Rehabilitation robots for the treatment of sensorimotor deficits: a neurophysiological perspective

Rehabilitation robots for the treatment of sensorimotor deficits: a neurophysiological perspective The past decades have seen rapid and vast developments of robots for the rehabilitation of sensorimotor deficits after damage to the central nervous system (CNS). Many of these innovations were technology-driven, limiting their clinical application and impact. Yet, rehabilitation robots should be designed on the basis of neurophysiological insights underlying normal and impaired sensorimotor functions, which requires interdisciplinary collaboration and background knowledge. Recovery of sensorimotor function after CNS damage is based on the exploitation of neuroplasticity, with a focus on the rehabilitation of movements needed for self-independence. This requires a physiological limb muscle activation that can be achieved through functional arm/hand and leg movement exercises and the activation of appropriate peripheral receptors. Such considerations have already led to the development of innovative rehabilitation robots with advanced interaction control schemes and the use of integrated sensors to continuously monitor and adapt the support to the actual state of patients, but many challenges remain. For a positive impact on outcome of function, rehabilitation approaches should be based on neurophysiological and clinical insights, keeping in mind that recovery of function is limited. Consequently, the design of rehabilitation robots requires a combination of specialized engineering and neurophysiological knowledge. When appropriately applied, robot- assisted therapy can provide a number of advantages over conventional approaches, including a standardized training environment, adaptable support and the ability to increase therapy intensity and dose, while reducing the physical burden on therapists. Rehabilitation robots are thus an ideal means to complement conventional therapy in the clinic, and bear great potential for continued therapy and assistance at home using simpler devices. This review summarizes the evolution of the field of rehabilitation robotics, as well as the current state of clinical evidence. It highlights fundamental neurophysiological factors influencing the recovery of sensorimotor function after a stroke or spinal cord injury, and discusses their implications for the development of effective rehabilitation robots. It thus provides insights on essential neurophysiological mechanisms to be considered for a successful development and clinical inclusion of robots in rehabilitation. Keywords: Robot-assisted therapy, Neurorehabilitation technology, Assist-as-needed, Stroke, Spinal cord injury, Locomotion, Upper limb function, Sensorimotor neurophysiology, Neuroplasticity * Correspondence: roger.gassert@hest.ethz.ch Department of Health Sciences and Technology, ETH Zurich, 8092 Zurich, Switzerland Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 2 of 15 Background from passive movements for the most severely impaired Rehabilitation robotics is a relatively young and rapidly patients, to active-assisted and active-resisted move- growing field, with increasing penetration into the clin- ments in moderately impaired patients. Furthermore, as- ical environment [1]. In the late 1980s and early 90s a sistance could be automatically adapted to the patient’s number of pioneering technological developments were performance. Around the same time, the Mirror Image launched, triggered by discoveries on training-induced Motion Enabler (MIME; [12]) was introduced, which recovery of sensorimotor function in animal models with supported paretic limb movements with a stiff industrial damage to the central nervous system (CNS). The goal robot, controlled by the non-paretic limb by means of a was to enhance the effects of functional training by pro- motion digitizer (mirror-image therapy mode). viding increased therapy intensity and adaptive support Developments of rehabilitation robots for the lower in a controlled way. extremity began in 1994, with the design of the Lokomat The idea of using machines for rehabilitation dates [13], combining body-weight supported treadmill- back much earlier. In a 1910 patent, Theodor Büdingen training (BWSTT) with the assistance of a robotic gait proposed a ‘movement cure apparatus’, a machine driven orthosis. The Gait Trainer [14] realized a similar concept by an electric motor to guide and support stepping based on an end-effector design. movements in patients with heart disease. In the 1930s, The decades since these pioneering developments have Richard Scherb developed the ‘meridian’, a cable-driven seen an explosion of novel rehabilitation robots for both apparatus to move joints for orthopaedic therapy. This the upper and lower extremities, which can broadly be human-powered mechanotherapy machine already sup- classified into grounded exoskeletons, grounded end- ported multiple interaction modes, ranging from passive effector devices, and wearable exoskeletons (Fig. 1). to active-assisted and active-resisted movements. A first These design approaches affect the level of control over robotic rehabilitation system was based on the concept the interaction (control of individual joints in exoskel- of continuous passive motion (CPM), a stiff interaction eton devices vs control over selected joints or limb seg- mode in which the robot moves the joints along a prede- ments in grounded end-effector devices) as well as the fined trajectory, independent of the contribution of the output impedance of the device (resulting from the patient [2]. mechanical structure as well as actuator and transmis- The first powered exoskeletons for therapeutic applica- sion properties) and the ability to modulate this imped- tions in SCI patients were introduced in the 1970s [3–5]. ance through control. Grounded end-effector devices These systems used pneumatic, hydraulic, or electromag- will typically achieve higher motion dynamics and allow netic (via cams and Bowden cables) actuators for position the rendering of a wider range of impedances than exo- servocontrol. They included advanced features, such as skeleton devices with a serial kinematic structure, where actuated ankle flexion/extension, and hip adduction/ab- proximal joints need to move distal joints. The latter re- duction for increased stability [6] or the ability of a therap- quires large reduction ratios and results in high inertia ist to control the motion of the exoskeleton worn by the and friction at the output where the patient is attached patient through his/her own movement (in a similar, con- [15, 16]. These dynamics can only partially be compen- nected exoskeleton) [7]. The first system for robot- sated through control. assisted therapy of stroke survivors [8] was based on a stiff The number of new developments has been dispropor- industrial manipulator and did not physically interact with tionate to the penetration of these technologies into the patients, but rather moved a pad that patients had to clinical setting, likely due to the technology-driven ap- touch to different locations. proach of many engineering groups and the limited, al- A new era of neurorehabilitation robotics began in beit increasing, exchange of the field with therapists and 1989 with the development of the MIT-MANUS [9], clinicians. While a few randomized-controlled trials have which was first tested clinically in 1994. Compared to in- confirmed efficacy of robot-assisted therapy equivalent dustrial manipulators, this planar manipulandum pre- to that of dose-matched conventional therapy [17–21], sents inherently low mechanical output impedance (a the majority of published devices were never clinically frequency-dependent resistance to motion perceived at evaluated, or such an evaluation was limited to pilot the interface between the human user and the robotic studies on a few patients. Interestingly, many of these system) and provides unloading of the upper limb studies unsuccessfully aimed to demonstrate superiority against gravity, thereby allowing to adapt support to the of robot-assisted as compared to conventional therapy, severity of the deficits. A few years later, force controlled despite the fact that there is currently no consensus on devices for bimanual, cooperative grasping [10] and lift- the optimal therapy program for an individual patient in ing [11] were introduced. This new generation of de- the clinical field. vices, using torque-controlled direct drive actuation, For a successful inclusion of robots in rehabilitation, allowed for more advanced interaction control, ranging fundamental knowledge about the physiological basis of Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 3 of 15 Fig. 1 Schematic representation and classification of rehabilitation robots. Besides the extremity that is trained, rehabilitation robots can be broadly classified into grounded exoskeletons, end-effector devices and wearable exoskeletons. While the first two are well established, the latter are currently entering clinical application [17–21, 95, 122, 130–140] the recovery of function is required. This knowledge is for upper and lower limb functions that should be con- widely distributed and difficult for engineers to access sidered for the design of effective rehabilitation robots, and translate into design considerations. Consequently, and underlines the need for transdisciplinary collabora- in our opinion and experience, a close cooperation be- tions for future developments. Before addressing aspects tween engineers, therapists and clinical neurophysiolo- specific to upper and lower limb rehabilitation, general gists/neurorehabilitation scientists is required from the neurophysiological considerations of relevance for the very beginning of a development, and was shown to be design of rehabilitation robots will be discussed. successful in previous developments (e.g. of the Lokomat with the involvement of VD, a neurologist/clinical Neurophysiological basis for the recovery of neurophysiologist [13]). sensorimotor function after CNS damage According to evidence from studies in cats [22], non- Stroke and spinal cord injury (SCI) are among the lead- human primates [23], and humans [24], recovery of sen- ing causes of adult long-term physical disability, with ap- sorimotor function after damage to the central nervous proximately 10 million people surviving a stroke and system (CNS) is based on the exploitation of neuroplas- over 250′000 surviving a spinal cord injury (of which ap- ticity. It relies on physiological limb activation during proximately 60% are incomplete) every year. Muscle the training of functional arm and hand movements, and weakness due to activation deficits represents the main the stimulation of appropriate peripheral receptors dur- disability following stroke and SCI, and frequently limits ing automatically performed leg movements such as self-independence. Furthermore, following CNS damage, stepping. Rehabilitation robots should therefore enable secondary effects such as spastic muscle tone (increased and support such functional training. resistance to passive stretch) develop. This review aims to provide historical and clinical The aim of neurorehabilitation is to improve outcome background of relevance to the field of rehabilitation ro- of function after damage to the CNS, such as stroke and botics for engineers, basic and clinical neurophysiolo- SCI, through intensive physical therapy. This goal is, gists and therapists interested in and entering this however, difficult to define as the effects of conventional exciting field. It introduces the neurophysiological basis therapy can hardly be separated from the spontaneous Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 4 of 15 recovery of function that occurs in parallel to the effects support during stance and gait in a stick-like manner of the rehabilitative treatment [25]. In stroke [26] and [39]. However, this only holds for moderately affected, SCI [27], most of the spontaneous recovery occurs mobile patients, while in severely disabled patients, spas- within the first three months. tic signs such as muscle cramps may become exagger- Therapy-induced recovery is mediated by neuroplasti- ated, requiring pharmaceutical interventions. city, and the goal of rehabilitation is thus to maximally There is currently no consensus on the optimal ther- exploit neuroplasticity in order to achieve an optimal apy programs to promote recovery of motor function outcome for the individual patient. However, neuroplas- following CNS damage, and the understanding of recov- ticity is limited, with most patients reaching a plateau ery mechanisms is limited. Nevertheless, current evi- after recovering approximately 70–80% of the initial im- dence suggests that recovery requires active physical pairment (stroke: [28–30]). Based on these studies it has participation of patients during therapy [40]. Addition- been suggested that most of the observed recovery is ally, intensity (number of repetitions per unit of time) spontaneous, without evidence for significant training ef- and dose (duration) of physical therapy are also thought fects on upper limb function. The recovery of neuro- to have a positive effect on outcome in both animal [22] logical deficits is similar in young and elderly subjects, and human [41–43] studies. These reports were chal- but the transfer into activities of daily living is reduced lenged by a study showing no intensity effect and min- in the elderly [31]. As recovery is incomplete, compensa- imal gains in chronic post-stroke subjects [44]. This tory movement strategies are also an important con- finding might be explained by the relatively low overall tributor to the mitigation of motor deficits [32], e.g. by dose, ranging from 13.6 h to 26.3 h on the mean, enabling mobility through technical aids such as a whereas by applying a very high dose of 300 h, clinically wheelchair. meaningful gains were described [45]. This suggests that The recovery of function in persons with CNS lesion is the doses provided in the standard of care are not suffi- much like a relearning process exploiting preserved sen- ciently high, with implications for the further application sorimotor circuits [33]. The relearning can be optimized of rehabilitation robots in the clinic and at home. Also, by providing appropriate proprioceptive input to the intensive task-specific multi-joint functional training spinal cord with the goal of maximally engaging preserved does not necessarily improve performance in ADL [46] neural circuits. The extent of recovery depends on the se- nor is it superior to single joint robotic training [47]. verity of CNS damage and the individual neural capacity Nevertheless, there is some evidence for a transfer of of a patient to regain a function. Cognition and motivation task-specific training effects to untrained tasks [48]. are important contributors to this relearning, especially Most of the factors that influence rehabilitation out- for the upper limbs [34], and must therefore be considered come are based on evidence from experiments in post- during rehabilitation. ‘Normal’ movement performance stroke patients as they represent a much larger patient can only rarely be restored after a stroke or SCI. There- group than patients with SCI. However, findings made in fore, the goal of rehabilitation is not primarily to re- stroke concerning lower limb, i.e. stepping function, are establish ‘normal’ movement patterns, but to enable usually also valid in SCI and can be transferred to this ‘simpler’, less well-organized movements to achieve opti- patient population. For example, the positive effect of mal outcome in mobility and independence during activ- training intensity on the outcome of ambulation in ities of daily living (ADL) for the individual patient [35]. stroke subjects [41] could recently be confirmed for sub- There are basic differences in the recovery of upper jects with SCI [49]. For the upper extremity, hand func- and lower limb function. For instance, the exploitation tion in SCI subjects is determined by the lesion level of neuroplasticity is quite limited for arm and hand and the combined damage of central and peripheral ner- movements after a stroke, especially when the corticosp- val structures after a cervical injury [24]. In contrast, in inal tract is damaged. In addition, there are differences post-stroke patients it greatly depends on the integrity of between cerebral and spinal cord damage. For example, the corticospinal tract. the success of rehabilitation depends on the integrity of cognitive function, which is often impaired in post- General implications for robot-assisted therapy stroke subjects. The general neurophysiological considerations provide Spasticity can contribute to the compensation of sen- a strong basis for the application of robots in re- sorimotor deficits [36–38], thereby assisting in the res- habilitation. Robot-assisted rehabilitation provides a toration of function. Spastic muscle tone can be used to standardized environment, in which both therapy in- partially compensate for the loss of limb activation in tensity and dose can be increased. In a conventional mobile patients. Consequently, movement generation setting, hemiparetic patients typically perform about takes place on a lower level of organization in the ab- 30 movement repetitions with their affected upper sence of cortical drive, i.e. spastic legs can provide body limb in a 45-min session [50], whereas robot-assisted Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 5 of 15 therapy has achieved over 1000 repetitions per session on the functional tasks to be trained, the targeted patient [18]. Active physical and cognitive engagement of pa- population and the severity of sensorimotor deficits. tients during therapy are crucial for recovery. This can be promoted through adaptive assistance [51], in Rehabilitation of arm/hand function a way to avoid slacking of the patient [52], as well as The versatility and complexity of arm and hand through cognitive challenge [53], automated task diffi- movements with unique functions such as unimanual culty adaptation [54, 55] and motivating feedback reaching, grasping and manipulation, as well as bi- [34]. Feedback about movement performance can not manual separate and cooperative movements, differ only enhance motivation but also facilitate plasticity fundamentally from stepping movements with a more in the motor cortex if it arrives synchronously with automatic movement control. Skilled hand and finger motor output [56]. movements reflect cultural achievements in the evolu- Severely affected patients can benefit from passive/ tion [59] that are associated with a specific cortico- highly-assisted movements and gravity support by motoneuronal control [60], i.e. direct projections from exoskeletons that provide control over all relevant the cortex to motoneurons in the spinal cord which joints (Figs. 2 and 3). In order to minimally interfere innervate arm/hand muscles. As a result, arm, and with and alter functional movements in less affected especially distal hand function are often severely patients [16], nor influence automated assessments impaired following CNS damage, greatly limiting basedonintegratedsensing [57], rehabilitation robots patients in their ability to perform ADL [61]. The should have low inherent impedance [58], or require severity of impairment and, consequently, any recov- the ability to adapt output impedance through control ery of function is related to the extent of damage of [16]. This requires a careful selection of kinematic struc- the corticospinal system [62, 63]. Functional training ture, actuation/transmission and integrated sensing based approaches and, consequently, devices supporting Fig. 2 Upper panel: Evolution of upper extremity rehabilitation robots. From stiff (high impedance) industrial manipulators to dedicated rehabilitation robots providing control at the distal effector or over each joint, including the rendering of virtual object dynamics resulting in somatosensory feedback. Further evolution of the technology will see wearable systems providing support not only during therapy sessions, but also during activities of daily living in the home environment, allowing physical interaction with real objects. Lower panel: Task-specific design of hand rehabilitation robots. Functional hand movement training should focus not only on unimanual, i.e. reach and grasp tasks (left), but should also include bimanual separate tasks (middle), as well as cooperative movement tasks that are employed, e.g., when opening a bottle (right) Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 6 of 15 Fig. 3 Evolution of lower extremity rehabilitation robots. Since their introduction, rehabilitation robots for the lower extremity have evolved from stiff industrial robot arms to guide the limb passively, without cognitive or physical involvement of the patient, to systems allowing for active engagement of patients through adapted support and body weight unloading in a vertical posture. Currently, wearable exoskeletons are being introduced into clinical practice, promoting even more active engagement of the patient, while balance is provided by crutches. Future exoskeletons will support balance to the degree needed. The three systems to the right are inspired by neurophysiological insights, stimulating afferent receptors through, e.g., weight loading, ground contact and assisted hip extension to trigger leg flexion movements. From left to right, patients require increasing functional abilities, while the robotic systems provide less support. Most patients will benefit from several of these systems (from leftto right) during different phases of recovery unilateral arm and hand movements [64, 65] should flexor hypertonia and extensor weakness that contrib- thus be directed towards the abilities patients require for utes to the inability to perform finger extension and ADL, i.e., most importantly unimanual and bimanual hand opening movements [60]. These patients also suf- reach and grasp tasks [66]. Furthermore compensatory fer from difficulties in the grasping and manipulation of approaches and assistive devices have to be considered for objects, while some proximal arm function is usually more severely impaired patients. preserved. Most reports show that in patients with dam- age to the CST, even with intensive rehabilitation mea- Neurophysiological factors influencing the recovery of sures, little recovery [28, 30], particularly of hand and upper limb function finger function [70], can be expected. In general, the recovery of arm/hand function following In contrast, the recovery in patients with an intact CNS damage is limited when compared to gait in post- CST is proportional to the initial impairment, with pa- stroke [41] and cervical spinal cord injured [67] subjects, tients recovering approximately 70–80% of the initial even if intensive therapy is applied. In patients with a impairment (proportional recovery rule) [28–30]. Some cervical SCI, arm function depends on the level of the studies indicate that training effects in these patients are lesion. An injury level at C5/C6 results in a ‘passive’ small or absent [46], i.e. only a minor dose-response ef- hand function (supination movement at the elbow joint fects occur [44]. However, there is also evidence that a for hand opening) or, frequently, at C6/7, in a tenodesis higher dose of practice, especially when applied early grasp. This grasp is defined as a hand function when after a stroke, leads to a better outcome of motor func- some forearm extensor muscle activation is preserved tion of the paretic arm [41, 43, 71]. [68]. It allows to close the hand by wrist extension Early after stroke flaccid arm muscle paresis prevails, i. movements with the fingers in a slightly flexed contrac- e. the limbs are weak and do not resist passive displace- ture position. Some spastic muscle tone is required to ment. With the development of some spastic muscle perform such simple grasp movements [24]. tone, needed to perform rudimentary grips, the training In post-stroke subjects, outcome of upper limb func- of residual muscle function can be initiated [24]. In this tion critically depends on the integrity of the corticosp- stage, the focus of therapy/training should be directed to inal tract (CST) [63, 69]. A stroke with damage to the enable the execution of simple reach and grasp move- CST results in lasting impairment of hand and finger ments. In the weeks following stroke, spastic muscle function and an unbalanced muscle tone with forearm tone usually becomes more pronounced in the forearm Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 7 of 15 flexor than in the extensor muscles, as the antigravity contralateral hemispheres become involved in the con- muscles have more muscle mass [39, 72]. This can again trol of each of the two hands during cooperative hand impair the execution of functional reach and grasp movements [84]. Consequently, in post-stroke patients movements. However, some spastic muscle tone in the during the training of cooperative hand tasks, the un- forearm muscles allows the performance of a tenodesis affected hemisphere supports movements of the paretic grasp, which is important for the execution of ADL, not hand and arm [85]. However, the effect of a cooperative only in SCI but also in post-stroke subjects. training on the outcome of hand function remains to be Patients typically compensate for their sensorimotor defi- determined. cits through the involvement of the non-paretic arm/hand, Finally, while the recovery of finger function is limited, leading to learned non-use of the paretic arm [64, 73]. basic functions such as opening and closing the hand Therefore, one important approach to rehabilitate hand should also be trained, as most of the interaction with function after stroke was presented in the form of the environment during ADL involves grasping and re- constraint-induced movement therapy (CIMT). This was leasing objects. Besides motor function, somatosensory based on the idea of enhancing recovery of function by re- function is also of importance during object grasping: ducing interhemispheric inhibition of the stroke hemi- shaping and maintaining a stable grasp during the ma- sphere [74]. By immobilizing the non-affected hand the nipulation of an object relies on the processing of som- patient is forced to use the paretic hand/arm for the per- atosensory input, determined by the mechanical formance of ADL [64]. However so far, a superior effect of properties of the manipulated object [86]. Somatosen- CIMT compared to other therapy approaches was not sory function is often impaired after CNS damage, lead- reported [75]. ing to a visual compensation of movement control. During the course of upper limb rehabilitation, the However, in some patients it can recover spontaneously support provided should always be kept to a minimum or through dedicated training [87]. in order to make the training challenging with a max- imum of individual effort and contribution to movement Implications for robot-assisted therapy of upper limb performance by the patient (for review [24]). However, function the optimal level of assistance also depends on the sever- The combination of kinematic complexity and functional ity of impairment [70]. Most stroke patients will benefit impairment makes the design of robotic devices to train from gravity support, allowing them to perform func- arm, hand and finger function after CNS damage par- tional movements by their own effort [76]. Without such ticularly challenging. Following the initial developments support, shoulder abduction, which is important for ob- based on stiff industrial manipulators, end-effector- ject manipulation, may limit elbow extension and result based devices for planar (MIT-MANUS; [88]) and 3D in concurrent elbow, wrist and finger flexion, i.e. so- (Gentle/S, [89]) reaching movements were introduced to called flexion synergies after stroke [77]. This can affect allow more active contribution of the patient while limit- the execution of functional hand movements. ing the apparent impedance of the robot. Subsequent Many upper limb movements involve the use of both developments focused on incorporating additional de- hands. However, only a few studies provide a neurophysio- grees of freedom (DOF) related to wrist [90] and hand logical basis for the training of bimanual movements [78]. opening/closing function (Gentle/G [91]). For the func- Bimanual training of reaching and grasping tasks in stroke tional training of three-dimensional arm movements patients has been suggested to be more effective in im- with guidance at the three proximal joints, ARMin, a proving unilateral execution of these tasks with the af- grounded, powered exoskeleton was developed, which fected arm than unilateral training alone [79]. This might also integrates grasp and release function [92, 93] (Fig. 2, be a result of stronger recruitment of the contralesional upper panel). hemisphere through bilateral compared to unilateral train- Independent of their kinematic configuration, all of ing [80]. However, there is currently no clear evidence that these systems can partially or fully unload the arm bimanual training is superior to CIMT [65, 81, 82], or un- against gravity. This approach reduces the effect of constrained unimanual training [83]. flexor synergies, and allows the performance of hand The involvement of the unaffected hemisphere in movements within a larger workspace. However, the movement control of the paretic hand might be even complex structure and geared actuators of such devices stronger in a special type of bimanual movement, where with their reflected inertia limit the interaction quality one hand supports the action of the other one by gener- and the ability to adapt the level of support [16]. The ating equal but opposed forces/torques, e.g. when open- large output impedance may render the active initiation ing a bottle or cutting bread. Such cooperative hand of movements more difficult, and potentially alter nat- movements are based on a task-specific control: a ural movement dynamics. Therefore, a trade-off between ‘neural coupling’ of the hemispheres, i.e. both ipsi- and the number of DOF and the quality of the physical Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 8 of 15 interaction exists, limiting the application of these de- cognitive nature of these tasks and motivate patients. In vices to specific stages of recovery. For example, training a meta-analysis, the application of virtual reality (VR) with a powered whole-arm exoskeleton is mainly indi- games was found to be potentially useful for the im- cated for stroke subjects with severe arm paresis early provement of arm function after stroke [34]. after the incident. Similar effects can also be achieved by In conclusion, a good, mainly spontaneous, recovery of using passive devices for gravity support to the upper upper limb function after a stroke can be expected when limb, to enable self-initiated movements [94]. the integrity of the CST is preserved. There is some evi- Robot-assisted approaches should also consider the dence that higher dose of practice leads to improved training of bimanual and cooperative movement tasks as function, especially early after stroke. Nevertheless, in they are important during ADL (Fig. 2, lower panel). Bi- cases with damaged CST the recovery is limited and nei- manual training was a focus of some early studies [10], ther depends on the approach nor on the dose of train- but its potential has not been sufficiently explored and de- ing. Unimanual robot-assisted therapy approaches serves further investigation. Many upper extremity sys- should be complemented by bimanual (cooperative) ap- tems developed and clinically evaluated so far could also proaches. These should also incorporate the training of be used for bimanual training, by combining two devices basic hand function and interaction with virtual object in a mirrored configuration. The training of cooperative dynamics that generate somatosensory feedback. In the hand movements (e.g. opening a bottle) has been pro- future, it will be possible to at least partially compensate posed using a dedicated device [84], and can also be for remaining deficits with wearable assistive robotics. achieved by virtually coupling two unimanual devices through control. Rehabilitation of locomotor function Due to the biomechanical and neural complexity of Locomotor movements are performed more automatic- hand and finger movements, robot-assisted rehabilitation ally than arm and hand movements. Corticospinal con- of hand and finger function became a focus only recently. trol mainly serves the goal to voluntarily alter the Most rehabilitation robots for hand function have been stepping rhythm, e.g. to correct the stepping direction or based on end-effector designs, used either independently amplitude to overcome obstacles. Accordingly, corti- or in combination with grounded exoskeletons or end- cospinal projections to lower limb motoneurons in effector type arm devices (Fig. 2). Several groups have also humans are stronger to the flexor than to the extensor made attempts to develop exoskeleton systems for the muscles [99, 100]. The rehabilitation of locomotor func- hand, some of which assist independent finger motion, tion is simpler than that of upper extremity function, generally resulting in highly complex devices that under- and basic mobility can usually be restored in post-stroke went none to little clinical evaluation. A review [19]found subjects by using the paretic limb as a stick to support that only 25% of 30 hand rehabilitation robots had been the body [24]. clinically tested, and many devices had been considered Passive orthoses can assist foot dorsiflexion in the too complex for clinical use. However, such complexity swing phase of stepping. In SCI subjects, some proximal might not be necessary when the focus is directed to the leg muscle activation is required for a successful loco- basic function of opening and closing the hand [95]. This motor training [101]. Besides this, the rehabilitation of might be sufficient given the limited potential for the re- locomotor function in post-stroke and SCI subjects is covery of finger function following CNS damage, while similar. In severely affected subjects, mobility can be re- remaining highly relevant for ADL. Finally, hand opening/ stored with a wheelchair or other mobility aids. Never- closing can also be supported through wearable assistive theless, the primary goal of rehabilitation is to restore technology, such as soft robotic gloves [96, 97], which sufficient lower-limb function for patients to ambulate could be worn during the performance of ADL. without walking aids. Interaction with the environment occurs mainly through the hands and generates somatosensory feed- Neurophysiological factors influencing the recovery of back. However, somatosensory function is often im- locomotor function paired after CNS damage. Therefore, neurorehabilitation Thirty years ago, rehabilitation after CNS damage was fo- devices for the upper extremity should train hand and, cused on the strengthening of leg muscles to a level where as far as possible, finger function, providing both visual patients were able to perform stepping movements on and haptic feedback [53]. Training should include tasks parallel bars with the support of their arms [102]. In the which are functionally relevant for ADL, such as grasp- early nineties, functional locomotor training with body ing and releasing objects with rendered virtual dynamics unloading of para−/tetraparetic SCI subjects was intro- to also train somatosensory function and sensorimotor duced. This was based on the observation that locomotor integration [98]. Finally, most upper limb training de- function in cat SCI models recovers quite well during vices are embedded in computer games to reflect the treadmill training with body-weight unloading (body Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 9 of 15 weight supported treadmill training, BWSTT) [103]. In in- activation [110]. This is in line with cat experiments, completely paralyzed SCI patients, BWSTT has been where flexor bursts were automatically generated at the shown to result in a similar outcome of stepping function end hip extension despite complete SCI [112, 113]. In compared to a conventional rehabilitation approach [104]. robot-assisted gait training systems, leg flexion move- In post-stroke subjects no gain in outcome was found dur- ments are usually imposed by the robot, leaving the sub- ing BWSTT compared to an unspecific physical exercise ject passive. Over time, this leads to a rarefaction program [105]. The BWSTT training of SCI subjects is [114]and dysfunction[115]ofleg flexor motoneu- physically demanding and requires two physiotherapists rons, i.e. the peripheral nervous system, deprived of who assist leg movements from both sides. As a result, supraspinal input, undergoes degenerative changes. In training time is limited to about half an hour per day, even completely paralyzed patients with an SCI who do though many patients would tolerate more therapy. Yet, not undergo a functional locomotor training, spinal such a dose increase has been associated with a better neuronal circuits underlying stepping movements be- outcome [41]. come silent even when appropriate proprioceptive in- Movement speed during locomotor training represents put is provided. On a longer term this results in a another factor that influences outcome. In ambulatory neuronal dysfunction below the level of the lesion in stroke patients, a successive increase (according to both rodents [116]and patients with SCI [117]. principles of sports physiology) of treadmill speed after a Today we know that bipeds use a quadrupedal coord- 4-week training period resulted in a better walking abil- ination during locomotion, i.e. arm movements repre- ity than conventional gait training [106]. Furthermore, sent an integral part of locomotion [118, 119] and, locomotor training was shown to be most efficient when therefore, might be included in locomotor training pro- delivered in a real-world environment [107]. grams. In fact, recent experiments indicate that arm In severely paralyzed patients with an SCI, automatic movements induce an increase in leg muscle activity stepping movements can be induced, associated with a during stepping [120]. physiological leg muscle activation (i.e. close-to normal timing of electromyography (EMG) patterns with re- duced amplitude), when patients stand on a moving Implications for robot-assisted therapy of lower limb treadmill with the body unloaded up to 80% [108, 109]. function This leg muscle activation is triggered by load receptor Functional gait training positively affects the recovery of input from contact forces during the stance phase of gait locomotor function, but is personnel-intensive and phys- [110]. Such a physiological limb muscle activation was ically demanding for the therapists. This situation trig- found to be the prerequisite for positive training effects gered the development of robotic devices to assist leg and improvement of locomotor function in rodents [22] movements during stepping on a treadmill with the body and patients with a stroke or SCI (for review [24]). With partially unloaded [13]. Robot-assisted BWSTT has been the onset of voluntary control of some proximal leg shown to be as effective as overground stepping with the muscles, body unloading can be reduced and self- support of physiotherapists [121]. Together with the fact induced stepping movements become possible. This is that training intensity has a positive effect on the recov- associated with an increase in strength of leg muscle ac- ery of locomotor ability in post-stroke [41] and SCI [49] tivation. Thus, during the course of training, body un subjects, this motivates the use of robot-assisted −/reloading has to be adapted to the actual degree of BWSTT, allowing longer training times (and thus higher paresis. dose) with less personnel. Furthermore, this approach Most of the recovery of function occurs during the provides a standardized training environment and allows first three months after CNS damage. However, also in an objective assessment of the changes achieved during chronic patients with an incomplete SCI and stroke a the course of rehabilitation [57]. A systematic review significant gain in gait velocity, endurance, and perform- that examined the effect of electromechanical and robot- ance can be achieved by an automated locomotor train- assisted gait training in post-stroke subjects showed that ing [102]. Further improvement of locomotor function patients receiving such training are more likely to after damage to the CNS is associated with only minor achieve independent walking than people who received changes in the leg muscle activity pattern, and relies gait training without such devices [122]. more on a better coordination between the legs and an During the course of rehabilitation, the physical support adapted spastic muscle tone (stroke: [111]; SCI: [37]). has to be continuously adapted to the actual needs of the Hip extension at the end of the stance phase is an es- patient, with the objective of maximizing active participa- sential stimulus for the leg muscle activation during tion of the patient by reducing and selectively providing locomotion, especially for initiating the stance to swing assistance [24]. With the recovery of locomotor function transition with an appropriate change in leg muscle (especially of proximal joints), a transition from a Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 10 of 15 grounded exoskeleton to a grounded end-effector device continue therapy after hospital rehabilitation, but also can take place (e.g. [123]). This can be followed by the to compensate for remaining deficits by providing ap- execution of stepping movements on normal ground with propriate assistance in ADL. reduced support from a wearable robotic exoskeleton or In conclusion, in post-stroke patients training leads to mobility aid. a good recovery of stepping function using spastic Initial developments for robot-assisted gait training muscle tone for body support. In SCI, some remaining were focused on patients with almost complete paralysis proximal leg muscle function is required to allow a suc- as a result of SCI, where training is most demanding for cessful training and recovery of function. The activation therapists. These patients can hardly actively contribute of load (re-loading of the body as far as possible) and to the leg movements. Therefore, high assistive torques hip-joint related (hip extension) receptors leads to a are required, typically resulting in robotic devices with physiological leg muscle activity pattern during stepping high output impedance. With increased penetration of and, consequently, to dose-dependent training effects. this technology into clinics, locomotor training using Accordingly, devices are required which can adapt the such devices was expanded to patients with a stroke or support and impedance of individual joints according to incomplete SCI, who require less and/or asymmetric as- patients’ impairment. The development of wearable ro- sistance. However, this is difficult to achieve by devices botic gait orthoses with integrated balance support will with a high output impedance, as these behave more like further promote functional training, engagement and a velocity than a torque source. Consequently, novel motivation, and lead to systems that can provide long- control approaches [124] and devices with low intrinsic term assistance in the home environment (Fig. 3). impedance [125] were developed to better adapt the physical support to these patient populations. These ef- Conclusions forts need to be continued, also to assure that automated Rehabilitation training of the upper and lower limb assessments reflect the current impairment level of a pa- should be founded on neurophysiological insights, inde- tient, and are not masked by the device dynamics [57]. pendent of whether it is performed conventionally, or The field of lower-extremity robot-assisted therapy is pro- with the support of robotic devices. After CNS damage, gressing towards wearable powered exoskeletons (Fig. 3). improvement of sensorimotor functions occurs to a large These combine the advantages of grounded devices with degree spontaneously and can further be achieved by an the ability to train in a real-world environment and provide exploitation of neuroplasticity. This is reflected in a higher levels of subject participation and challenge [58]. physiological limb muscle activation that might serve as Even more than in the case of grounded devices, it is a a marker for the achievement of training effects. This re- challenge to achieve low output impedance together with quires voluntarily performed upper limb movements, or the provision of sufficient assistive torque in wearable exo- an activation of appropriate receptors for a purposeful skeletons, as all links and joints are integrated and thus car- activation of lower limb muscles, i.e. during stepping ried by the moving exoskeleton. This results in weight and movements. complexity constraints that limit both the number of DOF The potential for a recovery of function differs not that can be actuated and the transparency (ability of the only between upper and lower limbs, but also between system to getout of theway)thatcan be achieved. The neurological disorders such as stroke and SCI, and re- high reduction ratios required to generate sufficient torque quires the development of technology accounting for increase the output impedance of the device, thus limiting these differences. Table 1 summarizes the main aspects the capacity to adapt the support to changing abilities of of neurorehabilitation and outcome in these disorders, the patient. With technological progress, it might become as well as the implications for rehabilitation technology. possible to modulate the output impedance of each joint. Training effects in upper compared to lower limbs are Through this approach, hip extension might be enforced to limited and are mainly determined by corticospinal tract trigger physiological leg flexion movements for the initi- integrity. Nevertheless, intensive, highly dosed training ation of the swing phase (cf. [110]), which the exoskeleton has beneficial effects, especially early after stroke. Training could passively follow. Such an approach would allow a bet- devices should unload (and gradually re-load) arm move- ter adaptation of the support to the individual patient, en- ments against gravity, incorporate hand function for reach able more dynamic motion, and prevent degenerative and grasp training and use the motivating and cognitively changes in the peripheral nervous system. engaging effects of virtual reality, with a focus on the ADL With the ability to partially support balance with that are most important to the individual. wearable exoskeletons, the hands become free from For the lower limbs, the effects of training on the holding crutches. This will facilitate arm swing, which recovery of sensorimotor function seem to depend on both represents an integral part of locomotion (for review their intensity and dose. Rehabilitation robots are ideal tools see [118]). These devices will not only allow to to complement classical functional therapy by allowing a Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 11 of 15 Table 1 Main aspects of neurorehabilitation and outcome, and their implications for rehabilitation technology Limb Condition Typical recovery course Goal Rehabilitation approach Technology UL stroke damaged CST little recovery, esp. prox. arm muscle prox. arm muscle therapy: passive chronic impairment of activation; avoidance of strengthening; mobilization (position hand/finger extension muscle contractures; use continuous passive limb control) or weight of impaired limb for motion; training of support for self- support/holding compensatory strategies initiated proximal function movements; active/passive hand module with extension bias assistance: supported arm/hand motion (admittance control) vial intention detection (e.g. force, EMG, gaze) intact CST spontaneous recovery of arm reaching and functional reach/grasp therapy: proximal gravity approx. 70–80% of intial simple grasping and bimanual support during reach/ arm/hand impairment function; uni−/bimanual (cooperative) hand grasp; training of ADL functions movements; individual joints using strengthening of wrist/ dedicated devices, finger extensors; simple including hand/fingers, movement training with as well as (cooperative) transfer to ADL; limited bimanual training (Fig. 2) dose-dependent training assistance: passive effects: subacute proximal gravity support > chronic stage combined with active wrist/finger support via residual function amplification (force/EMG control) SCI (incomplete) typical lesion spastic forearm flexor tenodesis grasp; assistance: active level C6/7 muscle tone impeding bimanual grasp exoskeleton/glove to the development of facilitate wrist and finger tenodesis grasp flexion/extension triggered by proximal arm motion (e.g. joint angle sensor) LL stroke hemiparesis spontaneous recovery; non-assisted ambulation generation of therapy: body-weight spastic muscle support; appropriate afferent support according to reduced level of input from load (body paresis; adapted stepping movement un/reloading) and hip movement support ability receptors (hip extension) (position/admittance during stepping; control for severe SCI (incomplete) paraparesis some prox. leg muscle assisted/independent importance of stepping impairment and variable function and spastic ambulation velocity and hip impedance control for muscle tone required for extension (initiation of mild/moderate stepping ability swing); dose-dependent impairment; Fig. 3); leg training effects flexor activation through robotically assisted hip extension UL upper limb, LL lower limb, SCI spinal cord injury, CST corticospinal tract, ADL activities of daily living, EMG electromyography standardized and intensive training with individually and to adapt their output impedance and physical support to continuously adapted physical support. Grounded exoskele- the actual state of the patient and the task at hand, with- tons and end-effector devices as well as wearable exoskele- out altering functional movement patterns through their tons seem to be equally effective in the improvement of apparent dynamics. Patients will likely train with differ- function. However, their suitability depends on the phase of ent devices throughout the recovery phase during re- recovery, and the individual impairment (Fig. 3). habilitation, to optimally adapt movement complexity Rehabilitation robots should always provide targeted and physical support to the current state and functional physical support adapted to the functional abilities of abilities of the patient (e.g. transitioning from left to the patient in a way to enable functional movements. right in Fig. 3). To deal with this challenge, the design of This has strong implications for the design, instrumenta- future robotic rehabilitation systems should also con- tion and control of such systems. These should be able sider the relevance of particular joints during functional Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 12 of 15 movement (e.g. the hip joint plays a larger role in loco- Ethics approval and consent to participate Not applicable. motion than the knee and ankle joints; [110]) and their potential for recovery (e.g. limited recovery of individual Competing interests finger movement). Volker Dietz is member of the Scientific Advisory Board of Hocoma AG. There are currently a number of novel and exciting developments in and around the fields of rehabilitation Publisher’sNote engineering and rehabilitation sciences. Advances in ma- Springer Nature remains neutral with regard to jurisdictional claims in terial sciences will allow lighter, more customizable published maps and institutional affiliations. structures with more tightly integrated actuation and Author details sensing. Furthermore, there is an increasing focus on 1 Department of Health Sciences and Technology, ETH Zurich, 8092 Zurich, combining robotics with non-invasive [126, 127] and Switzerland. Spinal Cord Injury Center, Balgrist University Hospital, 8008 Zurich, Switzerland. invasive [128] brain-machine interfaces or neuropros- thetics, with the aim of promoting independence during Received: 20 January 2018 Accepted: 7 May 2018 activities of daily living. These approaches are at an early stage and still face a number of challenges. Nevertheless, References even an optimal exploitation of neuroplasticity (cf. [24]) 1. Reinkensmeyer D, Dietz V. Neurorehabilitation Technology. 2nd ed. Springer, will not result in a full functional recovery. The field Cham; 2016. should therefore focus on wearable systems that not only 2. Khalili D, Zomlefer M. An intelligent robotic system for rehabilitation of joints and estimation of body segment parameters. IEEE Trans Biomed Eng. support functional therapy, but that can also serve as as- 1988;35(2):138–46. sistive devices to compensate for persisting sensorimotor 3. Rabishong P, Bel J, Hill J, Peruchon E, Simeon M, Screve J, et al. The AMOLL deficits. Advanced actuation, sensing and control ap- project (active modular orthosis for lower limbs). In: Proc Int Symp external control hum extremities, vol. 1975; 1975. p. 33–42. proaches will make these systems more robust and applic- 4. Seireg A, Grundmann J. Design of a multitask exoskeletal walking device for able for ADL tasks, and applicable both in the clinic and at paraplegics. Biomechanics of medical devices. 1981:569–644. home. In the future, it can be expected that wearable de- 5. Vukobratovic M, Hristic D, Stojiljkovic Z. Development of active anthropomorphic exoskeletons. Med Biol Eng. 1974;12(1):66–80. vices continuously adapt and reduce support until recovery 6. Ali H. Bionic exoskeleton: history, development and the future. In: plateaus, and then compensate for chronic impairments. International conference on advances in Engineering & Technology: IOSR A number of further challenges remain for the field of Journal of Mechanical and Civil Engineering (IOSR-JMCE); 2014;58–62. 7. Rabischong P. Pneumatic Exoskeleton Prosthesis http://cyberneticzoocom/ robot-assisted therapy and assistance, many of which will bionics/1976-pneumatic-exoskeleton-prosthesis-pierre-rabischong-french/. also require collaboration with industry and government bodies. Adoption of robotic devices is driven by cost and 8. Dijkers MP, de Bear PC, Erlandson RF, Kristy K, Geer DM, Nichols A. Patient and staff acceptance of robotic technology in occupational therapy: a pilot reimbursement systems, and research should therefore study. J Rehabil Res Dev. 1991;28(2):33–44. also focus on identifying the simplest, most effective tech- 9. Krebs HI, Hogan N, Aisen ML, Volpe BT. Robot-aided neurorehabilitation. nical solutions that can support the rehabilitation process. IEEE Trans Rehabil Eng. 1998;6(1):75–87. 10. Lum PS, Reinkensmeyer DJ, Lehman SL. Robotic assist devices for bimanual Finally, future rehabilitation approaches will not only physical therapy: preliminary experiments. IEEE Trans Rehabil Eng. 1993;1(3): profit from the inclusion of robots, but also from an ad- 185–91. vanced understanding of neurophysiological mechanisms 11. Lum PS, Lehman SL, Reinkensmeyer DJ. The bimanual lifting rehabilitator: an adaptive machine for therapy of stroke patients. IEEE Trans Rehabil Eng. underlying normal and impaired sensorimotor functions, 1995;3(2):166–74. enabled by the use of robots as scientific tools [129]. 12. Burgar CG, Lum PS, Shor PC, Machiel Van der Loos HF. Development of Resulting insights will benefit the development of ad- robots for rehabilitation therapy: the Palo alto VA/Stanford experience. J Rehabil Res Dev. 2000;37(6):663–73. vanced rehabilitation robots, and further promote col- 13. Colombo G, Joerg M, Schreier R, Dietz V. Treadmill training of paraplegic laboration between engineers, therapists and clinical patients using a robotic orthosis. J Rehabil Res Dev. 2000;37(6):693–700. neurophysiologists. 14. Hesse S, Uhlenbrock D. A mechanized gait trainer for restoration of gait. J Rehabil Res Dev. 2000;37(6):701–8. Acknowledgments 15. Just F, Baur K, Klamroth-Marganska V, Riener R, Rauter G. Motor inertia We thank Stefan Schneller for preparing the graphics included in this compensation of the ARMin rehabilitation robot. In: AUTOMED workshop manuscript, Raffaele Ranzani for his valuable and critical feedback on the 2016: Hochschule Wismar; 2016. manuscript, and Monica Stüssi for formatting the references. 16. Metzger JC, Lambercy O, Gassert, R. Performance comparison of interaction control strategies on a hand rehabilitation robot. IEEE International Funding Conference on Rehabilitation Robotics (ICORR). 2015;846–851. https://doi. The chair in Rehabilitation Engineering (RG) is supported by the ETH Zurich org/10.1109/ICORR.2015.7281308. Foundation in collaboration with Hocoma AG. 17. Klamroth-Marganska V, Blanco J, Campen K, Curt A, Dietz V, Ettlin T, et al. Three-dimensional, task-specific robot therapy of the arm after stroke: a Authors’ contributions multicentre, parallel-group randomised trial. Lancet Neurol. 2014;13(2):159–66. Both authors contributed equally to this manuscript. Each author initially 18. Lo AC, Guarino PD, Richards LG, Haselkorn JK, Wittenberg GF, Federman DG, focused on the parts related to his respective background (RG: rehabilitation et al. Robot-assisted therapy for long-term upper-limb impairment after robotics, VD: neurophysiological aspects of healthy and impaired functional stroke. N Engl J Med. 2010;362:1772–83. movement), which were then merged through numerous discussions and 19. Balasubramanian S, Klein J, Burdet E. Robot-assisted rehabilitation of hand iterations. Both authors read and approved the final manuscript. function. Curr Opin Neurol. 2010;23(6):661–70. Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 13 of 15 20. Kwakkel G, Kollen BJ, Krebs HI. Effects of robot-assisted therapy on upper 47. Milot MH, Spencer SJ, Chan V, Allington JP, Klein J, Chou C, et al. A limb recovery after stroke: a systematic review. Neurorehabil Neural Repair. crossover pilot study evaluating the functional outcomes of two different 2008;22(2):111–21. types of robotic movement training in chronic stroke survivors using the 21. Tefertiller C, Pharo B, Evans N, Winchester P. Efficacy of rehabilitation arm exoskeleton BONES. J Neuroeng Rehabil. 2013;10:112. robotics for walking training in neurological disorders: a review. J Rehabil 48. Schaefer SY, Patterson CB, Lang CE. Transfer of training between distinct Res Dev. 2011;48(4):387–416. motor tasks after stroke: implications for task-specific approaches to 22. Edgerton VR, Tillakaratne NJ, Bigbee AJ, de Leon RD, Roy RR. Plasticity of the upper-extremity neurorehabilitation. Neurorehabil Neural Repair. spinal neural circuitry after injury. Annu Rev Neurosci. 2004;27:145–67. 2013;27(7):602–12. 49. Wirz M, Mach O, Maier D, Benito-Penalva J, Taylor J, Esclarin A, et al. 23. Nudo RJ, Plautz EJ, Frost SB. Role of adaptive plasticity in recovery of function after damage to motor cortex. Muscle Nerve. 2001;24(8):1000–19. Effectiveness of automated locomotor training in patients with acute incomplete spinal cord injury: a randomized controlled multicenter trial. 24. Dietz V, Fouad K. Restoration of sensorimotor functions after spinal cord J Neurotrauma. 2017;34:1891–6. injury. Brain. 2014;137(Pt 3):654–67. 25. Zeiler SR, Krakauer JW. The interaction between training and plasticity in the 50. Lang CE, Macdonald JR, Reisman DS, Boyd L, Jacobson Kimberley T, poststroke brain. Curr Opin Neurol. 2013;26(6):609–16. Schindler-Ivens SM, et al. Observation of amounts of movement practice 26. Zarahn E, Alon L, Ryan SL, Lazar RM, Vry MS, Weiller C, et al. Prediction of provided during stroke rehabilitation. Arch Phys Med Rehabil. 2009;90(10): motor recovery using initial impairment and fMRI 48 h poststroke. Cereb 1692–8. Cortex. 2011;21(12):2712–21. 51. Riener R, Lunenburger L, Colombo G. Human-centered robotics applied to gait training and assessment. J Rehabil Res Dev. 2006;43(5):679–94. 27. Curt A, Van Hedel HJ, Klaus D, Dietz V. Recovery from a spinal cord injury: 52. Marchal-Crespo L, McHughen S, Cramer SC, Reinkensmeyer DJ. The effect of significance of compensation, neural plasticity, and repair. J Neurotrauma. haptic guidance, aging, and initial skill level on motor learning of a steering 2008;25(6):677–85. task. Exp Brain Res. 2010;201(2):209–20. 28. Prabhakaran S, Zarahn E, Riley C, Speizer A, Chong JY, Lazar RM, et al. Inter- individual variability in the capacity for motor recovery after ischemic stroke. 53. Metzger JC, Lambercy O, Califfi A, Conti FM, Gassert R. Neurocognitive Neurorehabil Neural Repair. 2008;22:64–71. robot-assisted therapy of hand function. IEEE Trans Haptics. 2014;7(2):140–9. 29. Kwakkel G, Kollen B, Lindeman E. Understanding the pattern of functional 54. Metzger JC, Lambercy O, Califfi A, Dinacci D, Petrillo C, Rossi P, et al. recovery after stroke: facts and theories. Restor Neurol Neurosci. 2004;22(3–5): Assessment-driven selection and adaptation of exercise difficulty in 281–99. robot-assisted therapy: a pilot study with a hand rehabilitation robot. 30. Winters C, van Wegen EE, Daffertshofer A, Kwakkel G. Generalizability of the J Neuroeng Rehabil. 2014;11:154. proportional recovery model for the upper extremity after an ischemic 55. Zimmerli L, Krewer C, Gassert R, Muller F, Riener R, Lunenburger L. stroke. Neurorehabil Neural Repair. 2015;29(7):614–22. Validation of a mechanism to balance exercise difficulty in robot-assisted 31. Jakob W, Wirz M, van Hedel HJ, Dietz V. Difficulty of elderly SCI subjects to upper-extremity rehabilitation after stroke. J Neuroeng Rehabil. 2012;9:6. 56. Stefan K, Kunesch E, Cohen LG, Benecke R, Classen J. Induction of plasticity in the translate motor recovery—“body function”–into daily living activities. J human motor cortex by paired associative stimulation. Brain. 2000;123:572–84. Neurotrauma. 2009;26(11):2037–44. 32. Huang VS, Krakauer JW. Robotic neurorehabilitation: a computational motor 57. Lambercy O, Maggioni S, Lünenburger L, Gassert R, Bolliger M. Robotic and learning perspective. J Neuroeng Rehabil. 2009;6:5. wearable sensor Technologies for Measurements/Clinical Assessments. In: 33. Krakauer JW. Motor learning: its relevance to stroke recovery and Reinkensmeyer D, Dietz V, editors. Neurorehabilitation Technology. Springer: neurorehabilitation. Curr Opin Neurol. 2006;19(1):84–90. Cham; 2016. 34. Saposnik G, Levin M, Group ORCW. Virtual reality in stroke rehabilitation: a 58. Pennycott A, Wyss D, Vallery H, Klamroth-Marganska V, Riener R. Towards meta-analysis and implications for clinicians. Stroke. 2011;42(5):1380–6. more effective robotic gait training for stroke rehabilitation: a review. 35. Latash M, Anson J. What are “normal movements” in atypic populations? J Neuroeng Rehabil. 2012;9:65. Behav Brain Sci. 1996;19:55–106. 59. Herder JG. Ideen zur Philiosophie der Geschichte der Menschheit, Bd. 1 edn. Leipzig.Hartknoch; 1785. 36. O'Dwyer NJ, Ada L, Neilson PD. Spasticity and muscle contracture following 60. Lemon RN. Descending pathways in motor control. Annu Rev Neurosci. stroke. Brain. 1996;119:1737–49. 2008;31:195–218. 37. Dietz V, Sinkjaer T. Secondary changes after CNS damage: the significance of spastic muscle tone in rehabilitation. In: Dietz V, Ward N, editors. Oxford 61. Jorgensen HS, Reith J, Nakayama H, Kammersgaard LP, Raaschou HO, Olsen textbook of neurorehabilitation. Oxford: Oxford University Press; 2015. TS. What determines good recovery in patients with the most severe 38. Katz RT, Rymer WZ. Spastic hypertonia: mechanisms and measurement. strokes? The Copenhagen Stroke Study Stroke. 1999;30(10):2008–12. Arch Phys Med Rehabil. 1989;70(2):144–55. 62. Ward NS, Newton JM, Swayne OB, Lee L, Thompson AJ, Greenwood RJ, 39. Dietz V, Sinkjaer T. Spastic movement disorder: impaired reflex function and et al. Motor system activation after subcortical stroke depends on altered muscle mechanics. Lancet Neurol. 2007;6(8):725–33. corticospinal system integrity. Brain. 2006;129(Pt 3):809–19. 63. Byblow WD, Stinear CM, Barber PA, Petoe MA, Ackerley SJ. Proportional 40. Ada L, Dorsch S, Canning CG. Strengthening interventions increase strength recovery after stroke depends on corticomotor integrity. Ann Neurol. 2015; and improve activity after stroke: a systematic review. Aust J Physiother. 78(6):848–59. 2006;52(4):241–8. 64. Taub E, Uswatte G, Pidikiti R. Constraint-induced movement therapy: a new 41. Kwakkel G, Wagenaar RC, Twisk JW, Lankhorst GJ, Koetsier JC. Intensity of leg family of techniques with broad application to physical rehabilitation–a and arm training after primary middle-cerebral-artery stroke: a randomised trial. clinical review. J Rehabil Res Dev. 1999;36(3):237–51. Lancet. 1999;354(9174):191–6. 42. Lohse KR, Lang CE, Boyd LA. Is more better? Using metadata to explore 65. Stoykov ME, Lewis GN, Corcos DM. Comparison of bilateral and unilateral dose-response relationships in stroke rehabilitation. Stroke. 2014;45(7): training for upper extremity hemiparesis in stroke. Neurorehabil Neural 2053–8. Repair. 2009;23:945–53. 43. Wu X, Guarino P, Lo AC, Peduzzi P, Wininger M. Long-term effectiveness of 66. Kilbreath SL, Heard RC. Frequency of hand use in healthy older persons. intensive therapy in chronic stroke. Neurorehabil Neural Repair. 2016;30(6):583–90. Aust J Physiother. 2005;51(2):119–22. 44. Lang CE, Strube MJ, Bland MD, Waddell KJ, Cherry-Allen KM, Nudo RJ, et al. 67. Zariffa J, Kapadia N, Kramer JL, Taylor P, Alizadeh-Meghrazi M, Zivanovic V, et al. Feasibility and efficacy of upper limb robotic rehabilitation in a subacute Dose response of task-specific upper limb training in people at least 6 cervical spinal cord injury population. Spinal Cord. 2012;50(3):220–6. months poststroke: a phase II, single-blind, randomized, controlled trial. Ann Neurol. 2016;80(3):342–54. 68. Dietz V, Curt A. Neurological aspects of spinal-cord repair: promises and challenges. Lancet Neurol. 2006;5(8):688–94. 45. McCabe J, Monkiewicz M, Holcomb J, Pundik S, Daly JJ. Comparison of robotics, functional electrical stimulation, and motor learning methods for 69. Stinear CM, Byblow WD, Ackerley SJ, Smith MC, Borges VM, Barber PA. treatment of persistent upper extremity dysfunction after stroke: a Proportional motor recovery after stroke: implications for trial design. Stroke. randomized controlled trial. Arch Phys Med Rehabil. 2015;96(6):981–90. 2017;48(3):795–8. 46. Waddell KJ, Strube MJ, Bailey RR, Klaesner JW, Birkenmeier RL, Dromerick 70. Rowe JB, Chan V, Ingemanson ML, Cramer SC, Wolbrecht ET, Reinkensmeyer AW, et al. Does task-specific training improve upper limb performance in DJ. Robotic assistance for training finger movement using a Hebbian model: a daily life Poststroke? Neurorehabil Neural Repair. 2017;31(3):290–300. randomized controlled trial. Neurorehabil Neural Repair. 2017;31(8):769–80. Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 14 of 15 71. Veerbeek JM, van Wegen E, van Peppen R, van der Wees PJ, Hendriks E, 96. In H, Kang BB, Sin M, Cho KJ. Exo glove: a wearable robot for the hand with Rietberg M, et al. What is the evidence for physical therapy poststroke? A a soft tendon routing system. IEEE Robotics & Automation Magazine 2015; systematic review and meta-analysis. PLoS One. 2014;9(2):e87987. 1(Mar 22):97-105. 72. Marciniak C, Rader L, Gagnon C. The use of botulinum toxin for spasticity after 97. Arata J, Ohmoto K, Gassert R, Lambercy O, Fujimoto H, Wada I. A new hand spinal cord injury. Am J Phys Med Rehabil. 2008;87(4):312–7. quiz 318-320, 329 exoskeleton device for rehabilitation using a three-layered sliding spring 73. Taub E, Crago JE, Burgio LD, Groomes TE, Cook EW 3rd, DeLuca SC, et al. An mechanism. IEEE International Conference on Robotics and Automation. operant approach to rehabilitation medicine: overcoming learned nonuse 2013;3902–3907. https://doi.org/10.1109/ICRA.2013.6631126. by shaping. J Exp Anal Behav. 1994;61(2):281–93. 98. Metzger JC, Lambercy O, Gassert R. High-fidelity rendering of virtual objects with the ReHapticKnob - Novel avenues in robot-assisted rehabilitation of 74. Di Pino G, Pellegrino G, Assenza G, Capone F, Ferreri F, Formica D, et al. hand function. IEEE Haptics Symposium. 2012;51–6. https://doi.org/10.1109/ Modulation of brain plasticity in stroke: a novel model for HAPTIC.2012.6183769. neurorehabilitation. Nat Rev Neurol. 2014;10(10):597–608. 75. Kwakkel G, Veerbeek JM, van Wegen EE, Wolf SL. Constraint-induced 99. Brouwer B, Ashby P. Corticospinal projections to lower limb motoneurons in movement therapy after stroke. Lancet Neurol. 2015;14(2):224–34. man. Exp Brain Res. 1992;89(3):649–54. 76. Beer RF, Ellis MD, Holubar BG, Dewald JP. Impact of gravity loading on post-stroke 100. Schubert M, Curt A, Jensen L, Dietz V. Corticospinal input in human gait: reaching and its relationship to weakness. Muscle Nerve. 2007;36(2):242–50. modulation of magnetically evoked motor responses. Exp Brain Res. 1997; 77. Lan Y, Yao J, Dewald J. Increased shoulder abduction loads decreases 115(2):234–46. volitional finger extension in individuals with chronic stroke: preliminary 101. Wirz M, van Hedel HJ, Rupp R, Curt A, Dietz V. Muscle force and gait findings. Annual International Conference of the IEEE Engineering in performance: relationships after spinal cord injury. Arch Phys Med Rehabil. Medicine and Biology Society (EMBC). 2014;5808–5811. 2006;87(9):1218–22. 102. Wirz M, Zemon DH, Rupp R, Scheel A, Colombo G, Dietz V, et al. Effectiveness 78. McCombe Waller S, Whitall J. Bilateral arm training: why and who benefits? of automated locomotor training in patients with chronic incomplete spinal NeuroRehabilitation. 2008;23(1):29–41. cord injury: a multicenter trial. Arch Phys Med Rehabil. 2005;86(4):672–80. 79. Mudie MH, Matyas TA. Can simultaneous bilateral movement involve the undamaged hemisphere in reconstruction of neural networks damaged by 103. Barbeau H, Wainberg M, Finch L. Description and application of a system stroke? Disabil Rehabil. 2000;22(1–2):23–37. for locomotor rehabilitation. Med Biol Eng Comput. 1987;25(3):341–4. 80. Luft AR, McCombe-Waller S, Whitall J, Forrester LW, Macko R, Sorkin JD, et 104. Dobkin B, Barbeau H, Deforge D, Ditunno J, Elashoff R, Apple D, et al. The al. Repetitive bilateral arm training and motor cortex activation in chronic evolution of walking-related outcomes over the first 12 weeks of stroke: a randomized controlled trial. JAMA. 2004;292(15):1853–61. rehabilitation for incomplete traumatic spinal cord injury: the multicenter 81. Lin KC, Chen YA, Chen CL, Wu CY, Chang YF. The effects of bilateral arm randomized spinal cord injury locomotor trial. Neurorehabil Neural Repair. training on motor control and functional performance in chronic stroke: a 2007;21(1):25–35. randomized controlled study. Neurorehabil Neural Repair. 2010;24(1):42–51. 105. Duncan PW, Sullivan KJ, Behrman AL, Azen SP, Wu SS, Nadeau SE, et al. Body-weight-supported treadmill rehabilitation after stroke. N Engl J Med. 82. vanDelden AL, Beek PJ,RoerdinkM,KwakkelG,Peper CL.Unilateral 2011;364(21):2026–36. and bilateral upper-limb training interventions after stroke have similar 106. Pohl M, Mehrholz J, Ritschel C, Ruckriem S. Speed-dependent treadmill effects on bimanual coupling strength. Neurorehabil Neural Repair. training in ambulatory hemiparetic stroke patients: a randomized controlled 2015;29(3):255–67. trial. Stroke. 2002;33(2):553–8. 83. Whitall J, McCombe Waller S, Silver KH, Macko RF. Repetitive bilateral arm training with rhythmic auditory cueing improves motor function in chronic 107. Sandler EB, Roach KE, Field-Fote EC. Dose-response outcomes associated hemiparetic stroke. Stroke. 2000;31(10):2390–5. with different forms of locomotor training in persons with chronic motor- 84. Dietz V, Macauda G, Schrafl-Altermatt M, Wirz M, Kloter E, Michels L. Neural incomplete spinal cord injury. J Neurotrauma. 2017;34(10):1903–8. coupling of cooperative hand movements: a reflex and FMRI study. Cereb 108. Dietz V, Colombo G, Jensen L. Locomotor activity in spinal man. Lancet. Cortex. 2015;25(4):948–58. 1994;344(8932):1260–3. 85. Schrafl-Altermatt M, Dietz V. Cooperative hand movements in stroke 109. Dietz V, Colombo G, Jensen L, Baumgartner L. Locomotor capacity of spinal patients: neural reorganization. Clin Neurophysiol. 2016;127(1):748–54. cord in paraplegic patients. Ann Neurol. 1995;37(5):574–82. 110. Dietz V, Muller R, Colombo G. Locomotor activity in spinal man: significance of 86. Johansson RS. How is grasping modifed by somatosensory input. Motor afferent input from joint and load receptors. Brain. 2002;125(Pt 12):2626–34. Control: Consepts and Issues. 1991;14:331–5. 87. Carey L, Macdonell R, Matyas TA. SENSe: study of the effectiveness of 111. Den Otter AR, Geurts AC, Mulder T, Duysens J. Gait recovery is not neurorehabilitation on sensation: a randomized controlled trial. associated with changes in the temporal patterning of muscle activity Neurorehabil Neural Repair. 2011;25(4):304–13. during treadmill walking in patients with post-stroke hemiparesis. Clin 88. Aisen ML, Krebs HI, Hogan N, McDowell F, Volpe BT. The effect of robot- Neurophysiol. 2006;117(1):4–15. assisted therapy and rehabilitative training on motor recovery following 112. Grillner S, Rossignol S. On the initiation of the swing phase of locomotion stroke. Arch Neurol. 1997;54(4):443–6. in chronic spinal cats. Brain Res. 1978;146(2):269–77. 89. Loureiro R, Amirabdollahian F, Topping M, Driessen B, Harwin W. Upper 113. Kriellaars DJ, Brownstone RM, Noga BR, Jordan LM. Mechanical entrainment of limb robot mediated stroke therapy - GENTLE/s approch. Auton Robot. fictive locomotion in the decerebrate cat. J Neurophysiol. 1994;71(6):2074–86. 2003;15(1):35–51. 114. Lin CS, Macefield VG, Elam M, Wallin BG, Engel S, Kiernan MC. Axonal changes in spinal cord injured patients distal to the site of injury. Brain. 90. Masia L, Krebs HI, Cappa P, Hogan N. Design and characterization of hand 2007;130(Pt 4):985–94. module for whole-arm rehabilitation following stroke. IEEE ASME Trans Mechatron. 2007;12(4):399–407. 115. Dietz V, Muller R. Degradation of neuronal function following a spinal 91. Loureiro RC, Lamperd B, Collin C, Harwin WS. Reach & grasp therapy: Effects cord injury: mechanisms and countermeasures. Brain. 2004;127(Pt 10): of the Gentle/G System assessing sub-acute stroke whole-arm rehabilitation. 2221–31. IEEE International Conference on Rehabilitation Robotics. 2009;755–760. 116. Beauparlant J, van den Brand R, Barraud Q, Friedli L, Musienko P, Dietz V, 92. Nef T, Klamroth-Marganska V, Keller U, Riener R. Three-dimensional multi- et al. Undirected compensatory plasticity contributes to neuronal degree-of-freedom arm therapy robot (ARMin). In Neurorehabilitation dysfunction after severe spinal cord injury. Brain. 2013;136(Pt 11):3347–61. Technology. Springer: Cham. 2016;351–374. 117. Dietz V, Grillner S, Trepp A, Hubli M, Bolliger M. Changes in spinal reflex and 93. Nef T, Riener R. Three-dimensional-multi-degree of freedom arm therapy locomotor activity after a complete spinal cord injury: a common robot (ARMin). In: Dietz V, Nef T, Rymer WZ, editors. Neurorehabilitation mechanism? Brain. 2009;132(Pt 8):2196–205. technology. London: Springer; 2012. 118. Dietz V. Do human bipeds use quadrupedal coordination? Trends Neurosci. 2002;25(9):462–7. 94. Housman SJ, Scott KM, Reinkensmeyer DJ. A randomized controlled trial of gravity-supported, computer-enhanced arm exercise for individuals with 119. Pauvert V, Pierrot-Deseilligny E, Rothwell JC. Role of spinal premotoneurones in severe hemiparesis. Neurorehabil Neural Repair. 2009;23(5):505–14. mediating corticospinal input to forearm motoneurones in man. J Physiol. 95. Lambercy O, Dovat L, Yun H, Wee SK, Kuah CW, Chua KS, et al. Effects of a 1998;508:301–12. robot-assisted training of grasp and pronation/supination in chronic stroke: 120. de Kam D, Rijken H, Manintveld T, Nienhuis B, Dietz V, Duysens J. Arm a pilot study. J Neuroeng Rehabil. 2011;8:63. movements can increase leg muscle activity during submaximal recumbent Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 15 of 15 stepping in neurologically intact individuals. J Appl Physiol (1985). 2013; 115(1):34–42. 121. Dobkin B, Apple D, Barbeau H, Basso M, Behrman A, Deforge D, et al. Weight-supported treadmill vs over-ground training for walking after acute incomplete SCI. Neurology. 2006;66(4):484–93. 122. Mehrholz J, Thomas S, Werner C, Kugler J, Pohl M, Elsner B. Electromechanical-assisted training for walking after stroke. Cochrane database Syst Rev. 2017;5:CD006185. 123. Hesse S, Schattat N, Mehrholz J, Werner C. Evidence of end-effector based gait machines in gait rehabilitation after CNS lesion. NeuroRehabilitation. 2013;33(1):77–84. 124. Duschau-Wicke A, von Zitzewitz J, Caprez A, Lünenburger L, Riener R. Path control: a method for patient-cooperative robot-aided gait rehabilitation. IEEE Trans Neural Syst Rehabil Eng. 2010;18:38–48. 125. Veneman JF, Kruidhof R, Hekman EE, Ekkelenkamp R, Van Asseldonk EH, van der Kooij H. Design and evaluation of the LOPES exoskeleton robot for interactive gait rehabilitation. IEEE Trans Neural Syst Rehabil Eng. 2007;15(3):379–86. 126. Buch E, Weber C, Cohen LG, Braun C, Dimyan MA, Ard T, et al. Think to move: a neuromagnetic brain-computer interface (BCI) system for chronic stroke. Stroke. 2008;39(3):910–7. 127. Carda S, Biasiucci A, Maesani A, Ionta S, Moncharmont J, Clarke S, et al. Electrically assisted movement therapy in chronic stroke patients with severe upper limb paresis: a pilot, single-blind, randomized crossover study. Arch Phys Med Rehabil. 2017;98(8):1628–35. e1622 128. Hochberg LR, Bacher D, Jarosiewicz B, Masse NY, Simeral JD, Vogel J, et al. Reach and grasp by people with tetraplegia using a neurally controlled robotic arm. Nature. 2012;485(7398):372–5. 129. Wolpert DM, Flanagan JR. Q&A: robotics as a tool to understand the brain. BMC Biol. 2010;8:92. 130. Loureiro RC, Harwin WS, Nagai K, Johnson M. Advances in upper limb stroke rehabilitation: a technology push. Med Biol Eng Comput. 2011;49(10):1103–18. 131. Maciejasz P, Eschweiler J, Gerlach-Hahn K, Jansen-Troy A, Leonhardt S. A survey on robotic devices for upper limb rehabilitation. J Neuroeng Rehabil. 2014;11:3. 132. Sheng B, Zhang Y, Meng W, Deng C, Xie S. Bilateral robots for upper-limb stroke rehabilitation: state of the art and future prospects. Med Eng Phys. 2016;38(7):587–606. 133. Lum PS, Godfrey SB, Brokaw EB, Holley RJ, Nichols D. Robotic approaches for rehabilitation of hand function after stroke. Am J Phys Med Rehabil. 2012;91(11 Suppl 3):S242–54. 134. Bos RA, Haarman CJ, Stortelder T, Nizamis K, Herder JL, Stienen AH, et al. A structured overview of trends and technologies used in dynamic hand orthoses. J Neuroeng Rehabil. 2016;13(1):62. 135. Mehrholz J, Pohl M, Platz T, Kugler J, Elsner B. Electromechanical and robot- assisted arm training for improving activities of daily living, arm function, and arm muscle strength after stroke. Cochrane Database of Systematic Reviews 2015, Issue 11. Art. No.: CD006876. https://doi.org/10.1002/14651858.CD006876.pub4. 136. Veerbeek JM, Langbroek-Amersfoort AC, van Wegen EE, Meskers CG, Kwakkel G. Effects of robot-assisted therapy for the upper limb after stroke. Neurorehabil Neural Repair. 2017;31(2):107–21. 137. Diaz I, Gil JJ, Sanchez E. Lower-limb robotic rehabilitation: literature review and challenges. J Robot. 2011;2011:1-11. https://doi.org/10.1155/2011/759764. 138. Benito-Penalva J, Edwards DJ, Opisso E, Cortes M, Lopez-Blazquez R, Murillo N, et al. Gait training in human spinal cord injury using electromechanical systems: effect of device type and patient characteristics. Arch Phys Med Rehabil. 2012;93(3):404–12. 139. Nam KY, Kim HJ, Kwon BS, Park JW, Lee HJ, Yoo A. Robot-assisted gait training (Lokomat) improves walking function and activity in people with spinal cord injury: a systematic review. J Neuroeng Rehabil. 2017;14(1):24. 140. Louie DR, Eng JJ. Powered robotic exoskeletons in post-stroke rehabilitation of gait: a scoping review. J Neuroeng Rehabil. 2016;13(1):53. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Journal of NeuroEngineering and Rehabilitation Springer Journals

Rehabilitation robots for the treatment of sensorimotor deficits: a neurophysiological perspective

Free
15 pages
Loading next page...
 
/lp/springer_journal/rehabilitation-robots-for-the-treatment-of-sensorimotor-deficits-a-iX8ZTx8w9E
Publisher
BioMed Central
Copyright
Copyright © 2018 by The Author(s).
Subject
Biomedicine; Neurosciences; Neurology; Rehabilitation Medicine; Biomedical Engineering
eISSN
1743-0003
D.O.I.
10.1186/s12984-018-0383-x
Publisher site
See Article on Publisher Site

Abstract

The past decades have seen rapid and vast developments of robots for the rehabilitation of sensorimotor deficits after damage to the central nervous system (CNS). Many of these innovations were technology-driven, limiting their clinical application and impact. Yet, rehabilitation robots should be designed on the basis of neurophysiological insights underlying normal and impaired sensorimotor functions, which requires interdisciplinary collaboration and background knowledge. Recovery of sensorimotor function after CNS damage is based on the exploitation of neuroplasticity, with a focus on the rehabilitation of movements needed for self-independence. This requires a physiological limb muscle activation that can be achieved through functional arm/hand and leg movement exercises and the activation of appropriate peripheral receptors. Such considerations have already led to the development of innovative rehabilitation robots with advanced interaction control schemes and the use of integrated sensors to continuously monitor and adapt the support to the actual state of patients, but many challenges remain. For a positive impact on outcome of function, rehabilitation approaches should be based on neurophysiological and clinical insights, keeping in mind that recovery of function is limited. Consequently, the design of rehabilitation robots requires a combination of specialized engineering and neurophysiological knowledge. When appropriately applied, robot- assisted therapy can provide a number of advantages over conventional approaches, including a standardized training environment, adaptable support and the ability to increase therapy intensity and dose, while reducing the physical burden on therapists. Rehabilitation robots are thus an ideal means to complement conventional therapy in the clinic, and bear great potential for continued therapy and assistance at home using simpler devices. This review summarizes the evolution of the field of rehabilitation robotics, as well as the current state of clinical evidence. It highlights fundamental neurophysiological factors influencing the recovery of sensorimotor function after a stroke or spinal cord injury, and discusses their implications for the development of effective rehabilitation robots. It thus provides insights on essential neurophysiological mechanisms to be considered for a successful development and clinical inclusion of robots in rehabilitation. Keywords: Robot-assisted therapy, Neurorehabilitation technology, Assist-as-needed, Stroke, Spinal cord injury, Locomotion, Upper limb function, Sensorimotor neurophysiology, Neuroplasticity * Correspondence: roger.gassert@hest.ethz.ch Department of Health Sciences and Technology, ETH Zurich, 8092 Zurich, Switzerland Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 2 of 15 Background from passive movements for the most severely impaired Rehabilitation robotics is a relatively young and rapidly patients, to active-assisted and active-resisted move- growing field, with increasing penetration into the clin- ments in moderately impaired patients. Furthermore, as- ical environment [1]. In the late 1980s and early 90s a sistance could be automatically adapted to the patient’s number of pioneering technological developments were performance. Around the same time, the Mirror Image launched, triggered by discoveries on training-induced Motion Enabler (MIME; [12]) was introduced, which recovery of sensorimotor function in animal models with supported paretic limb movements with a stiff industrial damage to the central nervous system (CNS). The goal robot, controlled by the non-paretic limb by means of a was to enhance the effects of functional training by pro- motion digitizer (mirror-image therapy mode). viding increased therapy intensity and adaptive support Developments of rehabilitation robots for the lower in a controlled way. extremity began in 1994, with the design of the Lokomat The idea of using machines for rehabilitation dates [13], combining body-weight supported treadmill- back much earlier. In a 1910 patent, Theodor Büdingen training (BWSTT) with the assistance of a robotic gait proposed a ‘movement cure apparatus’, a machine driven orthosis. The Gait Trainer [14] realized a similar concept by an electric motor to guide and support stepping based on an end-effector design. movements in patients with heart disease. In the 1930s, The decades since these pioneering developments have Richard Scherb developed the ‘meridian’, a cable-driven seen an explosion of novel rehabilitation robots for both apparatus to move joints for orthopaedic therapy. This the upper and lower extremities, which can broadly be human-powered mechanotherapy machine already sup- classified into grounded exoskeletons, grounded end- ported multiple interaction modes, ranging from passive effector devices, and wearable exoskeletons (Fig. 1). to active-assisted and active-resisted movements. A first These design approaches affect the level of control over robotic rehabilitation system was based on the concept the interaction (control of individual joints in exoskel- of continuous passive motion (CPM), a stiff interaction eton devices vs control over selected joints or limb seg- mode in which the robot moves the joints along a prede- ments in grounded end-effector devices) as well as the fined trajectory, independent of the contribution of the output impedance of the device (resulting from the patient [2]. mechanical structure as well as actuator and transmis- The first powered exoskeletons for therapeutic applica- sion properties) and the ability to modulate this imped- tions in SCI patients were introduced in the 1970s [3–5]. ance through control. Grounded end-effector devices These systems used pneumatic, hydraulic, or electromag- will typically achieve higher motion dynamics and allow netic (via cams and Bowden cables) actuators for position the rendering of a wider range of impedances than exo- servocontrol. They included advanced features, such as skeleton devices with a serial kinematic structure, where actuated ankle flexion/extension, and hip adduction/ab- proximal joints need to move distal joints. The latter re- duction for increased stability [6] or the ability of a therap- quires large reduction ratios and results in high inertia ist to control the motion of the exoskeleton worn by the and friction at the output where the patient is attached patient through his/her own movement (in a similar, con- [15, 16]. These dynamics can only partially be compen- nected exoskeleton) [7]. The first system for robot- sated through control. assisted therapy of stroke survivors [8] was based on a stiff The number of new developments has been dispropor- industrial manipulator and did not physically interact with tionate to the penetration of these technologies into the patients, but rather moved a pad that patients had to clinical setting, likely due to the technology-driven ap- touch to different locations. proach of many engineering groups and the limited, al- A new era of neurorehabilitation robotics began in beit increasing, exchange of the field with therapists and 1989 with the development of the MIT-MANUS [9], clinicians. While a few randomized-controlled trials have which was first tested clinically in 1994. Compared to in- confirmed efficacy of robot-assisted therapy equivalent dustrial manipulators, this planar manipulandum pre- to that of dose-matched conventional therapy [17–21], sents inherently low mechanical output impedance (a the majority of published devices were never clinically frequency-dependent resistance to motion perceived at evaluated, or such an evaluation was limited to pilot the interface between the human user and the robotic studies on a few patients. Interestingly, many of these system) and provides unloading of the upper limb studies unsuccessfully aimed to demonstrate superiority against gravity, thereby allowing to adapt support to the of robot-assisted as compared to conventional therapy, severity of the deficits. A few years later, force controlled despite the fact that there is currently no consensus on devices for bimanual, cooperative grasping [10] and lift- the optimal therapy program for an individual patient in ing [11] were introduced. This new generation of de- the clinical field. vices, using torque-controlled direct drive actuation, For a successful inclusion of robots in rehabilitation, allowed for more advanced interaction control, ranging fundamental knowledge about the physiological basis of Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 3 of 15 Fig. 1 Schematic representation and classification of rehabilitation robots. Besides the extremity that is trained, rehabilitation robots can be broadly classified into grounded exoskeletons, end-effector devices and wearable exoskeletons. While the first two are well established, the latter are currently entering clinical application [17–21, 95, 122, 130–140] the recovery of function is required. This knowledge is for upper and lower limb functions that should be con- widely distributed and difficult for engineers to access sidered for the design of effective rehabilitation robots, and translate into design considerations. Consequently, and underlines the need for transdisciplinary collabora- in our opinion and experience, a close cooperation be- tions for future developments. Before addressing aspects tween engineers, therapists and clinical neurophysiolo- specific to upper and lower limb rehabilitation, general gists/neurorehabilitation scientists is required from the neurophysiological considerations of relevance for the very beginning of a development, and was shown to be design of rehabilitation robots will be discussed. successful in previous developments (e.g. of the Lokomat with the involvement of VD, a neurologist/clinical Neurophysiological basis for the recovery of neurophysiologist [13]). sensorimotor function after CNS damage According to evidence from studies in cats [22], non- Stroke and spinal cord injury (SCI) are among the lead- human primates [23], and humans [24], recovery of sen- ing causes of adult long-term physical disability, with ap- sorimotor function after damage to the central nervous proximately 10 million people surviving a stroke and system (CNS) is based on the exploitation of neuroplas- over 250′000 surviving a spinal cord injury (of which ap- ticity. It relies on physiological limb activation during proximately 60% are incomplete) every year. Muscle the training of functional arm and hand movements, and weakness due to activation deficits represents the main the stimulation of appropriate peripheral receptors dur- disability following stroke and SCI, and frequently limits ing automatically performed leg movements such as self-independence. Furthermore, following CNS damage, stepping. Rehabilitation robots should therefore enable secondary effects such as spastic muscle tone (increased and support such functional training. resistance to passive stretch) develop. This review aims to provide historical and clinical The aim of neurorehabilitation is to improve outcome background of relevance to the field of rehabilitation ro- of function after damage to the CNS, such as stroke and botics for engineers, basic and clinical neurophysiolo- SCI, through intensive physical therapy. This goal is, gists and therapists interested in and entering this however, difficult to define as the effects of conventional exciting field. It introduces the neurophysiological basis therapy can hardly be separated from the spontaneous Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 4 of 15 recovery of function that occurs in parallel to the effects support during stance and gait in a stick-like manner of the rehabilitative treatment [25]. In stroke [26] and [39]. However, this only holds for moderately affected, SCI [27], most of the spontaneous recovery occurs mobile patients, while in severely disabled patients, spas- within the first three months. tic signs such as muscle cramps may become exagger- Therapy-induced recovery is mediated by neuroplasti- ated, requiring pharmaceutical interventions. city, and the goal of rehabilitation is thus to maximally There is currently no consensus on the optimal ther- exploit neuroplasticity in order to achieve an optimal apy programs to promote recovery of motor function outcome for the individual patient. However, neuroplas- following CNS damage, and the understanding of recov- ticity is limited, with most patients reaching a plateau ery mechanisms is limited. Nevertheless, current evi- after recovering approximately 70–80% of the initial im- dence suggests that recovery requires active physical pairment (stroke: [28–30]). Based on these studies it has participation of patients during therapy [40]. Addition- been suggested that most of the observed recovery is ally, intensity (number of repetitions per unit of time) spontaneous, without evidence for significant training ef- and dose (duration) of physical therapy are also thought fects on upper limb function. The recovery of neuro- to have a positive effect on outcome in both animal [22] logical deficits is similar in young and elderly subjects, and human [41–43] studies. These reports were chal- but the transfer into activities of daily living is reduced lenged by a study showing no intensity effect and min- in the elderly [31]. As recovery is incomplete, compensa- imal gains in chronic post-stroke subjects [44]. This tory movement strategies are also an important con- finding might be explained by the relatively low overall tributor to the mitigation of motor deficits [32], e.g. by dose, ranging from 13.6 h to 26.3 h on the mean, enabling mobility through technical aids such as a whereas by applying a very high dose of 300 h, clinically wheelchair. meaningful gains were described [45]. This suggests that The recovery of function in persons with CNS lesion is the doses provided in the standard of care are not suffi- much like a relearning process exploiting preserved sen- ciently high, with implications for the further application sorimotor circuits [33]. The relearning can be optimized of rehabilitation robots in the clinic and at home. Also, by providing appropriate proprioceptive input to the intensive task-specific multi-joint functional training spinal cord with the goal of maximally engaging preserved does not necessarily improve performance in ADL [46] neural circuits. The extent of recovery depends on the se- nor is it superior to single joint robotic training [47]. verity of CNS damage and the individual neural capacity Nevertheless, there is some evidence for a transfer of of a patient to regain a function. Cognition and motivation task-specific training effects to untrained tasks [48]. are important contributors to this relearning, especially Most of the factors that influence rehabilitation out- for the upper limbs [34], and must therefore be considered come are based on evidence from experiments in post- during rehabilitation. ‘Normal’ movement performance stroke patients as they represent a much larger patient can only rarely be restored after a stroke or SCI. There- group than patients with SCI. However, findings made in fore, the goal of rehabilitation is not primarily to re- stroke concerning lower limb, i.e. stepping function, are establish ‘normal’ movement patterns, but to enable usually also valid in SCI and can be transferred to this ‘simpler’, less well-organized movements to achieve opti- patient population. For example, the positive effect of mal outcome in mobility and independence during activ- training intensity on the outcome of ambulation in ities of daily living (ADL) for the individual patient [35]. stroke subjects [41] could recently be confirmed for sub- There are basic differences in the recovery of upper jects with SCI [49]. For the upper extremity, hand func- and lower limb function. For instance, the exploitation tion in SCI subjects is determined by the lesion level of neuroplasticity is quite limited for arm and hand and the combined damage of central and peripheral ner- movements after a stroke, especially when the corticosp- val structures after a cervical injury [24]. In contrast, in inal tract is damaged. In addition, there are differences post-stroke patients it greatly depends on the integrity of between cerebral and spinal cord damage. For example, the corticospinal tract. the success of rehabilitation depends on the integrity of cognitive function, which is often impaired in post- General implications for robot-assisted therapy stroke subjects. The general neurophysiological considerations provide Spasticity can contribute to the compensation of sen- a strong basis for the application of robots in re- sorimotor deficits [36–38], thereby assisting in the res- habilitation. Robot-assisted rehabilitation provides a toration of function. Spastic muscle tone can be used to standardized environment, in which both therapy in- partially compensate for the loss of limb activation in tensity and dose can be increased. In a conventional mobile patients. Consequently, movement generation setting, hemiparetic patients typically perform about takes place on a lower level of organization in the ab- 30 movement repetitions with their affected upper sence of cortical drive, i.e. spastic legs can provide body limb in a 45-min session [50], whereas robot-assisted Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 5 of 15 therapy has achieved over 1000 repetitions per session on the functional tasks to be trained, the targeted patient [18]. Active physical and cognitive engagement of pa- population and the severity of sensorimotor deficits. tients during therapy are crucial for recovery. This can be promoted through adaptive assistance [51], in Rehabilitation of arm/hand function a way to avoid slacking of the patient [52], as well as The versatility and complexity of arm and hand through cognitive challenge [53], automated task diffi- movements with unique functions such as unimanual culty adaptation [54, 55] and motivating feedback reaching, grasping and manipulation, as well as bi- [34]. Feedback about movement performance can not manual separate and cooperative movements, differ only enhance motivation but also facilitate plasticity fundamentally from stepping movements with a more in the motor cortex if it arrives synchronously with automatic movement control. Skilled hand and finger motor output [56]. movements reflect cultural achievements in the evolu- Severely affected patients can benefit from passive/ tion [59] that are associated with a specific cortico- highly-assisted movements and gravity support by motoneuronal control [60], i.e. direct projections from exoskeletons that provide control over all relevant the cortex to motoneurons in the spinal cord which joints (Figs. 2 and 3). In order to minimally interfere innervate arm/hand muscles. As a result, arm, and with and alter functional movements in less affected especially distal hand function are often severely patients [16], nor influence automated assessments impaired following CNS damage, greatly limiting basedonintegratedsensing [57], rehabilitation robots patients in their ability to perform ADL [61]. The should have low inherent impedance [58], or require severity of impairment and, consequently, any recov- the ability to adapt output impedance through control ery of function is related to the extent of damage of [16]. This requires a careful selection of kinematic struc- the corticospinal system [62, 63]. Functional training ture, actuation/transmission and integrated sensing based approaches and, consequently, devices supporting Fig. 2 Upper panel: Evolution of upper extremity rehabilitation robots. From stiff (high impedance) industrial manipulators to dedicated rehabilitation robots providing control at the distal effector or over each joint, including the rendering of virtual object dynamics resulting in somatosensory feedback. Further evolution of the technology will see wearable systems providing support not only during therapy sessions, but also during activities of daily living in the home environment, allowing physical interaction with real objects. Lower panel: Task-specific design of hand rehabilitation robots. Functional hand movement training should focus not only on unimanual, i.e. reach and grasp tasks (left), but should also include bimanual separate tasks (middle), as well as cooperative movement tasks that are employed, e.g., when opening a bottle (right) Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 6 of 15 Fig. 3 Evolution of lower extremity rehabilitation robots. Since their introduction, rehabilitation robots for the lower extremity have evolved from stiff industrial robot arms to guide the limb passively, without cognitive or physical involvement of the patient, to systems allowing for active engagement of patients through adapted support and body weight unloading in a vertical posture. Currently, wearable exoskeletons are being introduced into clinical practice, promoting even more active engagement of the patient, while balance is provided by crutches. Future exoskeletons will support balance to the degree needed. The three systems to the right are inspired by neurophysiological insights, stimulating afferent receptors through, e.g., weight loading, ground contact and assisted hip extension to trigger leg flexion movements. From left to right, patients require increasing functional abilities, while the robotic systems provide less support. Most patients will benefit from several of these systems (from leftto right) during different phases of recovery unilateral arm and hand movements [64, 65] should flexor hypertonia and extensor weakness that contrib- thus be directed towards the abilities patients require for utes to the inability to perform finger extension and ADL, i.e., most importantly unimanual and bimanual hand opening movements [60]. These patients also suf- reach and grasp tasks [66]. Furthermore compensatory fer from difficulties in the grasping and manipulation of approaches and assistive devices have to be considered for objects, while some proximal arm function is usually more severely impaired patients. preserved. Most reports show that in patients with dam- age to the CST, even with intensive rehabilitation mea- Neurophysiological factors influencing the recovery of sures, little recovery [28, 30], particularly of hand and upper limb function finger function [70], can be expected. In general, the recovery of arm/hand function following In contrast, the recovery in patients with an intact CNS damage is limited when compared to gait in post- CST is proportional to the initial impairment, with pa- stroke [41] and cervical spinal cord injured [67] subjects, tients recovering approximately 70–80% of the initial even if intensive therapy is applied. In patients with a impairment (proportional recovery rule) [28–30]. Some cervical SCI, arm function depends on the level of the studies indicate that training effects in these patients are lesion. An injury level at C5/C6 results in a ‘passive’ small or absent [46], i.e. only a minor dose-response ef- hand function (supination movement at the elbow joint fects occur [44]. However, there is also evidence that a for hand opening) or, frequently, at C6/7, in a tenodesis higher dose of practice, especially when applied early grasp. This grasp is defined as a hand function when after a stroke, leads to a better outcome of motor func- some forearm extensor muscle activation is preserved tion of the paretic arm [41, 43, 71]. [68]. It allows to close the hand by wrist extension Early after stroke flaccid arm muscle paresis prevails, i. movements with the fingers in a slightly flexed contrac- e. the limbs are weak and do not resist passive displace- ture position. Some spastic muscle tone is required to ment. With the development of some spastic muscle perform such simple grasp movements [24]. tone, needed to perform rudimentary grips, the training In post-stroke subjects, outcome of upper limb func- of residual muscle function can be initiated [24]. In this tion critically depends on the integrity of the corticosp- stage, the focus of therapy/training should be directed to inal tract (CST) [63, 69]. A stroke with damage to the enable the execution of simple reach and grasp move- CST results in lasting impairment of hand and finger ments. In the weeks following stroke, spastic muscle function and an unbalanced muscle tone with forearm tone usually becomes more pronounced in the forearm Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 7 of 15 flexor than in the extensor muscles, as the antigravity contralateral hemispheres become involved in the con- muscles have more muscle mass [39, 72]. This can again trol of each of the two hands during cooperative hand impair the execution of functional reach and grasp movements [84]. Consequently, in post-stroke patients movements. However, some spastic muscle tone in the during the training of cooperative hand tasks, the un- forearm muscles allows the performance of a tenodesis affected hemisphere supports movements of the paretic grasp, which is important for the execution of ADL, not hand and arm [85]. However, the effect of a cooperative only in SCI but also in post-stroke subjects. training on the outcome of hand function remains to be Patients typically compensate for their sensorimotor defi- determined. cits through the involvement of the non-paretic arm/hand, Finally, while the recovery of finger function is limited, leading to learned non-use of the paretic arm [64, 73]. basic functions such as opening and closing the hand Therefore, one important approach to rehabilitate hand should also be trained, as most of the interaction with function after stroke was presented in the form of the environment during ADL involves grasping and re- constraint-induced movement therapy (CIMT). This was leasing objects. Besides motor function, somatosensory based on the idea of enhancing recovery of function by re- function is also of importance during object grasping: ducing interhemispheric inhibition of the stroke hemi- shaping and maintaining a stable grasp during the ma- sphere [74]. By immobilizing the non-affected hand the nipulation of an object relies on the processing of som- patient is forced to use the paretic hand/arm for the per- atosensory input, determined by the mechanical formance of ADL [64]. However so far, a superior effect of properties of the manipulated object [86]. Somatosen- CIMT compared to other therapy approaches was not sory function is often impaired after CNS damage, lead- reported [75]. ing to a visual compensation of movement control. During the course of upper limb rehabilitation, the However, in some patients it can recover spontaneously support provided should always be kept to a minimum or through dedicated training [87]. in order to make the training challenging with a max- imum of individual effort and contribution to movement Implications for robot-assisted therapy of upper limb performance by the patient (for review [24]). However, function the optimal level of assistance also depends on the sever- The combination of kinematic complexity and functional ity of impairment [70]. Most stroke patients will benefit impairment makes the design of robotic devices to train from gravity support, allowing them to perform func- arm, hand and finger function after CNS damage par- tional movements by their own effort [76]. Without such ticularly challenging. Following the initial developments support, shoulder abduction, which is important for ob- based on stiff industrial manipulators, end-effector- ject manipulation, may limit elbow extension and result based devices for planar (MIT-MANUS; [88]) and 3D in concurrent elbow, wrist and finger flexion, i.e. so- (Gentle/S, [89]) reaching movements were introduced to called flexion synergies after stroke [77]. This can affect allow more active contribution of the patient while limit- the execution of functional hand movements. ing the apparent impedance of the robot. Subsequent Many upper limb movements involve the use of both developments focused on incorporating additional de- hands. However, only a few studies provide a neurophysio- grees of freedom (DOF) related to wrist [90] and hand logical basis for the training of bimanual movements [78]. opening/closing function (Gentle/G [91]). For the func- Bimanual training of reaching and grasping tasks in stroke tional training of three-dimensional arm movements patients has been suggested to be more effective in im- with guidance at the three proximal joints, ARMin, a proving unilateral execution of these tasks with the af- grounded, powered exoskeleton was developed, which fected arm than unilateral training alone [79]. This might also integrates grasp and release function [92, 93] (Fig. 2, be a result of stronger recruitment of the contralesional upper panel). hemisphere through bilateral compared to unilateral train- Independent of their kinematic configuration, all of ing [80]. However, there is currently no clear evidence that these systems can partially or fully unload the arm bimanual training is superior to CIMT [65, 81, 82], or un- against gravity. This approach reduces the effect of constrained unimanual training [83]. flexor synergies, and allows the performance of hand The involvement of the unaffected hemisphere in movements within a larger workspace. However, the movement control of the paretic hand might be even complex structure and geared actuators of such devices stronger in a special type of bimanual movement, where with their reflected inertia limit the interaction quality one hand supports the action of the other one by gener- and the ability to adapt the level of support [16]. The ating equal but opposed forces/torques, e.g. when open- large output impedance may render the active initiation ing a bottle or cutting bread. Such cooperative hand of movements more difficult, and potentially alter nat- movements are based on a task-specific control: a ural movement dynamics. Therefore, a trade-off between ‘neural coupling’ of the hemispheres, i.e. both ipsi- and the number of DOF and the quality of the physical Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 8 of 15 interaction exists, limiting the application of these de- cognitive nature of these tasks and motivate patients. In vices to specific stages of recovery. For example, training a meta-analysis, the application of virtual reality (VR) with a powered whole-arm exoskeleton is mainly indi- games was found to be potentially useful for the im- cated for stroke subjects with severe arm paresis early provement of arm function after stroke [34]. after the incident. Similar effects can also be achieved by In conclusion, a good, mainly spontaneous, recovery of using passive devices for gravity support to the upper upper limb function after a stroke can be expected when limb, to enable self-initiated movements [94]. the integrity of the CST is preserved. There is some evi- Robot-assisted approaches should also consider the dence that higher dose of practice leads to improved training of bimanual and cooperative movement tasks as function, especially early after stroke. Nevertheless, in they are important during ADL (Fig. 2, lower panel). Bi- cases with damaged CST the recovery is limited and nei- manual training was a focus of some early studies [10], ther depends on the approach nor on the dose of train- but its potential has not been sufficiently explored and de- ing. Unimanual robot-assisted therapy approaches serves further investigation. Many upper extremity sys- should be complemented by bimanual (cooperative) ap- tems developed and clinically evaluated so far could also proaches. These should also incorporate the training of be used for bimanual training, by combining two devices basic hand function and interaction with virtual object in a mirrored configuration. The training of cooperative dynamics that generate somatosensory feedback. In the hand movements (e.g. opening a bottle) has been pro- future, it will be possible to at least partially compensate posed using a dedicated device [84], and can also be for remaining deficits with wearable assistive robotics. achieved by virtually coupling two unimanual devices through control. Rehabilitation of locomotor function Due to the biomechanical and neural complexity of Locomotor movements are performed more automatic- hand and finger movements, robot-assisted rehabilitation ally than arm and hand movements. Corticospinal con- of hand and finger function became a focus only recently. trol mainly serves the goal to voluntarily alter the Most rehabilitation robots for hand function have been stepping rhythm, e.g. to correct the stepping direction or based on end-effector designs, used either independently amplitude to overcome obstacles. Accordingly, corti- or in combination with grounded exoskeletons or end- cospinal projections to lower limb motoneurons in effector type arm devices (Fig. 2). Several groups have also humans are stronger to the flexor than to the extensor made attempts to develop exoskeleton systems for the muscles [99, 100]. The rehabilitation of locomotor func- hand, some of which assist independent finger motion, tion is simpler than that of upper extremity function, generally resulting in highly complex devices that under- and basic mobility can usually be restored in post-stroke went none to little clinical evaluation. A review [19]found subjects by using the paretic limb as a stick to support that only 25% of 30 hand rehabilitation robots had been the body [24]. clinically tested, and many devices had been considered Passive orthoses can assist foot dorsiflexion in the too complex for clinical use. However, such complexity swing phase of stepping. In SCI subjects, some proximal might not be necessary when the focus is directed to the leg muscle activation is required for a successful loco- basic function of opening and closing the hand [95]. This motor training [101]. Besides this, the rehabilitation of might be sufficient given the limited potential for the re- locomotor function in post-stroke and SCI subjects is covery of finger function following CNS damage, while similar. In severely affected subjects, mobility can be re- remaining highly relevant for ADL. Finally, hand opening/ stored with a wheelchair or other mobility aids. Never- closing can also be supported through wearable assistive theless, the primary goal of rehabilitation is to restore technology, such as soft robotic gloves [96, 97], which sufficient lower-limb function for patients to ambulate could be worn during the performance of ADL. without walking aids. Interaction with the environment occurs mainly through the hands and generates somatosensory feed- Neurophysiological factors influencing the recovery of back. However, somatosensory function is often im- locomotor function paired after CNS damage. Therefore, neurorehabilitation Thirty years ago, rehabilitation after CNS damage was fo- devices for the upper extremity should train hand and, cused on the strengthening of leg muscles to a level where as far as possible, finger function, providing both visual patients were able to perform stepping movements on and haptic feedback [53]. Training should include tasks parallel bars with the support of their arms [102]. In the which are functionally relevant for ADL, such as grasp- early nineties, functional locomotor training with body ing and releasing objects with rendered virtual dynamics unloading of para−/tetraparetic SCI subjects was intro- to also train somatosensory function and sensorimotor duced. This was based on the observation that locomotor integration [98]. Finally, most upper limb training de- function in cat SCI models recovers quite well during vices are embedded in computer games to reflect the treadmill training with body-weight unloading (body Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 9 of 15 weight supported treadmill training, BWSTT) [103]. In in- activation [110]. This is in line with cat experiments, completely paralyzed SCI patients, BWSTT has been where flexor bursts were automatically generated at the shown to result in a similar outcome of stepping function end hip extension despite complete SCI [112, 113]. In compared to a conventional rehabilitation approach [104]. robot-assisted gait training systems, leg flexion move- In post-stroke subjects no gain in outcome was found dur- ments are usually imposed by the robot, leaving the sub- ing BWSTT compared to an unspecific physical exercise ject passive. Over time, this leads to a rarefaction program [105]. The BWSTT training of SCI subjects is [114]and dysfunction[115]ofleg flexor motoneu- physically demanding and requires two physiotherapists rons, i.e. the peripheral nervous system, deprived of who assist leg movements from both sides. As a result, supraspinal input, undergoes degenerative changes. In training time is limited to about half an hour per day, even completely paralyzed patients with an SCI who do though many patients would tolerate more therapy. Yet, not undergo a functional locomotor training, spinal such a dose increase has been associated with a better neuronal circuits underlying stepping movements be- outcome [41]. come silent even when appropriate proprioceptive in- Movement speed during locomotor training represents put is provided. On a longer term this results in a another factor that influences outcome. In ambulatory neuronal dysfunction below the level of the lesion in stroke patients, a successive increase (according to both rodents [116]and patients with SCI [117]. principles of sports physiology) of treadmill speed after a Today we know that bipeds use a quadrupedal coord- 4-week training period resulted in a better walking abil- ination during locomotion, i.e. arm movements repre- ity than conventional gait training [106]. Furthermore, sent an integral part of locomotion [118, 119] and, locomotor training was shown to be most efficient when therefore, might be included in locomotor training pro- delivered in a real-world environment [107]. grams. In fact, recent experiments indicate that arm In severely paralyzed patients with an SCI, automatic movements induce an increase in leg muscle activity stepping movements can be induced, associated with a during stepping [120]. physiological leg muscle activation (i.e. close-to normal timing of electromyography (EMG) patterns with re- duced amplitude), when patients stand on a moving Implications for robot-assisted therapy of lower limb treadmill with the body unloaded up to 80% [108, 109]. function This leg muscle activation is triggered by load receptor Functional gait training positively affects the recovery of input from contact forces during the stance phase of gait locomotor function, but is personnel-intensive and phys- [110]. Such a physiological limb muscle activation was ically demanding for the therapists. This situation trig- found to be the prerequisite for positive training effects gered the development of robotic devices to assist leg and improvement of locomotor function in rodents [22] movements during stepping on a treadmill with the body and patients with a stroke or SCI (for review [24]). With partially unloaded [13]. Robot-assisted BWSTT has been the onset of voluntary control of some proximal leg shown to be as effective as overground stepping with the muscles, body unloading can be reduced and self- support of physiotherapists [121]. Together with the fact induced stepping movements become possible. This is that training intensity has a positive effect on the recov- associated with an increase in strength of leg muscle ac- ery of locomotor ability in post-stroke [41] and SCI [49] tivation. Thus, during the course of training, body un subjects, this motivates the use of robot-assisted −/reloading has to be adapted to the actual degree of BWSTT, allowing longer training times (and thus higher paresis. dose) with less personnel. Furthermore, this approach Most of the recovery of function occurs during the provides a standardized training environment and allows first three months after CNS damage. However, also in an objective assessment of the changes achieved during chronic patients with an incomplete SCI and stroke a the course of rehabilitation [57]. A systematic review significant gain in gait velocity, endurance, and perform- that examined the effect of electromechanical and robot- ance can be achieved by an automated locomotor train- assisted gait training in post-stroke subjects showed that ing [102]. Further improvement of locomotor function patients receiving such training are more likely to after damage to the CNS is associated with only minor achieve independent walking than people who received changes in the leg muscle activity pattern, and relies gait training without such devices [122]. more on a better coordination between the legs and an During the course of rehabilitation, the physical support adapted spastic muscle tone (stroke: [111]; SCI: [37]). has to be continuously adapted to the actual needs of the Hip extension at the end of the stance phase is an es- patient, with the objective of maximizing active participa- sential stimulus for the leg muscle activation during tion of the patient by reducing and selectively providing locomotion, especially for initiating the stance to swing assistance [24]. With the recovery of locomotor function transition with an appropriate change in leg muscle (especially of proximal joints), a transition from a Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 10 of 15 grounded exoskeleton to a grounded end-effector device continue therapy after hospital rehabilitation, but also can take place (e.g. [123]). This can be followed by the to compensate for remaining deficits by providing ap- execution of stepping movements on normal ground with propriate assistance in ADL. reduced support from a wearable robotic exoskeleton or In conclusion, in post-stroke patients training leads to mobility aid. a good recovery of stepping function using spastic Initial developments for robot-assisted gait training muscle tone for body support. In SCI, some remaining were focused on patients with almost complete paralysis proximal leg muscle function is required to allow a suc- as a result of SCI, where training is most demanding for cessful training and recovery of function. The activation therapists. These patients can hardly actively contribute of load (re-loading of the body as far as possible) and to the leg movements. Therefore, high assistive torques hip-joint related (hip extension) receptors leads to a are required, typically resulting in robotic devices with physiological leg muscle activity pattern during stepping high output impedance. With increased penetration of and, consequently, to dose-dependent training effects. this technology into clinics, locomotor training using Accordingly, devices are required which can adapt the such devices was expanded to patients with a stroke or support and impedance of individual joints according to incomplete SCI, who require less and/or asymmetric as- patients’ impairment. The development of wearable ro- sistance. However, this is difficult to achieve by devices botic gait orthoses with integrated balance support will with a high output impedance, as these behave more like further promote functional training, engagement and a velocity than a torque source. Consequently, novel motivation, and lead to systems that can provide long- control approaches [124] and devices with low intrinsic term assistance in the home environment (Fig. 3). impedance [125] were developed to better adapt the physical support to these patient populations. These ef- Conclusions forts need to be continued, also to assure that automated Rehabilitation training of the upper and lower limb assessments reflect the current impairment level of a pa- should be founded on neurophysiological insights, inde- tient, and are not masked by the device dynamics [57]. pendent of whether it is performed conventionally, or The field of lower-extremity robot-assisted therapy is pro- with the support of robotic devices. After CNS damage, gressing towards wearable powered exoskeletons (Fig. 3). improvement of sensorimotor functions occurs to a large These combine the advantages of grounded devices with degree spontaneously and can further be achieved by an the ability to train in a real-world environment and provide exploitation of neuroplasticity. This is reflected in a higher levels of subject participation and challenge [58]. physiological limb muscle activation that might serve as Even more than in the case of grounded devices, it is a a marker for the achievement of training effects. This re- challenge to achieve low output impedance together with quires voluntarily performed upper limb movements, or the provision of sufficient assistive torque in wearable exo- an activation of appropriate receptors for a purposeful skeletons, as all links and joints are integrated and thus car- activation of lower limb muscles, i.e. during stepping ried by the moving exoskeleton. This results in weight and movements. complexity constraints that limit both the number of DOF The potential for a recovery of function differs not that can be actuated and the transparency (ability of the only between upper and lower limbs, but also between system to getout of theway)thatcan be achieved. The neurological disorders such as stroke and SCI, and re- high reduction ratios required to generate sufficient torque quires the development of technology accounting for increase the output impedance of the device, thus limiting these differences. Table 1 summarizes the main aspects the capacity to adapt the support to changing abilities of of neurorehabilitation and outcome in these disorders, the patient. With technological progress, it might become as well as the implications for rehabilitation technology. possible to modulate the output impedance of each joint. Training effects in upper compared to lower limbs are Through this approach, hip extension might be enforced to limited and are mainly determined by corticospinal tract trigger physiological leg flexion movements for the initi- integrity. Nevertheless, intensive, highly dosed training ation of the swing phase (cf. [110]), which the exoskeleton has beneficial effects, especially early after stroke. Training could passively follow. Such an approach would allow a bet- devices should unload (and gradually re-load) arm move- ter adaptation of the support to the individual patient, en- ments against gravity, incorporate hand function for reach able more dynamic motion, and prevent degenerative and grasp training and use the motivating and cognitively changes in the peripheral nervous system. engaging effects of virtual reality, with a focus on the ADL With the ability to partially support balance with that are most important to the individual. wearable exoskeletons, the hands become free from For the lower limbs, the effects of training on the holding crutches. This will facilitate arm swing, which recovery of sensorimotor function seem to depend on both represents an integral part of locomotion (for review their intensity and dose. Rehabilitation robots are ideal tools see [118]). These devices will not only allow to to complement classical functional therapy by allowing a Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 11 of 15 Table 1 Main aspects of neurorehabilitation and outcome, and their implications for rehabilitation technology Limb Condition Typical recovery course Goal Rehabilitation approach Technology UL stroke damaged CST little recovery, esp. prox. arm muscle prox. arm muscle therapy: passive chronic impairment of activation; avoidance of strengthening; mobilization (position hand/finger extension muscle contractures; use continuous passive limb control) or weight of impaired limb for motion; training of support for self- support/holding compensatory strategies initiated proximal function movements; active/passive hand module with extension bias assistance: supported arm/hand motion (admittance control) vial intention detection (e.g. force, EMG, gaze) intact CST spontaneous recovery of arm reaching and functional reach/grasp therapy: proximal gravity approx. 70–80% of intial simple grasping and bimanual support during reach/ arm/hand impairment function; uni−/bimanual (cooperative) hand grasp; training of ADL functions movements; individual joints using strengthening of wrist/ dedicated devices, finger extensors; simple including hand/fingers, movement training with as well as (cooperative) transfer to ADL; limited bimanual training (Fig. 2) dose-dependent training assistance: passive effects: subacute proximal gravity support > chronic stage combined with active wrist/finger support via residual function amplification (force/EMG control) SCI (incomplete) typical lesion spastic forearm flexor tenodesis grasp; assistance: active level C6/7 muscle tone impeding bimanual grasp exoskeleton/glove to the development of facilitate wrist and finger tenodesis grasp flexion/extension triggered by proximal arm motion (e.g. joint angle sensor) LL stroke hemiparesis spontaneous recovery; non-assisted ambulation generation of therapy: body-weight spastic muscle support; appropriate afferent support according to reduced level of input from load (body paresis; adapted stepping movement un/reloading) and hip movement support ability receptors (hip extension) (position/admittance during stepping; control for severe SCI (incomplete) paraparesis some prox. leg muscle assisted/independent importance of stepping impairment and variable function and spastic ambulation velocity and hip impedance control for muscle tone required for extension (initiation of mild/moderate stepping ability swing); dose-dependent impairment; Fig. 3); leg training effects flexor activation through robotically assisted hip extension UL upper limb, LL lower limb, SCI spinal cord injury, CST corticospinal tract, ADL activities of daily living, EMG electromyography standardized and intensive training with individually and to adapt their output impedance and physical support to continuously adapted physical support. Grounded exoskele- the actual state of the patient and the task at hand, with- tons and end-effector devices as well as wearable exoskele- out altering functional movement patterns through their tons seem to be equally effective in the improvement of apparent dynamics. Patients will likely train with differ- function. However, their suitability depends on the phase of ent devices throughout the recovery phase during re- recovery, and the individual impairment (Fig. 3). habilitation, to optimally adapt movement complexity Rehabilitation robots should always provide targeted and physical support to the current state and functional physical support adapted to the functional abilities of abilities of the patient (e.g. transitioning from left to the patient in a way to enable functional movements. right in Fig. 3). To deal with this challenge, the design of This has strong implications for the design, instrumenta- future robotic rehabilitation systems should also con- tion and control of such systems. These should be able sider the relevance of particular joints during functional Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 12 of 15 movement (e.g. the hip joint plays a larger role in loco- Ethics approval and consent to participate Not applicable. motion than the knee and ankle joints; [110]) and their potential for recovery (e.g. limited recovery of individual Competing interests finger movement). Volker Dietz is member of the Scientific Advisory Board of Hocoma AG. There are currently a number of novel and exciting developments in and around the fields of rehabilitation Publisher’sNote engineering and rehabilitation sciences. Advances in ma- Springer Nature remains neutral with regard to jurisdictional claims in terial sciences will allow lighter, more customizable published maps and institutional affiliations. structures with more tightly integrated actuation and Author details sensing. Furthermore, there is an increasing focus on 1 Department of Health Sciences and Technology, ETH Zurich, 8092 Zurich, combining robotics with non-invasive [126, 127] and Switzerland. Spinal Cord Injury Center, Balgrist University Hospital, 8008 Zurich, Switzerland. invasive [128] brain-machine interfaces or neuropros- thetics, with the aim of promoting independence during Received: 20 January 2018 Accepted: 7 May 2018 activities of daily living. These approaches are at an early stage and still face a number of challenges. Nevertheless, References even an optimal exploitation of neuroplasticity (cf. [24]) 1. Reinkensmeyer D, Dietz V. Neurorehabilitation Technology. 2nd ed. Springer, will not result in a full functional recovery. The field Cham; 2016. should therefore focus on wearable systems that not only 2. Khalili D, Zomlefer M. An intelligent robotic system for rehabilitation of joints and estimation of body segment parameters. IEEE Trans Biomed Eng. support functional therapy, but that can also serve as as- 1988;35(2):138–46. sistive devices to compensate for persisting sensorimotor 3. Rabishong P, Bel J, Hill J, Peruchon E, Simeon M, Screve J, et al. The AMOLL deficits. Advanced actuation, sensing and control ap- project (active modular orthosis for lower limbs). In: Proc Int Symp external control hum extremities, vol. 1975; 1975. p. 33–42. proaches will make these systems more robust and applic- 4. Seireg A, Grundmann J. Design of a multitask exoskeletal walking device for able for ADL tasks, and applicable both in the clinic and at paraplegics. Biomechanics of medical devices. 1981:569–644. home. In the future, it can be expected that wearable de- 5. Vukobratovic M, Hristic D, Stojiljkovic Z. Development of active anthropomorphic exoskeletons. Med Biol Eng. 1974;12(1):66–80. vices continuously adapt and reduce support until recovery 6. Ali H. Bionic exoskeleton: history, development and the future. In: plateaus, and then compensate for chronic impairments. International conference on advances in Engineering & Technology: IOSR A number of further challenges remain for the field of Journal of Mechanical and Civil Engineering (IOSR-JMCE); 2014;58–62. 7. Rabischong P. Pneumatic Exoskeleton Prosthesis http://cyberneticzoocom/ robot-assisted therapy and assistance, many of which will bionics/1976-pneumatic-exoskeleton-prosthesis-pierre-rabischong-french/. also require collaboration with industry and government bodies. Adoption of robotic devices is driven by cost and 8. Dijkers MP, de Bear PC, Erlandson RF, Kristy K, Geer DM, Nichols A. Patient and staff acceptance of robotic technology in occupational therapy: a pilot reimbursement systems, and research should therefore study. J Rehabil Res Dev. 1991;28(2):33–44. also focus on identifying the simplest, most effective tech- 9. Krebs HI, Hogan N, Aisen ML, Volpe BT. Robot-aided neurorehabilitation. nical solutions that can support the rehabilitation process. IEEE Trans Rehabil Eng. 1998;6(1):75–87. 10. Lum PS, Reinkensmeyer DJ, Lehman SL. Robotic assist devices for bimanual Finally, future rehabilitation approaches will not only physical therapy: preliminary experiments. IEEE Trans Rehabil Eng. 1993;1(3): profit from the inclusion of robots, but also from an ad- 185–91. vanced understanding of neurophysiological mechanisms 11. Lum PS, Lehman SL, Reinkensmeyer DJ. The bimanual lifting rehabilitator: an adaptive machine for therapy of stroke patients. IEEE Trans Rehabil Eng. underlying normal and impaired sensorimotor functions, 1995;3(2):166–74. enabled by the use of robots as scientific tools [129]. 12. Burgar CG, Lum PS, Shor PC, Machiel Van der Loos HF. Development of Resulting insights will benefit the development of ad- robots for rehabilitation therapy: the Palo alto VA/Stanford experience. J Rehabil Res Dev. 2000;37(6):663–73. vanced rehabilitation robots, and further promote col- 13. Colombo G, Joerg M, Schreier R, Dietz V. Treadmill training of paraplegic laboration between engineers, therapists and clinical patients using a robotic orthosis. J Rehabil Res Dev. 2000;37(6):693–700. neurophysiologists. 14. Hesse S, Uhlenbrock D. A mechanized gait trainer for restoration of gait. J Rehabil Res Dev. 2000;37(6):701–8. Acknowledgments 15. Just F, Baur K, Klamroth-Marganska V, Riener R, Rauter G. Motor inertia We thank Stefan Schneller for preparing the graphics included in this compensation of the ARMin rehabilitation robot. In: AUTOMED workshop manuscript, Raffaele Ranzani for his valuable and critical feedback on the 2016: Hochschule Wismar; 2016. manuscript, and Monica Stüssi for formatting the references. 16. Metzger JC, Lambercy O, Gassert, R. Performance comparison of interaction control strategies on a hand rehabilitation robot. IEEE International Funding Conference on Rehabilitation Robotics (ICORR). 2015;846–851. https://doi. The chair in Rehabilitation Engineering (RG) is supported by the ETH Zurich org/10.1109/ICORR.2015.7281308. Foundation in collaboration with Hocoma AG. 17. Klamroth-Marganska V, Blanco J, Campen K, Curt A, Dietz V, Ettlin T, et al. Three-dimensional, task-specific robot therapy of the arm after stroke: a Authors’ contributions multicentre, parallel-group randomised trial. Lancet Neurol. 2014;13(2):159–66. Both authors contributed equally to this manuscript. Each author initially 18. Lo AC, Guarino PD, Richards LG, Haselkorn JK, Wittenberg GF, Federman DG, focused on the parts related to his respective background (RG: rehabilitation et al. Robot-assisted therapy for long-term upper-limb impairment after robotics, VD: neurophysiological aspects of healthy and impaired functional stroke. N Engl J Med. 2010;362:1772–83. movement), which were then merged through numerous discussions and 19. Balasubramanian S, Klein J, Burdet E. Robot-assisted rehabilitation of hand iterations. Both authors read and approved the final manuscript. function. Curr Opin Neurol. 2010;23(6):661–70. Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 13 of 15 20. Kwakkel G, Kollen BJ, Krebs HI. Effects of robot-assisted therapy on upper 47. Milot MH, Spencer SJ, Chan V, Allington JP, Klein J, Chou C, et al. A limb recovery after stroke: a systematic review. Neurorehabil Neural Repair. crossover pilot study evaluating the functional outcomes of two different 2008;22(2):111–21. types of robotic movement training in chronic stroke survivors using the 21. Tefertiller C, Pharo B, Evans N, Winchester P. Efficacy of rehabilitation arm exoskeleton BONES. J Neuroeng Rehabil. 2013;10:112. robotics for walking training in neurological disorders: a review. J Rehabil 48. Schaefer SY, Patterson CB, Lang CE. Transfer of training between distinct Res Dev. 2011;48(4):387–416. motor tasks after stroke: implications for task-specific approaches to 22. Edgerton VR, Tillakaratne NJ, Bigbee AJ, de Leon RD, Roy RR. Plasticity of the upper-extremity neurorehabilitation. Neurorehabil Neural Repair. spinal neural circuitry after injury. Annu Rev Neurosci. 2004;27:145–67. 2013;27(7):602–12. 49. Wirz M, Mach O, Maier D, Benito-Penalva J, Taylor J, Esclarin A, et al. 23. Nudo RJ, Plautz EJ, Frost SB. Role of adaptive plasticity in recovery of function after damage to motor cortex. Muscle Nerve. 2001;24(8):1000–19. Effectiveness of automated locomotor training in patients with acute incomplete spinal cord injury: a randomized controlled multicenter trial. 24. Dietz V, Fouad K. Restoration of sensorimotor functions after spinal cord J Neurotrauma. 2017;34:1891–6. injury. Brain. 2014;137(Pt 3):654–67. 25. Zeiler SR, Krakauer JW. The interaction between training and plasticity in the 50. Lang CE, Macdonald JR, Reisman DS, Boyd L, Jacobson Kimberley T, poststroke brain. Curr Opin Neurol. 2013;26(6):609–16. Schindler-Ivens SM, et al. Observation of amounts of movement practice 26. Zarahn E, Alon L, Ryan SL, Lazar RM, Vry MS, Weiller C, et al. Prediction of provided during stroke rehabilitation. Arch Phys Med Rehabil. 2009;90(10): motor recovery using initial impairment and fMRI 48 h poststroke. Cereb 1692–8. Cortex. 2011;21(12):2712–21. 51. Riener R, Lunenburger L, Colombo G. Human-centered robotics applied to gait training and assessment. J Rehabil Res Dev. 2006;43(5):679–94. 27. Curt A, Van Hedel HJ, Klaus D, Dietz V. Recovery from a spinal cord injury: 52. Marchal-Crespo L, McHughen S, Cramer SC, Reinkensmeyer DJ. The effect of significance of compensation, neural plasticity, and repair. J Neurotrauma. haptic guidance, aging, and initial skill level on motor learning of a steering 2008;25(6):677–85. task. Exp Brain Res. 2010;201(2):209–20. 28. Prabhakaran S, Zarahn E, Riley C, Speizer A, Chong JY, Lazar RM, et al. Inter- individual variability in the capacity for motor recovery after ischemic stroke. 53. Metzger JC, Lambercy O, Califfi A, Conti FM, Gassert R. Neurocognitive Neurorehabil Neural Repair. 2008;22:64–71. robot-assisted therapy of hand function. IEEE Trans Haptics. 2014;7(2):140–9. 29. Kwakkel G, Kollen B, Lindeman E. Understanding the pattern of functional 54. Metzger JC, Lambercy O, Califfi A, Dinacci D, Petrillo C, Rossi P, et al. recovery after stroke: facts and theories. Restor Neurol Neurosci. 2004;22(3–5): Assessment-driven selection and adaptation of exercise difficulty in 281–99. robot-assisted therapy: a pilot study with a hand rehabilitation robot. 30. Winters C, van Wegen EE, Daffertshofer A, Kwakkel G. Generalizability of the J Neuroeng Rehabil. 2014;11:154. proportional recovery model for the upper extremity after an ischemic 55. Zimmerli L, Krewer C, Gassert R, Muller F, Riener R, Lunenburger L. stroke. Neurorehabil Neural Repair. 2015;29(7):614–22. Validation of a mechanism to balance exercise difficulty in robot-assisted 31. Jakob W, Wirz M, van Hedel HJ, Dietz V. Difficulty of elderly SCI subjects to upper-extremity rehabilitation after stroke. J Neuroeng Rehabil. 2012;9:6. 56. Stefan K, Kunesch E, Cohen LG, Benecke R, Classen J. Induction of plasticity in the translate motor recovery—“body function”–into daily living activities. J human motor cortex by paired associative stimulation. Brain. 2000;123:572–84. Neurotrauma. 2009;26(11):2037–44. 32. Huang VS, Krakauer JW. Robotic neurorehabilitation: a computational motor 57. Lambercy O, Maggioni S, Lünenburger L, Gassert R, Bolliger M. Robotic and learning perspective. J Neuroeng Rehabil. 2009;6:5. wearable sensor Technologies for Measurements/Clinical Assessments. In: 33. Krakauer JW. Motor learning: its relevance to stroke recovery and Reinkensmeyer D, Dietz V, editors. Neurorehabilitation Technology. Springer: neurorehabilitation. Curr Opin Neurol. 2006;19(1):84–90. Cham; 2016. 34. Saposnik G, Levin M, Group ORCW. Virtual reality in stroke rehabilitation: a 58. Pennycott A, Wyss D, Vallery H, Klamroth-Marganska V, Riener R. Towards meta-analysis and implications for clinicians. Stroke. 2011;42(5):1380–6. more effective robotic gait training for stroke rehabilitation: a review. 35. Latash M, Anson J. What are “normal movements” in atypic populations? J Neuroeng Rehabil. 2012;9:65. Behav Brain Sci. 1996;19:55–106. 59. Herder JG. Ideen zur Philiosophie der Geschichte der Menschheit, Bd. 1 edn. Leipzig.Hartknoch; 1785. 36. O'Dwyer NJ, Ada L, Neilson PD. Spasticity and muscle contracture following 60. Lemon RN. Descending pathways in motor control. Annu Rev Neurosci. stroke. Brain. 1996;119:1737–49. 2008;31:195–218. 37. Dietz V, Sinkjaer T. Secondary changes after CNS damage: the significance of spastic muscle tone in rehabilitation. In: Dietz V, Ward N, editors. Oxford 61. Jorgensen HS, Reith J, Nakayama H, Kammersgaard LP, Raaschou HO, Olsen textbook of neurorehabilitation. Oxford: Oxford University Press; 2015. TS. What determines good recovery in patients with the most severe 38. Katz RT, Rymer WZ. Spastic hypertonia: mechanisms and measurement. strokes? The Copenhagen Stroke Study Stroke. 1999;30(10):2008–12. Arch Phys Med Rehabil. 1989;70(2):144–55. 62. Ward NS, Newton JM, Swayne OB, Lee L, Thompson AJ, Greenwood RJ, 39. Dietz V, Sinkjaer T. Spastic movement disorder: impaired reflex function and et al. Motor system activation after subcortical stroke depends on altered muscle mechanics. Lancet Neurol. 2007;6(8):725–33. corticospinal system integrity. Brain. 2006;129(Pt 3):809–19. 63. Byblow WD, Stinear CM, Barber PA, Petoe MA, Ackerley SJ. Proportional 40. Ada L, Dorsch S, Canning CG. Strengthening interventions increase strength recovery after stroke depends on corticomotor integrity. Ann Neurol. 2015; and improve activity after stroke: a systematic review. Aust J Physiother. 78(6):848–59. 2006;52(4):241–8. 64. Taub E, Uswatte G, Pidikiti R. Constraint-induced movement therapy: a new 41. Kwakkel G, Wagenaar RC, Twisk JW, Lankhorst GJ, Koetsier JC. Intensity of leg family of techniques with broad application to physical rehabilitation–a and arm training after primary middle-cerebral-artery stroke: a randomised trial. clinical review. J Rehabil Res Dev. 1999;36(3):237–51. Lancet. 1999;354(9174):191–6. 42. Lohse KR, Lang CE, Boyd LA. Is more better? Using metadata to explore 65. Stoykov ME, Lewis GN, Corcos DM. Comparison of bilateral and unilateral dose-response relationships in stroke rehabilitation. Stroke. 2014;45(7): training for upper extremity hemiparesis in stroke. Neurorehabil Neural 2053–8. Repair. 2009;23:945–53. 43. Wu X, Guarino P, Lo AC, Peduzzi P, Wininger M. Long-term effectiveness of 66. Kilbreath SL, Heard RC. Frequency of hand use in healthy older persons. intensive therapy in chronic stroke. Neurorehabil Neural Repair. 2016;30(6):583–90. Aust J Physiother. 2005;51(2):119–22. 44. Lang CE, Strube MJ, Bland MD, Waddell KJ, Cherry-Allen KM, Nudo RJ, et al. 67. Zariffa J, Kapadia N, Kramer JL, Taylor P, Alizadeh-Meghrazi M, Zivanovic V, et al. Feasibility and efficacy of upper limb robotic rehabilitation in a subacute Dose response of task-specific upper limb training in people at least 6 cervical spinal cord injury population. Spinal Cord. 2012;50(3):220–6. months poststroke: a phase II, single-blind, randomized, controlled trial. Ann Neurol. 2016;80(3):342–54. 68. Dietz V, Curt A. Neurological aspects of spinal-cord repair: promises and challenges. Lancet Neurol. 2006;5(8):688–94. 45. McCabe J, Monkiewicz M, Holcomb J, Pundik S, Daly JJ. Comparison of robotics, functional electrical stimulation, and motor learning methods for 69. Stinear CM, Byblow WD, Ackerley SJ, Smith MC, Borges VM, Barber PA. treatment of persistent upper extremity dysfunction after stroke: a Proportional motor recovery after stroke: implications for trial design. Stroke. randomized controlled trial. Arch Phys Med Rehabil. 2015;96(6):981–90. 2017;48(3):795–8. 46. Waddell KJ, Strube MJ, Bailey RR, Klaesner JW, Birkenmeier RL, Dromerick 70. Rowe JB, Chan V, Ingemanson ML, Cramer SC, Wolbrecht ET, Reinkensmeyer AW, et al. Does task-specific training improve upper limb performance in DJ. Robotic assistance for training finger movement using a Hebbian model: a daily life Poststroke? Neurorehabil Neural Repair. 2017;31(3):290–300. randomized controlled trial. Neurorehabil Neural Repair. 2017;31(8):769–80. Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 14 of 15 71. Veerbeek JM, van Wegen E, van Peppen R, van der Wees PJ, Hendriks E, 96. In H, Kang BB, Sin M, Cho KJ. Exo glove: a wearable robot for the hand with Rietberg M, et al. What is the evidence for physical therapy poststroke? A a soft tendon routing system. IEEE Robotics & Automation Magazine 2015; systematic review and meta-analysis. PLoS One. 2014;9(2):e87987. 1(Mar 22):97-105. 72. Marciniak C, Rader L, Gagnon C. The use of botulinum toxin for spasticity after 97. Arata J, Ohmoto K, Gassert R, Lambercy O, Fujimoto H, Wada I. A new hand spinal cord injury. Am J Phys Med Rehabil. 2008;87(4):312–7. quiz 318-320, 329 exoskeleton device for rehabilitation using a three-layered sliding spring 73. Taub E, Crago JE, Burgio LD, Groomes TE, Cook EW 3rd, DeLuca SC, et al. An mechanism. IEEE International Conference on Robotics and Automation. operant approach to rehabilitation medicine: overcoming learned nonuse 2013;3902–3907. https://doi.org/10.1109/ICRA.2013.6631126. by shaping. J Exp Anal Behav. 1994;61(2):281–93. 98. Metzger JC, Lambercy O, Gassert R. High-fidelity rendering of virtual objects with the ReHapticKnob - Novel avenues in robot-assisted rehabilitation of 74. Di Pino G, Pellegrino G, Assenza G, Capone F, Ferreri F, Formica D, et al. hand function. IEEE Haptics Symposium. 2012;51–6. https://doi.org/10.1109/ Modulation of brain plasticity in stroke: a novel model for HAPTIC.2012.6183769. neurorehabilitation. Nat Rev Neurol. 2014;10(10):597–608. 75. Kwakkel G, Veerbeek JM, van Wegen EE, Wolf SL. Constraint-induced 99. Brouwer B, Ashby P. Corticospinal projections to lower limb motoneurons in movement therapy after stroke. Lancet Neurol. 2015;14(2):224–34. man. Exp Brain Res. 1992;89(3):649–54. 76. Beer RF, Ellis MD, Holubar BG, Dewald JP. Impact of gravity loading on post-stroke 100. Schubert M, Curt A, Jensen L, Dietz V. Corticospinal input in human gait: reaching and its relationship to weakness. Muscle Nerve. 2007;36(2):242–50. modulation of magnetically evoked motor responses. Exp Brain Res. 1997; 77. Lan Y, Yao J, Dewald J. Increased shoulder abduction loads decreases 115(2):234–46. volitional finger extension in individuals with chronic stroke: preliminary 101. Wirz M, van Hedel HJ, Rupp R, Curt A, Dietz V. Muscle force and gait findings. Annual International Conference of the IEEE Engineering in performance: relationships after spinal cord injury. Arch Phys Med Rehabil. Medicine and Biology Society (EMBC). 2014;5808–5811. 2006;87(9):1218–22. 102. Wirz M, Zemon DH, Rupp R, Scheel A, Colombo G, Dietz V, et al. Effectiveness 78. McCombe Waller S, Whitall J. Bilateral arm training: why and who benefits? of automated locomotor training in patients with chronic incomplete spinal NeuroRehabilitation. 2008;23(1):29–41. cord injury: a multicenter trial. Arch Phys Med Rehabil. 2005;86(4):672–80. 79. Mudie MH, Matyas TA. Can simultaneous bilateral movement involve the undamaged hemisphere in reconstruction of neural networks damaged by 103. Barbeau H, Wainberg M, Finch L. Description and application of a system stroke? Disabil Rehabil. 2000;22(1–2):23–37. for locomotor rehabilitation. Med Biol Eng Comput. 1987;25(3):341–4. 80. Luft AR, McCombe-Waller S, Whitall J, Forrester LW, Macko R, Sorkin JD, et 104. Dobkin B, Barbeau H, Deforge D, Ditunno J, Elashoff R, Apple D, et al. The al. Repetitive bilateral arm training and motor cortex activation in chronic evolution of walking-related outcomes over the first 12 weeks of stroke: a randomized controlled trial. JAMA. 2004;292(15):1853–61. rehabilitation for incomplete traumatic spinal cord injury: the multicenter 81. Lin KC, Chen YA, Chen CL, Wu CY, Chang YF. The effects of bilateral arm randomized spinal cord injury locomotor trial. Neurorehabil Neural Repair. training on motor control and functional performance in chronic stroke: a 2007;21(1):25–35. randomized controlled study. Neurorehabil Neural Repair. 2010;24(1):42–51. 105. Duncan PW, Sullivan KJ, Behrman AL, Azen SP, Wu SS, Nadeau SE, et al. Body-weight-supported treadmill rehabilitation after stroke. N Engl J Med. 82. vanDelden AL, Beek PJ,RoerdinkM,KwakkelG,Peper CL.Unilateral 2011;364(21):2026–36. and bilateral upper-limb training interventions after stroke have similar 106. Pohl M, Mehrholz J, Ritschel C, Ruckriem S. Speed-dependent treadmill effects on bimanual coupling strength. Neurorehabil Neural Repair. training in ambulatory hemiparetic stroke patients: a randomized controlled 2015;29(3):255–67. trial. Stroke. 2002;33(2):553–8. 83. Whitall J, McCombe Waller S, Silver KH, Macko RF. Repetitive bilateral arm training with rhythmic auditory cueing improves motor function in chronic 107. Sandler EB, Roach KE, Field-Fote EC. Dose-response outcomes associated hemiparetic stroke. Stroke. 2000;31(10):2390–5. with different forms of locomotor training in persons with chronic motor- 84. Dietz V, Macauda G, Schrafl-Altermatt M, Wirz M, Kloter E, Michels L. Neural incomplete spinal cord injury. J Neurotrauma. 2017;34(10):1903–8. coupling of cooperative hand movements: a reflex and FMRI study. Cereb 108. Dietz V, Colombo G, Jensen L. Locomotor activity in spinal man. Lancet. Cortex. 2015;25(4):948–58. 1994;344(8932):1260–3. 85. Schrafl-Altermatt M, Dietz V. Cooperative hand movements in stroke 109. Dietz V, Colombo G, Jensen L, Baumgartner L. Locomotor capacity of spinal patients: neural reorganization. Clin Neurophysiol. 2016;127(1):748–54. cord in paraplegic patients. Ann Neurol. 1995;37(5):574–82. 110. Dietz V, Muller R, Colombo G. Locomotor activity in spinal man: significance of 86. Johansson RS. How is grasping modifed by somatosensory input. Motor afferent input from joint and load receptors. Brain. 2002;125(Pt 12):2626–34. Control: Consepts and Issues. 1991;14:331–5. 87. Carey L, Macdonell R, Matyas TA. SENSe: study of the effectiveness of 111. Den Otter AR, Geurts AC, Mulder T, Duysens J. Gait recovery is not neurorehabilitation on sensation: a randomized controlled trial. associated with changes in the temporal patterning of muscle activity Neurorehabil Neural Repair. 2011;25(4):304–13. during treadmill walking in patients with post-stroke hemiparesis. Clin 88. Aisen ML, Krebs HI, Hogan N, McDowell F, Volpe BT. The effect of robot- Neurophysiol. 2006;117(1):4–15. assisted therapy and rehabilitative training on motor recovery following 112. Grillner S, Rossignol S. On the initiation of the swing phase of locomotion stroke. Arch Neurol. 1997;54(4):443–6. in chronic spinal cats. Brain Res. 1978;146(2):269–77. 89. Loureiro R, Amirabdollahian F, Topping M, Driessen B, Harwin W. Upper 113. Kriellaars DJ, Brownstone RM, Noga BR, Jordan LM. Mechanical entrainment of limb robot mediated stroke therapy - GENTLE/s approch. Auton Robot. fictive locomotion in the decerebrate cat. J Neurophysiol. 1994;71(6):2074–86. 2003;15(1):35–51. 114. Lin CS, Macefield VG, Elam M, Wallin BG, Engel S, Kiernan MC. Axonal changes in spinal cord injured patients distal to the site of injury. Brain. 90. Masia L, Krebs HI, Cappa P, Hogan N. Design and characterization of hand 2007;130(Pt 4):985–94. module for whole-arm rehabilitation following stroke. IEEE ASME Trans Mechatron. 2007;12(4):399–407. 115. Dietz V, Muller R. Degradation of neuronal function following a spinal 91. Loureiro RC, Lamperd B, Collin C, Harwin WS. Reach & grasp therapy: Effects cord injury: mechanisms and countermeasures. Brain. 2004;127(Pt 10): of the Gentle/G System assessing sub-acute stroke whole-arm rehabilitation. 2221–31. IEEE International Conference on Rehabilitation Robotics. 2009;755–760. 116. Beauparlant J, van den Brand R, Barraud Q, Friedli L, Musienko P, Dietz V, 92. Nef T, Klamroth-Marganska V, Keller U, Riener R. Three-dimensional multi- et al. Undirected compensatory plasticity contributes to neuronal degree-of-freedom arm therapy robot (ARMin). In Neurorehabilitation dysfunction after severe spinal cord injury. Brain. 2013;136(Pt 11):3347–61. Technology. Springer: Cham. 2016;351–374. 117. Dietz V, Grillner S, Trepp A, Hubli M, Bolliger M. Changes in spinal reflex and 93. Nef T, Riener R. Three-dimensional-multi-degree of freedom arm therapy locomotor activity after a complete spinal cord injury: a common robot (ARMin). In: Dietz V, Nef T, Rymer WZ, editors. Neurorehabilitation mechanism? Brain. 2009;132(Pt 8):2196–205. technology. London: Springer; 2012. 118. Dietz V. Do human bipeds use quadrupedal coordination? Trends Neurosci. 2002;25(9):462–7. 94. Housman SJ, Scott KM, Reinkensmeyer DJ. A randomized controlled trial of gravity-supported, computer-enhanced arm exercise for individuals with 119. Pauvert V, Pierrot-Deseilligny E, Rothwell JC. Role of spinal premotoneurones in severe hemiparesis. Neurorehabil Neural Repair. 2009;23(5):505–14. mediating corticospinal input to forearm motoneurones in man. J Physiol. 95. Lambercy O, Dovat L, Yun H, Wee SK, Kuah CW, Chua KS, et al. Effects of a 1998;508:301–12. robot-assisted training of grasp and pronation/supination in chronic stroke: 120. de Kam D, Rijken H, Manintveld T, Nienhuis B, Dietz V, Duysens J. Arm a pilot study. J Neuroeng Rehabil. 2011;8:63. movements can increase leg muscle activity during submaximal recumbent Gassert and Dietz Journal of NeuroEngineering and Rehabilitation (2018) 15:46 Page 15 of 15 stepping in neurologically intact individuals. J Appl Physiol (1985). 2013; 115(1):34–42. 121. Dobkin B, Apple D, Barbeau H, Basso M, Behrman A, Deforge D, et al. Weight-supported treadmill vs over-ground training for walking after acute incomplete SCI. Neurology. 2006;66(4):484–93. 122. Mehrholz J, Thomas S, Werner C, Kugler J, Pohl M, Elsner B. Electromechanical-assisted training for walking after stroke. Cochrane database Syst Rev. 2017;5:CD006185. 123. Hesse S, Schattat N, Mehrholz J, Werner C. Evidence of end-effector based gait machines in gait rehabilitation after CNS lesion. NeuroRehabilitation. 2013;33(1):77–84. 124. Duschau-Wicke A, von Zitzewitz J, Caprez A, Lünenburger L, Riener R. Path control: a method for patient-cooperative robot-aided gait rehabilitation. IEEE Trans Neural Syst Rehabil Eng. 2010;18:38–48. 125. Veneman JF, Kruidhof R, Hekman EE, Ekkelenkamp R, Van Asseldonk EH, van der Kooij H. Design and evaluation of the LOPES exoskeleton robot for interactive gait rehabilitation. IEEE Trans Neural Syst Rehabil Eng. 2007;15(3):379–86. 126. Buch E, Weber C, Cohen LG, Braun C, Dimyan MA, Ard T, et al. Think to move: a neuromagnetic brain-computer interface (BCI) system for chronic stroke. Stroke. 2008;39(3):910–7. 127. Carda S, Biasiucci A, Maesani A, Ionta S, Moncharmont J, Clarke S, et al. Electrically assisted movement therapy in chronic stroke patients with severe upper limb paresis: a pilot, single-blind, randomized crossover study. Arch Phys Med Rehabil. 2017;98(8):1628–35. e1622 128. Hochberg LR, Bacher D, Jarosiewicz B, Masse NY, Simeral JD, Vogel J, et al. Reach and grasp by people with tetraplegia using a neurally controlled robotic arm. Nature. 2012;485(7398):372–5. 129. Wolpert DM, Flanagan JR. Q&A: robotics as a tool to understand the brain. BMC Biol. 2010;8:92. 130. Loureiro RC, Harwin WS, Nagai K, Johnson M. Advances in upper limb stroke rehabilitation: a technology push. Med Biol Eng Comput. 2011;49(10):1103–18. 131. Maciejasz P, Eschweiler J, Gerlach-Hahn K, Jansen-Troy A, Leonhardt S. A survey on robotic devices for upper limb rehabilitation. J Neuroeng Rehabil. 2014;11:3. 132. Sheng B, Zhang Y, Meng W, Deng C, Xie S. Bilateral robots for upper-limb stroke rehabilitation: state of the art and future prospects. Med Eng Phys. 2016;38(7):587–606. 133. Lum PS, Godfrey SB, Brokaw EB, Holley RJ, Nichols D. Robotic approaches for rehabilitation of hand function after stroke. Am J Phys Med Rehabil. 2012;91(11 Suppl 3):S242–54. 134. Bos RA, Haarman CJ, Stortelder T, Nizamis K, Herder JL, Stienen AH, et al. A structured overview of trends and technologies used in dynamic hand orthoses. J Neuroeng Rehabil. 2016;13(1):62. 135. Mehrholz J, Pohl M, Platz T, Kugler J, Elsner B. Electromechanical and robot- assisted arm training for improving activities of daily living, arm function, and arm muscle strength after stroke. Cochrane Database of Systematic Reviews 2015, Issue 11. Art. No.: CD006876. https://doi.org/10.1002/14651858.CD006876.pub4. 136. Veerbeek JM, Langbroek-Amersfoort AC, van Wegen EE, Meskers CG, Kwakkel G. Effects of robot-assisted therapy for the upper limb after stroke. Neurorehabil Neural Repair. 2017;31(2):107–21. 137. Diaz I, Gil JJ, Sanchez E. Lower-limb robotic rehabilitation: literature review and challenges. J Robot. 2011;2011:1-11. https://doi.org/10.1155/2011/759764. 138. Benito-Penalva J, Edwards DJ, Opisso E, Cortes M, Lopez-Blazquez R, Murillo N, et al. Gait training in human spinal cord injury using electromechanical systems: effect of device type and patient characteristics. Arch Phys Med Rehabil. 2012;93(3):404–12. 139. Nam KY, Kim HJ, Kwon BS, Park JW, Lee HJ, Yoo A. Robot-assisted gait training (Lokomat) improves walking function and activity in people with spinal cord injury: a systematic review. J Neuroeng Rehabil. 2017;14(1):24. 140. Louie DR, Eng JJ. Powered robotic exoskeletons in post-stroke rehabilitation of gait: a scoping review. J Neuroeng Rehabil. 2016;13(1):53.

Journal

Journal of NeuroEngineering and RehabilitationSpringer Journals

Published: Jun 5, 2018

References

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off