A systematic review and meta- 37. Sheng B, Zhang Y, Meng W, Deng C, Xie S. Bilateral robots for
analysis. Clin Neurosci 2018;48:11-17. upper-limb stroke rehabilitation: State of the art and future
20. Lo K, Stephenson M, Lockwood C. Effectiveness of robotic assisted prospects. Med Eng Phys 2016;38:587-606.
rehabilitation for mobility and functional ability in adult stroke 38. Veerbeek JM, Langbroek-Amersfoort AC, van Wegen EE,
patients: A systematic review. JBI Database System Rev Implement Meskers CG, Kwakkel G. Effects of robot-assisted therapy for
Rep 2017;15:3049-3091. the upper limb after stroke. Neurorehabil Neural Repair 2017;
21. Ada L, Dean CM, Vargas J, Ennis S. Mechanically assisted walking 31:107-121.
with body weight support results in more independent walking 39. Bertani R, Melegari C, De Cola MC, Bramanti A, Bramanti P,
than assisted overground walking in non-ambulatory patients early Calabrò RS. Effects of robot-assisted upper limb rehabilitation in
after stroke: A systematic review. J Physiother 2010;56:153-161. stroke patients: A systematic review with meta-analysis. Neurol
22. Swinnen E, Beckwée D, Meeusen R, Baeyens JP, Kerckhofs E. Does Sci 2017;38:1561-1569.
robot-assisted gait rehabilitation improve balance in stroke pa- 40. Zhang C, Li-Tsang CW, Au RK. Robotic approaches for the reha-
tients? A systematic review. Top Stroke Rehabil 2014;21:87-100. bilitation of upper limb recovery after stroke: A systematic review
23. Louie DR, Eng JJ. Powered robotic exoskeletons in post-stroke and meta-analysis. Int J Rehabil Res 2017;40:19-28.
rehabilitation of gait: A scoping review. J Neuroeng Rehabil 41. Singh H, Unger J, Zariffa J, et al. Robot-assisted upper extremity
2016;13:53. rehabilitation for cervical spinal cord injuries: A systematic
24. Esquenazi A, Talaty M, Jayaraman A. Powered exoskeletons for scoping review. Disabil Rehabil Assist Technol 2018:1-12.
walking assistance in persons with central nervous system injuries: 42. Morone G, Paolucci S, Cherubini A, et al. Robot-assisted gait
A narrative review. PM R 2017;9:46-62. training for stroke patients: Current state of the art and per-
25. Louie DR, Eng JJ, Lam T. Spinal Cord Injury Research Evidence spectives of robotics. Neuropsychiatr Dis Treat 2017;13:1303-1311.
(SCIRE) Research Team. Gait speed using powered robotic exo- 43. Fritz H, Patzer D, Galen SS. Robotic exoskeletons for reengaging in
skeletons after spinal cord injury: A systematic review and everyday activities: Promises, pitfalls, and opportunities. Disabil
correlational study. J Neuroeng Rehabil 2015;12:82. Rehabil 2017:1-4.
26. Forrest GF, Lorenz DJ, Hutchinson K, et al. Ambulation and bal- 44. Pons JL. Rehabilitation exoskeletal robotics. The promise of an
ance outcomes measure different aspects of recovery in in- emerging field. IEEE Eng Med Biol Mag 2010;29:57-63.
dividuals with chronic, incomplete spinal cord injury. Arch Phys 45. Molteni F, Gasperini G, Gaffuri M, et al. Wearable robotic
Med Rehabil 2012;93:1553-1564. exoskeleton for overground gait training in sub-acute and chronic
27. Miller LE, Zimmermann AK, Herbert WG. Clinical effectiveness and hemiparetic stroke patients: Preliminary results. Eur J Phys
safety of powered exoskeleton-assisted walking in patients with Rehabil Med 2017;53:676-684.
spinal cord injury: Systematic review with meta-analysis. Med 46. Krucoff MO, Rahimpour S, Slutzky MW, Edgerton VR, Turner DA.
Devices (Auckl) 2016;9:455-466. Enhancing nervous system recovery through neurobiologics, neural
28. Wessels M, Lucas C, Eriks I, de Groot S. Body weight-supported gait interface training, and neurorehabilitation. Front Neurosci 2016;
training for restoration of walking in people with an incomplete 10:584.
spinal cord injury: A systematic review. J Rehabil Med 2010;42: 47. Plautz EJ, Milliken GW, Nudo RJ. Effects of repetitive motor
513-519. training on movement representations in adult squirrel monkeys:
29. Morawietz C, Moffat F. Effects of locomotor training after incom- Role of use versus learning. Neurobiol Learn Mem 2000;74:27-55.
plete spinal cord injury: A systematic review. Arch Phys Med 48. Peters S, Handy TC, Lakhani B, Boyd LA, Garland SJ. Motor and
Rehabil 2013;94:2297-2308. visuospatial attention and motor planning after stroke:
S188 Exoskeleton and End-Effector Robots

Considerations for the rehabilitation of standing balance and gait. 54. Müller S, Vallence AM, Winstein C. Investigation of perceptual-
Phys Ther 2015;95:1423-1432. motor behavior across the expert athlete to disabled patient
49. Di Pino G, Maravita A, Zollo L, Guglielmelli E, Di Lazzaro V. skill continuum can advance theory and practical application. J
Augmentation-related brain plasticity. Front Syst Neurosci 2014; Mot Behav 2017:1-11.
8:109. 55. Yue Z, Zhang X, Wang J. Hand rehabilitation robotics on poststroke
50. Jarrassé N, Proietti T, Crocher V, et al. Robotic exoskeletons: A motor recovery. Behav Neurol 2017;2017:3908135.
perspective for the rehabilitation of arm coordination in stroke 56. Pazzaglia M, Galli G, Scivoletto G, Molinari M. A functionally
patients. Front Hum Neurosci 2014;8:947. relevant tool for the body following spinal cord injury. PLoS One
51. Wang W, Collinger JL, Perez MA, et al. Neural interface technology 2013;8:e58312.
for rehabilitation: Exploiting and promoting neuroplasticity. Phys 57. Hellman RB, Chang E, Tanner J, Helms Tillery SI, Santos VJ. A
Med Rehabil Clin N Am 2010;21:157-178. robot hand testbed designed for enhancing embodiment
52. Nocchi F, Gazzellini S, Grisolia C, et al. Brain network involved in and functional neurorehabilitation of body schema in subjects
visual processing of movement stimuli used in upper limb robotic with upper limb impairment or loss. Front Hum Neurosci 2015;
training: An fMRI study. J Neuroeng Rehabil 2012;9:49. 9:26.
53. Makin TR, de Vignemont F, Faisal AA. Neurocognitive barriers to 58. Pazzaglia M, Molinari M. The embodiment of assistive devices-from
the embodiment of technology. Nat Biomed Eng 2017;1. wheelchair to exoskeleton. Phys Life Rev 2016;16:163-175.

Disclosure

F.M. Valduce Hospital “Villa Beretta” Rehabilitation Center, Costa Masnaga, G.C. Valduce Hospital “Villa Beretta” Rehabilitation Center, Costa Masnaga,
Italy Italy
Disclosures related to this publication: grant, Ekso Bionics, Rewalk; disclosures Disclosure: nothing to disclose
outside this publication: Ekso Bionics
E.G. Valduce Hospital “Villa Beretta” Rehabilitation Center, Via N. Sauro 17,
G.G. Valduce Hospital “Villa Beretta” Rehabilitation Center, Costa Masnaga, Costa Masnaga, Italy. Address correspondence to: E.G.; e-mail: eleonora.
Italy guanziroli@gmail.com
Disclosure: nothing to disclose Disclosure: nothing to disclose