Converging Clinical and Engineering Research on Neurorehabilitation III

Stroke is the main cause of permanent and complex long-term disability in adults. RETRAINER S2 is a system able to recover and support person’s ability to perform Activities of Daily Living (ADL) in early stage after stroke. The system is based on exercises for hand and wrist performed using Neuro Muscular Electrical Stimulation (NMES). This work describes the preliminary results of a multi-center Randomized Controlled Trial (RCT) aimed at evaluating effectiveness of the system. The preliminary results were calculated on 18 patients who completed the protocol. Data is promising, the RETRANER S2 system seems to be a good tool for stroke rehabilitation. Data confirms also a general good usability of the system.

Collaborations between universities and industry are a driving force for innovation. While the benefits for both parties involved in such collaboration are numerous, so are the challenges that come with it. This paper shines a light on both – benefits and challenges - experienced in the H2020 project RETRAINER, a multi-national project in the strongly regulated field of medical devices.

Stroke is a leading cause of disability worldwide, and home-based solutions at low costs are required to provide effective rehabilitation to patients and decrease the economic burden. The interactive objects are one of the components of the RETRAINER system, and we exploited them to produce a stand-alone home rehabilitation tool. Patients own objects are equipped with RF sensitive tags and an antenna connected to a RF reader is placed on patient’s hand. Using the relation between the distance of RF sources and the RSSI the contact with the object is detected and the task completion is identified. A backend system autonomously guides the execution of the exercises designed by therapists and record the data about the performance of the execution. Preliminary results demonstrate the feasibility of the approach and show a good degree of acceptability by patients and clinicians.

This contribution describes wireless IMU- and EMG-sensors for the control of Functional Electrical Stimulation that have been developed within the European project RETRAINER. Combined IMU- and EMG-sensors shall be integrated into the RETRAINER exoskeleton (S1 system) for estimation of the arm posture and for EMG-triggered stimulation. IMU-sensors shall be used within the RETRAINER hand neuroprosthesis (S2 system) to estimate the finger and hand motion. Both sensor types are Bluetooth 5 ready and incorporate a powerful Cortex M4F processor. Data preprocessing, like orientation estimation and adaptive linear prediction EMG filtering for estimating residual volitional muscle activity, can be implemented directly on the wireless sensors. This reduces the data rate significantly down to the stimulation frequency.

Upper extremity exoskeletons are useful for humans in different ways: for motor rehabilitation, as assistive devices, or for the reduction of work-related loads on the musculoskeletal system. This paper describes the design of a passive modular and light-weight arm exoskeleton with gravity support and discusses possible fields of application. Tests, carried out with enabled gravity support show reduced muscle activations and forces compared to the same movements with disabled gravity support, indicting the effectiveness of the design.

The RETRAINER (Reaching and grasping Training based on Robotic hybrid AssIstance for Neurological patients: End users Real life evaluation) project is an Innovation Action funded by the European Commission under the H2020 research framework programme. The project aims at a full technology transfer of the results of a previous FP7 project, MUNDUS, aimed at the development of upper limb assistive technologies, to a robotic system for upper limb and hand rehabilitation to be tested in a wide clinical trial with stroke survivors in two clinical centers. The final result of the project is the design of a validated system suitable to address the rehabilitation market. Along this project’s path, several issues affecting both development and validation have been pointed out and are here summarized to serve as lesson learnt for prospective projects and challenges.

Restore a lost function is a special experience for people affected by neuromuscular evolutive diseases as muscular dystrophy. Upper limb stiffness and activity limitations have a crucial role in reducing patients’ autonomy and worsening quality of life. Even if the commercial products might assure a benefit to some users and meet most of their requirements, so far a validation of the use of such devices by people with neuromuscular diseases is missing. We aim at field-testing the improvement in arm functions provided by the use of two commercial devices (Jaeco Wrex and Armon Ayura) and assessing their impact on users’ quality of life and independence. This step is essential to assure a widespread access to these devices for most of the potential users, possibly presenting direction and guidance to health providers. The results acquired from the first three subjects, with a different disease progression, showed that the functional improvements gained with the use of these exoskeletons are limited and largely depends on the user’s impairment. Results showed that if the patient is severely impaired, the exoskeletons are not sufficient to gain functional movements. In contrast, if the patient is moderately impaired, both devices help the subject, even if some limitations of the movements occur. Finally, if the subject is slightly impaired, both devices decrease the performance. However, all the patients have appreciated the good usability of both devices.

Spinal cord stimulation (SCS) is a widely-used therapy for chronic pain management. Computational models provide a valuable tool to study the effects of SCS on the nervous system. However, it is critical that these models include sufficient detail to correlate model predictions with clinical effects, including patient-specific data. Therefore, in this study, we developed a patient-specific computer model for a patient undergoing SCS to treat neuropathic pain. The patient-specific model predicted sensory threshold estimates that were consistent with the corresponding clinical measurements. These results demonstrate the potential for patient-specific computer models to quantitatively describe the axonal response to SCS and to address scientific questions related to clinical SCS.

Spinal cord injury (SCI) disrupts the communication between the brain and spinal sensorimotor circuits below the lesion, leading to paralysis. Epidural electrical stimulation (EES) applied dorsally to the spinal cord modulates the activity of spared spinal circuits by supplying excitatory inputs via the direct recruitment of large myelinated afferent fibers running in posterior spinal roots. EES applied at the cervical level could promote upper-limb function after SCI, but its ability to engage specific arm and hand muscles remains largely unknown. Here we developed an anatomically realistic computational model to evaluate the influence of electrode positioning on the recruitment of cervical afferent fibers. Our results show that laterally-positioned electrode active sites recruit specific dorsal roots with higher selectivity than centrally-positioned active sites, opening a development path for efficient epidural electrode arrays tailored to the cervical cord.

Leg amputees suffer the lack of sensory feedback from a prosthesis, which is provoking their low confidence during the walking, falls and low mobility. Electrical peripheral nerve stimulation (ePNS) of upper-limb amputee’s residual nerves has shown the ability to restore the sensations from the missing limb into the proper sensorimotor scheme. Physiologically plausible stimulation protocols targeting lower limb sciatic nerve holds promise to induce sensory feedback restoration that should facilitate close-to-natural sensorimotor integration and therefore fall avoidance and walking corrections. The sciatic nerve, innervating the foot and lower leg, has very different dimensions and density/disposal of mechanoreceptors, respect to upper-limb nerves. Therefore there is a need to develop a computational model of its behaviour. Different types of neural interfaces and their different designs have been implemented. This computational modelling suggests the optimal interface to use in human subjects.

Mind controlled bionic limbs promise to replace mechanical function of lost biological extremities and restore amputees’ motor capacity. State of the art approaches use machine learning for establishing a mapping function between electromyography (EMG) and joint kinematics. However, current approaches require frequent recalibration with lack of robustness, thus providing control paradigms that are sensitive to external conditions. This paper presents an alternative method based on the authors’ recent findings. That is, a biomimetic decoder comprising a computational model that explicitly synthesizes the dynamics of the musculoskeletal system as controlled by EMG-derived neural activation signals.

The thalamus tends to be presented as a sensory gateway, a relay where little computation takes place. This view, combined with the difficulty of accessing this area related to cortex or peripheral nervous system, resulted in a relative paucity of thalamic computational models being detailed, complete and efficient enough to subserve neuroengineering applications. Here we present a novel model of thalamocortical interaction where both areas are simulated with adaptive integrate and fire spiking neurons. The model is able to reproduce information transfer from thalamus to cortex in both awake and asleep state, as shown by the local field potentials matching those observed experimentally in the two dynamics regimes. The applications in neuroengineering of such a simple and complete model range from simulations of sensory feedback injected directly in the thalamus for tetraplegic patients, to simulations of the effects on cortical activity of DBS stimulation delivered in the basal ganglia or directly in thalamus.

Despite the complex nature of human hands, neuroscientific studies suggested a simplified kinematic control underpinning motion generation, resulting in principal joint angle co-variation patterns, usually called postural hand synergies. Such a low dimensional description was observed in common grasping tasks, and was proven to be preserved also for grasps performed by exploiting the external environment (e.g., picking up a key by sliding it on a table). In this paper, we extend this analysis to the force domain. To do so, we performed experiments with six subjects, who were asked to grasp objects from a flat surface while force/torque measures were acquired at fingertip level through wearable sensors. The set of objects was chosen so that participants were forced to interact with the table to achieve a successful grasp. Principal component analysis was applied to force measurements to investigate the existence of co-variation schemes, i.e. a synergistic behavior. Results show that one principal component explains most of the hand force distribution. Applications to clinical assessment and robotic sensing are finally discussed.

State-of-the-art prosthetic hands allow separate control of all digits. Restoring natural hand use with these systems requires simultaneous and proportional control of all fingers. Regression algorithms might be able to predict any combination of degrees of freedom after training them separately. However, to the best of our knowledge, this has yet to be shown online. Twelve able-bodied participants were instructed to reach predefined target forces representing either single or combined finger presses, following a system training session consisting of only individual finger presses. Myoelectric control was implemented using linear ridge regression. The results demonstrated that myoelectric control allowed participants to reach both single finger, and combination targets, with hit rates of 88% and 54% respectively. These findings suggest that simultaneous control of multiple fingers is possible, even when these movements are not included in the training set.

State of art of modern hand prosthesis is populated by sophisticate hi-tech poly-articular hands which usually offer a broader set of movement capabilities, with the possibility to control up to 4 or 5 motors and achieve several different postures. Unfortunately these device are not so easy to control. A novel emerging trend is oriented towards a strong simplification of the mechanical design (through i.e. underactuation mechanisms), but still maintaining a good level of performance. A successful example is the SoftHand2 Pro, a 19 Degrees of Freedom (DoF) anthropomorphic hand which, using two motors, can move along two different synergistic directions, to perform either power grasp, precision grasp and index point. The combination of this multi-synergistic prosthetic hand with advanced controls, as myoelectric pattern recognition algorithms, allows to get promising results toward a more natural and intuitive control, introducing novel features as the possibility of a continuous switch between gestures. Preliminary experimental results are presented, demonstrating the effectiveness of the idea.

We have developed a soft prosthetic hand with features addressing the need of upper-limb amputees using soft robotics techniques. The designed hand is ultra-light and easy to manufacture. It is readily customisable for different hand size due to parameterised CAD design and using 3D printing techniques. The user-friendly control of the hand is achieved by using combination of designed-in behaviour of the finger and optimised movement of the fingers. This enables users to grasp a wide range of objects in one specific hand preshape eliminating the need for multiple switching between different grasps. The performance of the designed hand is evaluated through evaluation criteria used for prosthetic hands.

Stroke disrupts motor and sensory pathways, affects the ability to sense without distorsions body and peripersonal space, to decide to act, and to control efficiently the body. This paper describes the contextual requirements for the design of a grasp rehabilitation system for goal directed exercises. An implementation of a compatible system is shown.

This work presents the preliminary results of a randomized controlled trial aimed at evaluating the efficacy of a novel hybrid robotic system for arm rehabilitation after stroke. The system was developed within the European project RETRAINER and consists of a passive exoskeleton for weight relief combined with an arm EMG-triggered neuroprosthesis. Up to now, 39 patients completed the 9-week intervention: patients in the experimental group achieved a significantly better effect in the motoric outcome measures with respect to control subjects receiving only conventional therapy. These promising results need to be confirmed on a larger sample.

In this paper, we evaluate the performance of a low-cost surface electromyographic (sEMG) device in the classification of different hand-grip features. To that end, hand-pressure information has been measured together with the sEMG activity of the forearm during the performance of grasping activities with three different finger groups. Results show that affordable sEMG sensoring can accurately differentiate between different grasping poses and is also capable, at the same time, of decoding hand force levels.

This paper presents the progress towards the development of a sensorized transradial prosthesis designed and developed to be a reliable and easy to maintain prosthesis, aimed at sustaining long-term home studies. The hand builds around a transmission mechanism that implements a semi-independent actuation of the abduction/adduction of the thumb and of the flexion/extension of the index, by means of a single actuator. The new version of the hand includes three bi-axial force sensors in the thumb, index and middle fingers, respectively, allowing to measure the grasping and load forces at the fingertips. The mechanical transmission was revised increasing the grip force to 70 N (maximum single fingertip force) and still guaranteeing fast grasps (closing time ~500 ms). The current version of the prosthesis underwent durability tests and a preclinical evaluation involving one trans-radial amputee.

In this study we considered the role of the inhibitory interneurons known as Renshaw cells (RC) in the activity of a simulated locomotor neural network. We used a n integrate-and fire-model to reproduce RCs experimental three-phases responses, consisting of a fast activation, a relaxation time and a slow activation. Simulations of RCs within a model of muscle spindle reflex neural network highlighted multiple roles of Renshaw cells in locomotion. We found that RCs synchronize the pool of motor neurons (MNs) they act on, and regulate the relative duration of the antagonist muscle bursts during the gait cycle. This refined model can be used to simulate the interaction between electrodes and spinal circuits to improve the efficacy of spinal cord stimulation protocols.

Cervical spinal cord injury and stroke severely impact grasping movements required for activities of daily living. Intraneural peripheral nerve stimulation (PNS) enables specific activation of passing axons and may restore fine grasping in paralysed hands. Here, we assess the feasibility of using intrafascicular electrical stimulation of the arm peripheral nerves to produce precise hand movements in the non-human primate (NHP). Neuroanatomical studies of the NHP arm nerves, computational simulations based on realistic PNS models, and electrophysiology experiments show evidences for the use of intraneural PNS to induce fine hand movements.

Clinical studies have so far not been able to show if robotic therapy is superior to conventional methods. The personalization of robot-assisted therapy according to the individual motor deficits might contribute to reach this goal. Here we present a statistical approach to automatically personalize robotic rehabilitation. Our method uses different motor performance measures to estimate motor improvement and adapt the motor task within a treatment session. This approach was tested with a pilot sub-acute stroke patient and the outcome was compared to a similar patient who underwent conventional physical therapy. Pilot results showed better outcomes in clinical tests, kinematics and muscle activity for the subject who trained using the personalized robotic approach.

Partial hand amputations are the most common amputations worldwide, yet their prostheses, especially battery-powered ones, are only slowly progressing. As a result, only few clinical solutions are available. We developed a prototype of a powered prosthetic finger (dubbed S-Finger) that is equipped with a mechanical transmission alternative to the already available solutions which comprises of a miniaturized non-back drivable mechanism. Here we present the design criteria and the details of an optimized design that comprises of a non-backdrivable mechanism and a miniaturized Oldham joint.

Penetrating neural interfaces, connecting peripheral nerves to robotic devices (e.g., hand prostheses), could be inserted through tungsten needles, which are able to minimize damages and scarring due to the puncture wounds. Unfortunately, puncturing needles may fail independently on the material fracture toughness. In addition, independently on internal biotic causes, needles’ performances may decrease during in vivo trials. External biotic causes seems to be related to these effects, even if the exact genesis of phenomena, decreasing the in vivo reliability, is still partially unknown. Therefore, this work provides a hybrid computational approach, simultaneously using theoretical relationships and novel fast silico models of nerves. This framework is able to lower computational times needed to predict in vivo performances by using in vitro reliability and local differences between in vivo and in vitro mechanical response of nerves.

Current fitting procedures of powered partial hand prostheses heavily rely on manual “trial & error” approaches by skilled prosthetists without taking advantage of more modern engineering enabling technologies. Here we preliminary present an optimization method for the placement of the powered digits based on a biomechanical simplified model. The mathematical model is built taking into account the geometry of the residual limb, the functional conditions of the stump and the kinematics of the prosthesis. The optimization refers to the distance, on the palm plane, between the basements of two opposing digits. In particular it focused on the opposition capability of a prosthetic index digit with an unpaired thumb in a pinch grasp. The optimal posture is the one that allows the maximum control accuracy of the force along the direction of the grasp. We argue that the proposed method could be used to predict the quality of the fitting and to standardize the process.

This paper presents a novel mathematical approach to decode information from neural spike trains recorded using high-density surface EMG for the control of a robotic hand. During precision grasps characterised by low firing rates, the algorithm uses a point-process approach to extract the information encoded in the spike times. When performing power grasps, the increased firing rate makes the point-process less reliable and the algorithm automatically and gradually switches to an approach based on the EMG power. Unlike current point-process decoding algorithm relying on a Gaussian approximations of the observation model, this model adopts a fully Gaussian state-space representation thanks to a parametrization of the rate of an inverse Gaussian distribution and by modelling the inter-spike interval directly.

The level of physical stress rules the adaptative response of peripheral nerves, which is crucial to assess their physiological and pathological states. To this aim, in this work, a hybrid computational approach was presented to model the stress response of in vitro peripheral nerves undergoing longitudinal stretch. In addition, the rate of change of stiffness with strain was investigated and quantified through the secant modulus in order to achieve a quantitative framework, providing crucial information for both physical rehabilitation and biomaterial science.

In the state-of-the-art, only few studies analyzed volume changes in residual lower limbs. Most of them focused on transtibial amputees, while data on transfemoral ones are largely missing. This study aims at investigating how transfemoral stumps change over time. A measurement system based on a portable 3D optical scanning technique was used to measure volume fluctuations in 6 transfemoral amputees prior to and 20 min after physical activity. Such technique allowed to measure the stump volume once outside of the prosthetic socket. Investigating stump volume changes over time after prosthesis doffing is necessary to take into account this source of error during tests. Data show that volume fluctuations range from −1.25% to +3.09%. These values were enough to cause patient discomfort. This study highlights the need to properly characterize volume changes with the aim of designing more efficient socket solutions, based on such data.

Robot-Assisted Gait Training (RAGT) has gathered significant attention in the past years, nevertheless the results of its application on the stroke population are inconsistent. One of the reasons behind the mixed success of RAGT is believed to be its failure in promoting active participation from the patients. Herein is discussed the potential use of the Error Augmentation paradigm to RAGT. Error Augmentation is a training paradigm which utilize perturbations to amplify biomechanical errors (or their associated feedback). The augmented error leads to the generation of a motor adaptation that compensates the perturbation and, once the perturbation is removed, reduces the original error. So far the Error Augmentation paradigm has been applied to gait rehabilitation only in the split-belt treadmill paradigm, where it is used to induce compensation for altered step symmetry. The application of such paradigm to RAGT, although technically challenging, has the potential of increasing the gait parameters that can be targeted in a training modality that, by design, is highly patient-specific and demands the active participation of the patients during the therapy sessions.

Developing embedded systems tailored for resource-constrained platforms enables the design of robust frameworks for controlling artificial arms in prosthetic applications. This work presents preliminary results of the implementation of a novel platform for EMG-based gesture recognition application based on Hyper dimensional Computing (HDC), a novel brain-inspired classifier. HDC reaches classification accuracy comparable with traditional statistical learning algorithms, while its training phase is one order of magnitude faster, resulting suitable for the implementation on low-power and low-cost digital platforms. The proposed setup acquires EMG data from 8 sensors, performs training in 1.2 s on the embedded microcontroller and classifies 5 gestures with 88% accuracy, a latency of 10ms and energy consumption of just 0.65 mJ per classification.

The RETRAINER S1 system is an upper limb rehabilitation device designed to be used in repetitive task-oriented training. While the device itself is intrinsically controlled by the wearer, the execution of the training exercises is automatically controlled by a finite-state machine. This contribution discusses three different control strategies tested in a clinical environment.

Advances in neurophysiology are enabling understanding the neural processing underlying human movement, i.e. the recruitment of spinal motor neurons and the transmission of the resulting neural drive to the innervated muscle fibers. Similarly, advances in musculoskeletal modeling are enabling understanding movement mechanics at the level of muscle forces. However, despite detailed knowledge at the individual neural and musculoskeletal levels, our understanding of the neuro-mechanical interplay underlying movement is still limited. This paper presents recent techniques for probing the activity of spinal motor neuron pools as well as how this translates into musculoskeletal mechanical function. We then translate this in the context of robotic exoskeletons for establishing a class of human-machine interfaces that can open a window into human neuromuscular states. This represents an important step for the creation of symbiotic exoskeletons.

Robotic manipulators can be controlled in an autonomous way with great precision and dexterity. At the same time they can be equipped with sensors capable of conveying highly precise information on the surroundings, many times superior to that of a human sensory system. However, our limited capacity of interfacing these robots with the human body makes current prosthetic systems to be perceived by the users as simple tools rather than limbs. After decades of developments, osseointegration, selective nerve transfers, and nerve electrodes for sensory feedback have all been clinically tested in humans and are opening a new gateway for implementation of novel control strategies. Here, an overview of the most promising myocontrol and myoelectric signal processing technics to pave the way to longer-term visions of true limb replacement are presented.

Exosuits and exoskeletons have been shown to reduce the metabolic cost of walking by over 20%. While many studies have shown promising group-averaged data, there has often been high variability in individual response to assistance even when using standardized procedures and device platforms. Here, we aim to investigate the biomechanical differences in adaptation to exosuit assistance, and present initial data from seven subjects that show differences between the joint kinematics and kinetics, and muscle activations of individuals who display small vs large metabolic benefits. Future work will focus on expanding this work to understand general principles for predicting metabolic benefit and understanding sensorimotor adaptations over longer time-scales. We believe this effort will be instrumental towards understanding the potential for exosuit and exoskeleton assistance in gait augmentation and in clinical settings, for both rehabilitative and assistive applications.

Measuring the force exerted by patients in the exercise for rehabilitation after neurological injuries is important: in quantifying the patient’s motion capabilities, to ensure safety and to provide the appropriate amount of assistance, among others. Adding a force sensor for this purpose at the end-effector of a rehabilitation robot can add considerable cost. When a robotic device is dynamically transparent and mechanically backdrivable, a force estimator based on the model of the system can be used to estimate the force applied by the patient without using the explicit force sensor. This work validates the effectiveness of a model-based force estimator, derived from the literature, within the context of rehabilitation robotics, through a successful validation the strategy on the EMU upper-limb rehabilitation robot.

In this paper we aim at demonstrating the feasibility of a bilateral training protocol supported by a dual arm exoskeleton, ALEx-RS. Three subjects performed the same reaching task in three conditions: with and without the exoskeleton, and in the bilateral mode. EMG data recorded during the experimental sessions showed that ALEx-RS features a high level of transparency and all subjects easily accomplished the bilateral task in an intuitive way. These findings demonstrated the suitability of ALEx-RS for being applied in a rehabilitation protocol in which also bilateral strategies are employed.

The objective of this study is to evaluate the different timing of upper limb motor recovery in subacute and chronic stroke patients by using kinematic parameters combined with clinical evaluations.Twenty-five subacute and twenty-three chronic stroke patients participated in this study. The Chedoke-McMaster Stroke Assessment Scale, Modified Ashworth Scale, Fugl-Meyer Scale and Motricity Index were used as clinical outcome measures. The mean speed, maximum speed, number of speed peaks, normalized reaching speed, jerk metric, acceleration metric and mean distance were computed.Significant improvements were found in both groups in Fugl-Meyer scores. Motricity Index scores improved significantly only in the subacute group. Significant changes were found in all kinematic parameters in the subacute group. In the chronic group significant improvements were found in the mean and maximum speed, and in normalized reaching speed values.This study confirms that the upper limb robot-assisted therapy is effective both in subacute and chronic stage after stroke.

This paper proposes the AI Supervisor which controls the treadmill speed effectively like an expert such as personal trainer or physical therapist based on real-time sensor data and physical information on the user, and AI decision making. It makes a decision to control the speed of a treadmill during exercise or rehabilitation by measuring the heart rate. The decision is processed by the Deep Neural Network (DNN) with a dataset of 8 people, the accurate decision rate is 94.6%.

Robot-assisted exercises often use controllers which automatically adapt task parameters to the user’s performance. One problem with these controllers is how to accommodate the wide variety of degrees of impairment and to properly track the user’s improvement in spite of the inherent variability of performance that is typical of these tasks. Here we describe an adaptive controller model which uses reinforcement learning to maintain a model of user’s performance and uses it to continuously regulate the task parameter. We show that the model rapidly identifies the user’s model parameters and then smoothly tracks performance improvements.

In this work we present a bilateral rehabilitation system for the hand based on a novel underactuated hand exoskeleton to assist hand opening/closing and a pair of pressure-sensorized graspable objects. In particular the novel hand exoskeleton provides self-adaptability to different hand sizes and a more effective transmission of forces. In the system, the grasping force measured at the healthy side is used to modulate the assistance of the hand exoskeleton at the impaired side. Utilizing simpler and more reliable pressure-sensorized graspable objects instead of biosignals might allow the system to adapt to residual motor capabilities of the impaired hand. System performance of the robotic assisted, bilateral grasping tasks have been experimented with 3 healthy subjects.

Motivation is an important topic in rehabilitation and frequently used as a determinant of rehabilitation outcome. Several factors can influence patient motivation and so improve exercise adherence. This paper presents some techniques to improve patient motivation and maximize outcome during technology-assisted rehabilitation. In particular, we present some examples of multimodal augmented feedback. This approach provide feedback on the performance during training through the stimulation of different sensory channels such as vision, audition, proprioception etc. It is believed to have several advantages over unimodal feedback and it can promote the patient’s motivation and maximize outcome.

We propose a system for the EEG-EOG-EMG recording in stroke subjects during robotic rehabilitation task. This system was designed for obtaining safe recording conditions, high-quality data, triggering signals to track the task and to align EEG segments to motor performance, friendly visualization and management of the data during the signal acquisition and subsequent analysis.We recorded EEG data from a stroke subject during resting state and robotic rehabilitation task before and after a program of 30 sessions.Results showed high-quality EEG data recorded in the 4 patients with about 80% of artifact-free EEG epochs during robotic performance.Globally, the relatively high percentage of artifact-free EEG epochs represents a good first index of the quality of the EEG recordings. Furthermore, the analysis of EEG power density spectrum revealed typical features of human cortical EEG oscillatory activity during resting state and engaging events.

Proprioception, the sense of body position, is essential for movement planning and postural control. However, its clinical assessment after stroke is generally rudimentary, limiting the design of personalized therapy regimes. In this study, we analyze the sensitivity of robotic metrics in comparison to clinical scales for proprioceptive assessment. Thirteen patients were screened with the Erasmus MC Nottingham Sensory Assessment and with a passive Joint Position Matching task by means of a wrist robot at hospital admission, discharge and 1-month follow-up. While the clinical test did not show significant differences across sessions, robotic metrics revealed that proprioceptive accuracy improved during the rehabilitation period, while improvements in proprioceptive precision required a longer time to occur and were noticeable in the follow-up session.

The integrity of muscle synergies patterns has been proposed as a physiological marker of cortical damage but how to modify and train muscle synergies to relearn movement is still an open question. Here we present our recent results about the modifications that the forces induced by robots can cause on upper limb muscle synergies after stroke. Our results show that a single exposure to forces provided by a robotic device can impact muscle synergies activation. Moreover, a prolonged robot-aided motor training can promote an enduring longitudinal reorganization of upper limb muscle synergies. Finally, we discuss the application of muscle synergies in neurorehabilitation to both assess the effectiveness of treatments and design novel protocols.

Robot-assisted rehabilitation has emerged as an advantageous tool in the recovery of patients suffering from abnormal motor conditions. The assessment of the electrophysiological behavior of the muscles involved in this rehabilitation process can provide important hints of the patient functional recovery. Such analysis has been commonly performed through the processing of bipolar surface electromyography (sEMG). Here we introduce recent techniques based on the acquisition of high-density electromyography (HD-sEMG) to be used as a tool for the assessment of robot-assisted rehabilitation. To that end, the activity of the tibialis anterior was measured during the performance of periodical ankle dorsiflexion with different frequencies and amplitudes. Intramuscular motor unit coherence was then computed showing peaks at the frequency of movement (delta band) but more interestingly at the beta band. These changes in amplitude and position of the coherence peaks may provide meaningful metrics for gait rehabilitation procedures where the muscle contraction is as well periodical and variable in strength and speed.

Neurorehabilitation may greatly benefit from computational approaches. We present a computational model of the trial-by-trial dynamics of recovery through task-specific, robot-assisted exercise. The model assumes that recovery is driven by movement performance. The model explicitly addresses the extent to which training in one direction affects performance of subsequent movements in other directions. We fitted the model to data from a rehabilitation trial based on a task-specific exercise, involving reaching movements to different directions. The model reproduces the trial-by-trial speed and smoothness time series. These findings suggest that the model can be used to interpret the evolution of performance, and to formulate testable hypotheses on the recovery mechanisms, at the individual subjects’ level. Therefore, it can be used to adaptively customize the robot-aided exercise based on each patient’s direction-specific impairment.

Most of hemiplegic subjects need foot drop correction in order to decrease the risk of falling during walking. Since compensatory movements such as circumduction and steppage gait are considered to be developed to achieve foot clearance in the swing phase, such movements might be improved if enough ankle dorsiflexion is developed in the swing phase. In this paper, hip joint and ankle joint angles of a hemiplegic subject were measured during walking with inertial sensors, when FES foot drop correction was applied. As the ankle dorsiflexion angle decreased by muscle fatigue, hip flexion angle begun to increase earlier and timings of the heel off and the toe off were detected earlier, which were similar to gait movement of the subject without FES foot drop correction. These changes were considered to be a preparatory movement for achieving foot clearance by increasing hip flexion. This result suggested that the compensatory movements by the hip joint could be improved by FES foot drop correction.

We developed a myoelectric hand prosthesis equipped with the tacit learning system (TLS) to auto-regulate forearm rotation in response to upper extremity movement patterns. We evaluated the tacit learning effects on the central nervous system during a prosthesis control exercise. The experienced prosthetic user performed a series of simple mechanical tasks with the system inactivated (the baseline, normal condition) and then with it activated (the enhanced, experimental condition). The process was video recorded. Video 1 documents the ordinary TLS-inactive condition while Video 2 documents the TLS-active condition. Subsequently, the participant viewed each video under experimental conditions during magnetoencephalography and electroencephalography recording. The connections between the motor area and the other cortical areas were observed with a significant increase in the coherence value. Results suggest integration and interoperability as the process undergirding tacit learning promotes motor function related adaptive neuroplasticity.

Rehabilitative devices promote functional improvement of post-stroke patients through movements and tasks, thereby retraining the patient. Augmented reality (AR) is playing a vital role in this area through numerous novel interfaces. Various techniques and interactive technologies have been studied in clinical rehabilitation trails. However, there are insistent needs to design interactive home-based games to support patients’ motivation and their rehabilitation needs in home settings. This paper presents a framework that can be implemented to design AR games-based rehabilitative system and improve the patients biofeedback.

We investigate the use of submaximal force production as a targeted functionality in a rehabilitation task for early rehabilitation after stroke. We present the detailed assessment of related metrics of force production and position control, their correlation with submaximal force production control learning and their feasibility for robot–mediated therapy after stroke.

Technological innovations are allowing rehabilitation to move toward more integrated processes, with improved efficiency and less long-term impairments. In particular, robot-mediated neurorehabilitation is a rapidly advancing field, which uses robotic systems to define new methods for treating neurological injuries, especially stroke. The use of robots in motor recovery can enhance rehabilitation, but it needs to be used according to well-defined neuroscientific principles. The field of robot-mediated neurorehabilitation brings challenges to both bioengineering and clinical practice (motor and cognitive). The aim of this scoping review is to better understand the role of the technology into application of rehabilitation principle, overcoming the dichotomy between traditional and technological intervention.

Despite a proliferation of new rehabilitation technologies and a developing body of clinical evidence supporting their efficacy over the past 15 years, the uptake of rehabilitation technologies into clinical practice has been poor. In this qualitative study, we interviewed twenty seven rehabilitation clinicians to explore what helps and hinders the uptake of rehabilitation technologies into clinical practice. The findings illustrate that three inter-related factors are at play; the extent to which the technology addresses a clinical need, its clinical effectiveness and its usability. However, the interaction and weighting of these factors is influenced by the clinicians’ beliefs about the purpose of rehabilitation and what constitutes ‘valid’ rehabilitation practice, perceived professional roles, and whose perspective is given priority when evaluating technology. Understanding the context, process and values which drive decision-making may support the development of technologies which more readily translate into rehabilitation practice.

In recent years it has become evident that, in a number of functional movements, synergistically acting limbs become task-specifically linked by a soft-wired ‘neural coupling’ mechanism (e.g., the legs during balancing, the arms and legs during gait [1, 2] and both arms during cooperative hand movements [3]). Experimentally this mechanism became evident by the analysis of reflex responses as a marker for a neural coupling. It is reflected by the task-specific appearance of reflex EMG responses to non-noxious nerve stimulation, not only in muscles of the stimulated limb, but also, with same long latency, in muscles of meaningful coupled (contralateral) limb(s). After a stroke, nerve stimulation of the unaffected limb during such cooperative tasks is followed by EMG responses in muscles of the (contralateral) coupled affected limb, i.e. unaffected motor centres support synergistically acting movements of the paretic limb. In contrast, following stimulation of the affected limb, no contralateral responses appear due to defective processing of afferent input [4]. As a consequence, it may be therapeutically possible to strengthen the influence of unaffected motor centres on the performance of affected limb movements through training of cooperative limb movements required during activities of daily living.

Falls are acknowledged as one of the most serious causes of health problems in the elderly population. Thus, there is an urgent need to develop novel technological solutions to counteract the balance loss and prevent falls. In this pilot study, two healthy older volunteers were enrolled to investigate the effectiveness of different assistive torque patterns provided by a pelvis exoskeleton in promoting balance recovery after slippages. Our results revealed that the best strategy to counteract the balance loss after a slippage consists in slowing the slipping foot and minimizing the downward movement of the center of mass by providing opposite torque patterns between unperturbed and perturbed hips (i.e., flex/extensor assistive torques, respectively).

The purpose of this work was to analyze the variation of some gait descriptors when additional tasks (motor and cognitive with low and high load levels) were performed during walking, in 10 young healthy subjects. From one inertial sensor located at lumbar level, spatio-temporal parameters and measures of acceleration and jerk root mean square were extracted. The latter were normalized to make them independent from gait speed variations occurred during the different trials. Statistical analysis showed significant differences on the most of parameters between single and high cognitive level walking: the variations in acceleration-related measures were associated to a worsen in the walking balance, and those reported in jerk ones to a change in the smoothness, leading to an alteration in gait stability. The other two conditions presented just a trend, maybe due to the too low complexity of the tasks for the chosen population.

Gait disorders are major problems among patients with cerebellar ataxia (CA) and Parkinson’s disease (PD). Little is known on how damages to the cerebellum or the basal ganglia affect the recovery responses to external gait perturbations. Six patients with CA, six patients with PD and six healthy controls (HC) were exposed to 9 blocks of 8 antero-posterior (AP) and medio-lateral (ML) perturbations. AP and ML Base of support (BoS) were compared between groups at perturbation onset and the following five heel strikes. HCs showed longer AP BoS than the CA and PD groups. Patients with CA showed a wider ML BoS than the HC and PD groups. Patients with PD were unable to take longer or larger steps when necessary (i.e., in reaction to anterior and medial perturbations). These findings could help to design specific perturbation-based training in patients with different degenerative neurological diseases.

The ankle joint muscles can contribute to balance during walking by modulating the center of pressure and ground reaction forces through an ankle moment. This is especially effective in the sagittal plane through ankle plantar- or dorsiflexion. If the ankle joints would be physically blocked to make an ankle strategy ineffective, there would be no functional contribution of these muscles to balance during walking, nor would these muscles generate afferent output regarding ankle joint rotation. Consequently, ankle muscle activation for the purpose of balance control would be expected to disappear.To investigate human balance control, we have performed an experiment in which subjects received anteroposterior pelvis perturbations during walking while the ankle joints could not contribute to the balance recovery. The latter was realized by physically blocking the ankle joints through a pair of modified ankle-foot orthoses. Here, we present the lower-limb muscle activity responses in reaction to these perturbations. Of particular interest are the tibialis anterior and gastrocnemius medialis muscles, as these could not contribute to the balance recovery through the ankle joint, nor encode muscle length changes caused by ankle joint rotation. Yet, these muscles showed long-latency responses approximately 100 ms after perturbation onset, even though the ankle joints were blocked. The response amplitudes were dependent on the perturbation magnitude and direction, as well as the state of the leg. The results suggest a centralized regulation of balance control involving supra-spinal neural structures, without the need for changes in ankle muscle proprioceptive information.

Turning difficulty is a sign of balance instability and may be indicative of elevated fall risk. Features extracted from the 90° turn suggest that this turn type is the most unstable type of turn in older adults with compromised balance control. Since the 90° turn is also the most common type of turn executed during activities of daily living, we recommend targeting movement strategies specific to 90° turning during therapeutic intervention. Specific neuro-rehabilitation strategies to improve/optimize turning ability in individuals with compromised stability may significantly contribute to fall-risk reduction. The adoption of quantitative tools for the assessment and monitoring of turning quality is advisable.

This paper presents a preliminary study on the effectiveness of using a virtual environment for automation of the Fugl-Meyer Assessment (FMA) scoring. Also, a study of the feasibility of introducing a smoothness parameter to improve the scoring is described. A Kinect-based virtual environment was used for data acquisition and to enable a friendly method for the test administration. Movement smoothness was analyzed using the Spectral Arc length (SPARC).

Robotic gait training is a promising tool to improve walking ability after neurological disorders. Despite its benefits, its therapeutic effects might depend on the customization of the robotic training to the patient’s capabilities. In the last years, various approaches focused on the assist-as-needed (AAN) paradigm, have been proposed to improve the effectiveness of robotic rehabilitation. However, in most of these cases the amount of robotic assistance is chosen by a therapist, and therefore, subjective decisions could affect the adaptation of the therapy to the patient’s needs. This contribution presents the implementation of a novel performance-based adaptive controller, which automatically adjusts robotic assistance for diverse subtasks of gait based on the user’s performance during training. A pilot study testing this controller in one stroke survivor, shows the potential of the proposed tool to be included in future robot-based gait training protocols.

Patient progress in rehabilitation after stroke is measured with standard clinical assessments. In questionnaires or test settings a therapist encourages the patient to perform a set of tasks. These tasks typically do not match what the patient does in daily life. To measure the upper limb function in daily life movements, a sensor-based system has been developed. The study investigates, with a full body sensor-based system the difference between standard clinical assessments and daily life monitoring. Four stroke patients were included in the analyses. A change in arm use during rehabilitation and the difference between clinical assessments and daily life measures were observed rendering the latter as potentially more sensitive candidates for outcome measures.

Following stroke, altered gait biomechanics contribute to inefficient walking. Aside from assisting the ankle, there is also potential in augmenting locomotion by actuating the hip. We describe a proof-of-concept wearable, soft, hip exosuit, that assists with paretic hip flexion during late stance and swing phases. The exosuit allows for adjustment of force and timing parameters. Parameter tuning was guided by a physical therapist based on their observations during study sessions. In a sample of survivors of stroke, we demonstrated that the hip exosuit was able to deliver low, comfortable levels of hip flexion assistive forces. When powered, the exosuit effectively increased thigh angle and angular velocity during paretic swing phase compared to unpowered condition. These preliminary findings highlight the feasibility for clinicians to use a hip exosuit to directly target deficits in limb advancement in hemiparetic gait.

Nonlinear stiffness actuator is a key point in the research field of neurorehabilitation robot. An actuator with good nonlinear stiffness characteristics can simulate the natural features of human interaction with the outside environment better, so as to achieve better rehabilitation training effect. In this paper, a novel nonlinear stiffness actuator was designed, mainly including two rollers, two couples of customized concave blocks with opposite directions, partial gear train, and a central torsion spring. Thereinto, the profile curve of customized concave block was mapped by a predefined deflection-torque trajectory with the natural interaction feature. And the central torsion spring was directly obtained from the existing products. Through this novel actuator, the human-machine interaction characteristics of “low load, low stiffness and high load, high stiffness” [1] can be realized.

Neuromotor training program are recommended as part of essential activity programs for older adults. With the growth of wearable sensor technology, wearable sensor provides a low-cost monitoring method for objective exercise assessment. Customized remote sensors combination with off-the-shelf smart phone provides a high performance and low-cost monitoring platform. A connection-oriented sensor network with Bluetooth Low Energy (BLE) provides high bandwidth encrypted communication. With multiple remotes sensors, synchronization between sensors is fundamental. However, the accuracy with the generic BLE time service is limited. In this paper, we present a novel method that can potentially synchronize multiple sensors with an unmodified Android Device. The method is based on timestamping the Connection Event (Anchor Point) on the Slave side. The results show an absolute error is below 1 ms across two sensors, which is ideal for motion sensing.

Soft wearable robots promise to be the new frontier for assistance and augmentation of human motor abilities. In this work, we present the design, controller and a preliminary assessment of a soft, textile based glove for assisting hand movements. The device is shown to reduce the muscular effort required for grasping an object in healthy subjects, for forces up to 25 N but slows hand movements in free space.

Cycling induced by Functional electrical stimulation (FES) was proposed as a method to improve motor recovery after stroke. This study aimed at assessing the effects of this training on motor patterns of muscle co-activations, referred as muscle synergies. Seven post-acute patients underwent 15 sessions of FES-cycling training. Outcome measures were Motricity Index and gait speed collected before (T0) and after (T1) the intervention. Subjects were also involved in a cycling test during which four muscle synergies were extracted from 9 lower limb muscles and compared with a healthy control group. Results showed a significant improvement in terms of gait speed, which increased from a median value of 81.0 cm/s at T0 to 100.0 cm/s at T1. Trends of improvement in both spatial and temporal components of muscle synergies were also achieved. More data are needed to drive final conclusions about the effects of FES-cycling on muscle synergies but first results are promising.

To design an underactuated wrist prosthesis, a preliminary study has been conducted to identify the relationship between the Degrees of Freedom (DoFs) of the wrist during the execution of tasks of daily living. After the identification of the principal orientations of the wrist describing the tasks, polynomial functions were used to define a synergetic relationship between the DoFs. The latter can be implemented in a prosthetic wrist featuring one actuator to obtain motion along three DoFs, with the purpose of reducing compensatory movements.

Although human movements are extremely complex, our nervous system is able to implement effective control strategies, leveraging on a generalized simplification approach. Several works described this behavior within the framework of synergies, which can be regarded as basis ingredients for motion generation through dimensionality reduction. Focusing on hand kinematics, this concept allowed to dramatically improve our understanding of the neuro-physiology of hand motor system, offering effective mathematical tools to identify pathological deviations from the physiological case. At the same time, these observations have found a fertile application field in robotics, suggesting simple yet effective manners to design and control artificial systems, with a reduced number of actuators or inputs. However, while much has been said about kinematic hand synergies and their implications for engineering, there are still open issues to tackle. Solving these issues could give better insights on the synergistic organization embedded within the human body, finally impacting the future development of robotic devices. In this paper, we will explicitly focus on the role that hand postural synergies play for grasp force control, and on preliminary observations on a synergy-based organization for upper limb motion generation. Applications of these neuroscientific findings for devising a principled simplification approach in assistive and rehabilitation robotics are finally discussed.

Diseases affecting human sensorimotor system are usually characterized by abnormal patterns of muscle activation. Therefore, their monitoring can evaluate and improve motor recovery. In developing new platforms for robot-aided rehabilitation, it is fundamental to understand if the platform itself negatively influence muscular patterns. This paper aims at verifying if a new developed end-effector robotic platform provided with an arm support and based on an assistance-as-needed approach alters patients movements in terms of muscle synergies. Preliminary tests have been carried out on eight healthy subjects performing 16 point-to-point movements. Muscular synergies have been extracted using the non-negative matrix factorization algorithm. Experimental results demonstrated that the proposed platform does not produce significant variations in muscular activation patterns ( $$p>0.05$$ ).

We discuss the strategies that the motor system may adopt while learning to control a body-machine interface (BMI) and how a novel approach to modulate the parameters of the BMI could be effective in facilitating learning of an appropriate feedforward control strategy. The same mechanism could potentially be used for teaching different control strategies to the user.

Pedobarographic data is valuable in understanding locomotor function in both research and clinical settings. However, commercially available software do not enable easy customisation of data processing and presentation. We have developed a toolbox for processing plantar pressure data using the cross-platform MATLAB programming language. The main features of the toolbox include data visualisation, sensor masking, time series, center of pressure and multi-segment analysis along with report generation. The toolbox can be used with an intuitive graphical user interface (GUI) without working with the underlying code. However, a functional approach to code implementation ensures that the toolbox can be used as a set of independent functions and new functions can easily be added. The code of the toolbox is open-source under the MIT License and available online at https://github.com/GallVp/footPress .

In this paper, we present a method and software that measure the trunk movement of the patients during the robot-mediated rehabilitation for upper limbs. Usually, people with reduced mobility such as stroke patients unintentionally use their trunk to satisfy goals of trainings (e.g. reaching specific points), which reduces outcomes of the rehabilitation therapies. Hence it’s important to monitor trunk movement during trainings. The method utilizes two IMUs (Inertial Measurement Unit) placed in back and affected upper arm. The developed software displays the trunk movement in 3D avatar and plots while providing the estimated percentage of trunk involvement in therapy. To evaluate the method and software, they have been implemented in the UHP upper limb rehabilitation robot and tested through reaching exercises with a group of 25 healthy people. The results show that the method works properly in measuring trunk motion during training.

Visual feedback is a common method to enhance motor learning and is widely used in rehabilitation institutes. Although balance training with visual feedback has shown positive results, the most effective way to provide the feedback is still open to research. We investigated the effect of visual distortion during end-point excursion training on an unstable surface with healthy participants. We demonstrated that visually reduced amplitude of motion during a training period can challenge participants to increase their maximal end-point excursion and has the potential to decrease the visual feedback dependency.

The upper limb function is affected in more than 50% of spinal cord. The upper limb motor recovery can be promoted by providing high intensity, repetitive and task-orientated training by means of using virtual reality applications and motion capture technologies. The objective of the present study is to present the preliminary development of two serious games for rehabilitation purpose of patients who have suffered tetraplegia. The technology used for registering hands movements is the Leap Motion Controller. The results on 8 people (4 healthy and 4 tetraplegic patients) have shown that the performance of patients is minor than healthy people, reaching a less number of goals. As conclusion these versions of both serious games are feasible in the rehabilitation of gross manual dexterity and the pinching motor skills of the fingers in tetraplegic patients.

In this paper, we investigate the possibility of a machine-learning algorithm using the Support Victor Machine Regression (SVMR) to predict the motor functional recovery of moderate post stroke patients during their rehabilitation program. To train the model, we used the recorded electromyography (EMG) signals from the upper limb muscles of the patients during their initial rehabilitation sessions. Then we tested the trained model to predict the later muscles performance of the patient during the same sessions. The results of this pilot study were promising; data were, to some extent, predictable. We believe such research direction could be essential to motivate the patient to complete the designed rehabilitation program and can assist the therapist to innovate proper rehabilitation menu for individual patients.

The soft extra muscle (SEM) Glove is a robotic hand device with the potential to serve as an orthotic and rehabilitation tool for people with spinal cord injury (SCI). In this longitudinal study 15 participants with a high neurological level (C2–C8) chronic SCI will be selected to trial the SEM Glove for 12 weeks at home to assist with activities of daily living (ADL). The participants will be scheduled to attend assessment sessions at Week 0 (Initial), Week 6, Week 12 and Week 18. Here we present progress with results related to grip strength and the Toronto Rehabilitation Institute hand function test (TRI-HFT) from eight participants who have completed the study. There was significant improvement in gross hand grip strength between Initial and Week 12 (p = 0.045). The TRI-HFT revealed significant improvement in hand function between Initial and, Week 6 (p ~ 0.01), Week 12 (p ~ 0.02). These results suggest that home use of devices designed to improve grip function has a direct effect on hand function.

A control strategy based on the application of the information on intention of movement, obtained through sEMG, to a Direct Force Control algorithm, for the bio-cooperative control of an exoskeleton for hand rehabilitation, is presented. The aim is to implement the assist-as-need paradigm and to provide the appropriate assistance to the patient while performing therapies to restore the capacity of opening and closing the hand. The results obtained from normalization and feature extraction of sEMG signals are satisfactory and will allow us to advance on the implementation of the designed controller.

In recent years, as a result of multidisciplinary studies in textile science and technology, sensor characteristics are added to the textile products. Thanks to the many features of e-textile products, motion monitoring can be done in novel ways. The prototype developed in this paper, is low-cost, low-power, easy-to-integrate, and fully capable of communicating with mobile devices in a wireless manner. The proposed sensor system can be used to monitor the healing process of patients with neurological injuries.

With the increase of rehabilitation costs in the last years, tele-rehabilitation systems are becoming popular. These systems allow patients to exercise at home, without the need to commute to a rehabilitation center. In this paper we introduce a web-based platform to support and manage the entire cycle of a rehabilitation program that can be carried out autonomously at home. In particular, the platform lets therapists schedule a set of exercises and tune their difficulty level, according to patient’s residual capabilities. Moreover, it provides clinicians with the trend over time of specific indexes (accuracy, speed, amplitude, strength), computed over the whole pool of exercises carried out daily. Therapists may review patients’ activities through an avatar animated with previously acquired motion data. The platform is currently being tested on patients suffering from multiple sclerosis; however, it can be used also in other rehabilitation domains.

An experimental protocol that integrates psychological and physiological assessment of volunteers was designed and developed to assess the degree of user acceptance of an exoskeleton. The main goal of this study was to determine how to increase the efficiency of employing these devices to improve the users’ social re-integration.

Gait and balance deficits are major impairments for Charcot-Marie-Tooth 1A (CMT1A) patients. However, motor rehabilitation is a challenge in CMT1A patients. There are poor evidences of clinical efficacy and low responsive outcome measures. Instrumented assessment (like inertial sensors) might be new tools to detect patient’s changes. A small sample of CMT1A was assessed before and after the rehabilitation period. Traditional and instrumented assessment measures of gait and balance were used. Results shown that only six-minute walking test showed an improvement after rehabilitation, in all patients. The Instrumented Timed Up and Go subtasks duration did not show responsiveness in CMT1A patients.

Effectiveness of robotic gait training in rehabilitation of stroke patients remains inconclusive. A reason could be that the current robotic gait trainers do not initiate motor learning principles enough. To encourage active participation of the patient and therefore motor learning, assist-as-needed (AAN) support strategies have been implemented in the robotic gait trainer LOPESII. Aim of the current study was to examine the effect of assist-as-needed support on metabolic cost.Ten chronic stroke patients completed three 6-min walking trials in LOPESII, with zero support, AAN-support for stiff knee gait and complete-support. Metabolic parameters were measured and compared between support conditions.No significant differences in net metabolic power were observed between zero-support, AAN-support and complete support.No evidence was found that AAN-support asks a higher metabolic cost of the participant.

Fall and related injuries are one of the main health problems in people with balance disorders. Thus, the assessment of gait stability is extremely important in the field of rehabilitation. Among the many stability indicators, the Margin of Stability (MoS) could be a good candidate for clinical employment, considering its reliability also on short datasets. However, its correlation with gait parameters is still partially investigated. In this study, the relationship between the MoS and the walking speed was investigated while subjects walked on a treadmill at different speeds. As expected, results revealed that gait speed affected the MoS in the plane of progression: the higher the walking speed, the higher (absolute values) the MoS. Accordingly, in future studies, this metric could provide quantitative measure of stability during activities of daily living, as well, by using inertial motion units.

We present a novel mobile robotic platform for overground gait rehabilitation post-stroke. In addition to its primary role in preventing falls, our device combines two therapeutic interventions: assistance or resistance along the patient’s direction of walking, and partial body weight support. The robotic therapy is complemented by functional electrical stimulation (FES) to stimulate the ankle dorsiflexors during swing phase. Real-time gait event detection using inertial measurement units is performed in order to drive the FES. We discuss the design of the robotic platform and its control methods, and also present preliminary results from an experimental study involving stroke patients receiving gait training with our robotic system.

Prompt detection of movement intention is fundamental to increase the reliability in prostheses control and increase the effectiveness of robotic devices for rehabilitation. A machine learning approach is proposed here to perform activity recognition (overground walking vs rest) using muscles or brain activity. EMG and EEG signals were preprocessed, features in the frequency domain were extracted and the achieved decoding accuracy was around 90%. Although extensive validation is still required, the results constitute a first step towards the goal of predicting gait initiation in real life.

This work presents the evaluation of a new control system of Functional Electrical Stimulation (FES) for applying strategies for artificial activation of the ankle joint muscles. In particular, we present an investigation of the direct effects of event-driven open loop FES assistance of the ankle muscles during walking on the lower limb kinematics in able-bodied individuals (n = 5) and one incomplete spinal cord injured (iSCI) patient. We analyzed gait kinematics between walking with and without FES assistance. Effects consistent with normal physiological gait were obtained, but some unexpected alterations also occurred. Therefore, further studies are needed to help adjust the FES strategy.

The use of an active knee orthosis can cause postural instability in hemiparetic users due to its weight and the uncompensated dynamics of the paralyzed limb. Here, we have proposed a gait assistance system that combines a haptic cane and an active knee orthosis for post-stroke over-ground gait training. The haptic cane is composed of a tilted conventional cane mounted on a motorized wheel, and the active knee orthosis device is based on a novel kinematic design. The proposed system can provide postural stability through proprioceptive augmentation, while providing support to the affected lower limb. To identify the effects and potential implications of the proposed gait assistance system, a pilot study was performed with one young healthy subject wearing an ankle weight (4 kg) on one leg. Improvements were observed in mediolateral and anteroposterior trunk sway. The proposed gait assistance system exhibited promising outcomes in this pilot study. The translation of this enhanced gait stability to hemiparetic stroke suffering individuals is warranted.

In traditional rehabilitation a therapist is observing and modifying various kinds of states of a patient. Therefore, to develop a system to help with tasks of therapist, we are facing a complex problem: how to integrate the various devices with diverse inputs and outputs to function as one system? In this paper we represent a systematic approach to build neuro-rehabilitation systems with an integration framework. We discuss a system design issues related to performance and the demanding requirements of rehabilitation. The introduced framework is available for download at: https://github.com/riken-ibcu/bclab .

This paper presents a user study to evaluate the system’s performance by measuring objective indicators and subjective perception between the two versions of a planar rehabilitation robotic device: (i) PupArm system, called RoboTherapist 2D system for commercial purpose, designed and developed for clinical settings; and (ii) Homerehab system, developed for home use. Homerehab system is a home rehabilitation robotic platform developed inside the EU HOMEREHAB-Echord++ project framework. Nine patients with different neurological disorders participate in the study. Based on the analysis of subjective assessments of usability and the data acquired objectively by the robotic devices, we can conclude that the performance and user experience with both systems are very similar. This finding will be the base of more extensively studies to demonstrate that home-therapy with HomeRehab could be as efficient as therapy in clinical settings assisted by PupArm robot.

For assistive devices to seamlessly and promptly assist users with activities of daily living (ADL), it is important to understand the user’s intention. Current assistive systems are mostly driven by unimodal sensory input which hinders their accuracy and responses. In this paper, we propose a context-aware sensor fusion framework to detect intention for assistive robotic devices which fuses information from a wearable video camera and wearable inertial measurement unit (IMU) sensors. A Naive Bayes classifier is used to predict the intent to move from IMU data and the object classification results from the video data. The proposed approach can achieve an accuracy of 85.2% in detecting movement intention.

The assessment of muscle fatigue could provide crucial information to monitor the progression of a neuromuscular disease, as well as evidences about the efficacy of an eventual therapeutic approach. The aim of the present work is to test the feasibility of a novel method based on a robotic wrist device and sEMG, on neuromuscular pediatric subjects for a future implementation in clinical practice. The proposed paradigm has been previously validated on a population of healthy subjects. Five children (age 8.8 ± 1.3 yo) with Duchenne Muscular Dystrophy (DMD) performed the test and their right flexor and extensor carpi radialis were recorded to obtain an objective indicator of muscle fatigue.

This is a pilot study into the feasibility and effectiveness of robot-assisted gait training (RAGT) provided with a mobile exoskeleton for overground walking practice. In total, 7 chronic stroke survivors were recruited and allocated to either the RAGT or a conventional therapy group. Both groups were provided with 16 additional hours of therapy, however clinical relevant improvements in walking ability were observed in the RAGT group only. Training was well-accepted and no adverse events occurred. Greater, well-designed trials are needed to confirm these preliminary results.

People of industrialized countries are living longer and consequently the incidence of disability coming from medical conditions such as stroke is increasing. The hand rehabilitation process following a stroke accident requires very intense rehabilitation sessions. It is crucial to improve the outcomes of hand physical therapy by providing patients with an orthosis aid designed to assist their movement and performing the role of exercising the hand during the early stages of recovery. The aim of this work is to describe the system architecture and the mechanical design of a new tendon-like orthosis actuated by shape-memory-alloy (SMA) wires and controlled by myoelectric signals.

This paper presents preliminary comparison of CoM and CoP paths between the healthy and stroke patient in walking. The results show that single support with affected leg of stroke patient causes wide stepping of swing leg in lateral direction rather than in forward direction, resulting in an asymmetrical gait pattern.

Stroke is a leading cause of upper-limb impairment and a motivating factor behind the development of robotic technologies for arm assessment and rehabilitation. The shoulder is one of the most complex joints in the body and a potential source of human-exoskeleton misalignment. This paper presents histograms of shoulder displacement during 18 activities of daily living (ADLs) and demonstrates the shoulder joint translations that must be accommodated if misalignment is to be avoided during natural reach and grasp movements of the arm.

Brain state-dependent afferent stimulation constitutes a promising technique for rehabilitation of upper limb motor functions after stroke. Previous studies have shown that the associative stimulation with certain temporary restrictions between the user’s movement intent and the electrical stimuli delivered peripherally can generate plastic changes. However, this approach applied to the functional recovery of the upper limb has not been tested yet. The objective of this study is to ascertain whether the application of the functional electrical stimulation at the precise moment at which the subject wishes to move or at a different time interval (before or after), leads to significant changes in neurophysiological, clinical and kinematic measures as a result of undergoing multi-sessions intervention. The proposed intervention, which consists of a reaching movement, is composed of nine sessions divided into three weeks.

Post-stroke functional impairments of the arm effect a large and growing population of individuals. Numerous arm exoskeletons have been developed to help address the increasing therapy demands, but many have simplified the shoulder in ways that significantly limit its mobility. In this paper, a 2 degree-of-freedom (DOF) shoulder module is developed that can add sternoclavicular and acromioclavicular articulations to existing 3-DOF exoskeleton shoulders.

BLUE SABINO is an NSF-funded collaborative project between researchers at the University of Idaho, UCLA, and St. Luke’s Rehabilitation Institute. The goal of the project is to develop a BiLateral Upper-extremity Exoskeleton for Simultaneous Assessment of Biomechanical and Neuromuscular Output (BLUE SABINO). The project combines a dual-arm exoskeleton system with off-the-shelf electromyographic (EMG) and electroencephalographic (EEG) measurement equipment. A preliminary CAD model, featuring innovative new designs and components adapted from several previous systems, has been developed to address the needs of therapists and patients in post-stroke neurological assessment.

The use of robotic systems during the rehabilitation of spinal cord injured patients can contribute to achieve significant outcomes. The aim of this study is to investigate the effects of an overground robot-assisted gait training in complete spinal cord injured patients. Sixteen SCI patients (mean age: 35.3 ± 10.3 yrs) underwent 20 robot-assisted gait training sessions based on an overground robotic exoskeleton.The preliminary results of this study show that the robot-assisted gait training in complete SCI patients can improve patient-robot interaction.

In neuro-motion rehabilitation it is important for the subjects to plan a neural and motion rehabilitation program at home. Telerehabilitation could represent a valid aid for subjects involved in a rehabilitation process at home. Motion and neural rehabilitation are strictly correlated.Prompt motion rehabilitation is essential to ensure good recovery performance. Remote therapy, which involves a tele-rehabilitation program, should monitor daily motion activity. The integration of step-counters in the process could improve the care, providing daily quantitative measurements.Three different step counters (a) have been proposed and (b) integrated in a kit (c) with a study aimed at designing a guideline for the optimal assignment on the basis of the pathology.

The aim of this study is to test the feasibility of a new protocol of upper limb treatment, using a set of four robotic and sensor-based devices in sub-acute stroke patients in a multicenter study aimed to evaluate the efficacy of robotic therapy compared with conventional therapy. Participants are recruited in 8 centers and randomized to either the robotic group or the conventional group. The outcomes were selected according to the International Classification of Functioning, Disability and Health domains, to evaluate body function, activities and participation. A total of 247 participants were randomized, 224 of whom received the allocated intervention. Our results demonstrate the feasibility of our rehabilitation protocol in a multicenter clinical trial and in clinical practice.

Rehabilitation is essential after stroke and, besides conventional rehabilitation, technological one has had big growth in clinical practice. There is a growing interest in cognitive reserve (CR) that summarizes pre-morbid life of each patient and has a key role in a sudden change of individual lifestyle (for example, after a stroke). Our preliminary data suggested that CR impacts on motor rehabilitation outcome. We hypothesized that CR may help in the complex choice between technological or conventional rehabilitation. The aim of this study is to evaluate whether the CR influences the motor outcome in patients after stroke treated with conventional or robotic therapy and if CR may address towards one treatment rather than another.

The goal of this study is to analyse the effects of combined transcranial direct current stimulation (tDCS) and wrist robot-assisted therapy in subacute stroke patients.Thirty-nine patients were included in this study and randomly assigned to the experimental or control group. All participants performed wrist robot-assisted training (a) in conjunction with tDCS (real stimulation for experimental group) or (b) without tDCS (sham stimulation for control group). Clinical outcome measure and kinematic parameters were used as assessment metrics.The clinical outcome measures, except the Modified Ashworth Scale showed significant increase after treatment in both groups, but no significant difference in the average changes after treatment between groups was observed.Movement velocity and smoothness showed some significant increases after training, however no significant difference between groups was observed.The combination of wrist robotic training and tDCS did not show any additional effects in comparison with wrist robot-assisted training only in subacute stroke patients.

Tetraplegia is a devastating condition, resulting in severe disability and isolation from social activities and entertainment. Drones may provide a severely paralyzed individual the possibility of participation in drone-sports and thereby in social interaction and further it may give a sense of freely moving. However, individuals with tetraplegia currently lack options for controlling a drone. Researchers at Aalborg University have developed a wireless intraoral tongue computer interface (ITCI) for disabled users. This study investigates the possibility of controlling a drone by the ITCI. One able-bodied experimental participant controlled the drone using a standard keyword, the ITCI while keeping it in the hand, and by mounting the ITCI inside the mouth and using the tongue. The performance of the ITCI was compared with respect to the keyboard. The mean distance from the target and the mean flight time were 36% and 16% greater when using the ITCI inside the mouth with respect to controlling by the keyboard.

The objective of this study is to compare the effects of robot-assisted hand training (RAHT) against non-robotic training on upper limb activity, hand dexterity and muscle activity in patients with MS. 34 persons with MS (EDSS: 1.5–8) and hand dexterity deficits were enrolled. The RAHT group (n = 17) received robot-assisted hand training and the control group (n = 17) received non-robotic training.Training consisted of 50 min/session, 2 sessions/week, 5 weeks. Before, after, and at 1-month follow-up patients were evaluated by a blinded rater by a comprehensive tests series investigating ICF domains and surface electromyography of the upper limbs.No significant between-group differences in primary and secondary outcomes were found. An increased amount of upper limb use and increased EMG activity in the extensor carpi were observed only in the RAHT group. Although training effects were comparable between the two groups, the RAHT demonstrated remarkable effects on upper limb use and muscle activity.

The aim of this two-part paper is to find clues for designing a rehabilitation protocol for robot-assisted gait training, based on the principles of “assisting patient movements only when necessary” and “eliciting a normative sensory input to improve motor output”, applied through neurorehabilitation techniques to facilitate the functional recovery of gait in spinal cord lesion patients.For that purpose, in the first part we review the specific role of the foot sole receptor input on the modulation of the spinal cord motor output for gait. From this knowledge, the second part will explore the applicability of the foot sole dynamometric map concept in a case-study of gait of a patient with low-lumbar level myelomeningocele. The goal is to evaluate whether this type of map could supply spatiotemporal data linked to that foot sole receptors role, useful for promoting neurorehabilitation through robotic gait training, in line with the aforementioned therapeutic principles.

Knowledge from the review of the first part of this paper is employed here to explore the applicability of the foot sole dynamometric map concept to gait, as well as its possible relation to the function of the foot sole receptors during walking. For this purpose, the spatiotemporal plantar pressure distribution and its variability throughout the gait stance phase for a case study of a patient with low-lumbar level myelomeningocele is determined and compared with those of a healthy subject. The results indicate a significant spatial variability at the stance phase with a limited contact of the forefoot, probably related to an ankle plantar flexor muscle weakness shown by this patient. In future researches we will analyse deeper whether this type of information could be used to design a rehabilitation protocol for robot-assisted gait training, oriented to promote the functional recovery of gait in patients with spinal cord lesion.

In this study, we inferred the changes in proprioception through kinematic data analysis in repetitive robot-assisted reach exercise. It was assumed that proprioception could be inferred by observing the variability of point-to-point movements during exercise. Ten chronic stroke survivors performed the robot-assisted reach exercise for six weeks. The averaged distance between points passing through a virtual plane near a return target position and their centroid during the reach exercise was calculated as an index of the movement variability. The results showed that the variability was reduced by 15.7%. Using clinical assessment tools, functional arm movement improved by 6%. We can infer the changes in proprioception from the exercise based on the decrease of the variability.

The motoBOTTE is a robot designed for assistive rehabilitation. The identification of a mathematical model for the system is shown; then we design and implement a nonlinear control law for tracking a reference signal. The nonlinear controller design is achieved by combining an exact convex representation with the direct method of Lyapunov. Results are given in terms of linear matrix inequalities (LMI); simulation and real-time results are shown as well.

Regaining balance function is often one of the key goals of stroke rehabilitation. Improvements in balance function can be the result of restitution or compensational strategies. In previous studies, the processes of restitution and compensational strategies have been established for straight-line walking. The development of these processes, however, are still largely unknown for other gait activities such as turning and side-stepping. Here, we present a fully ambulant gait analysis system that can be used for an objective evaluation of balance during functional tasks. The results of two individuals are presented: one individual with adequate balance function and one individual with impaired balance function. The analysis showed that the individual with adequate balance function was able to walk with increased instability when compared to the individual with impaired balance function. Based on these results, we conclude that the fully ambulant system is feasible for an objective quantification of balance function.

Spasticity, characterized by hyperreflexia, is a common motor impairment following hemispheric stroke. Botulinum toxin (BT) injections are widely used to reduce spasticity. BT acts by disrupting action potential propagation between a motor neuron and its innervated muscle fibers. Here we quantified the magnitude and time course of a BT injection on the stretch reflex responses, by analyzing the root mean squared amplitude of tap evoked surface EMG (sEMG) potentials from stroke affected biceps brachii (BB), before the BT injection and up to 18 weeks post-injection. We used a 16 × 8 sEMG electrode grid and mapped the tap-evoked potentials. We found, (a) the maximum reduction of the RMS EMG amplitude of the reflex response to occur six weeks post-BT injection and (b) a gradual recovery of the RMS EMG amplitude by 18 weeks post-injection. Unexpectedly, even at 18 weeks post injection, the tap-evoked reflex activity did not fully recover to its baseline (pre-injection recordings). These findings have a potential impact on the timing of BT administration for repeated injections.

A novel metric for estimating connectivity between brain areas, namely the Phase Linearity Measurement (PLM), is presented. The purpose consists in measuring the amount of information exchanged between brain areas. Such scope is achieved by analyzing the similarities between the recorded signal phases. The PLM has been designed for exploiting both Electroencephalographic (EEG) and Magnetoencephalographic (MEG) data. We compared the results achieved by PLM in case of real MEG data with a widely adopted phase based connectivity metric, the Phase Lag Index (PLI). The PLM is characterized by interesting results, mainly in terms of noise resiliency.

The aim of this study is to evaluate the changes that Alzheimer’s disease (AD) and mild cognitive impairment (MCI) cause in the neural patterns of information flow and in the brain network properties. For this purpose, phase-slope index (PSI) was applied to spontaneous electroencephalographic activity from 32 AD patients, 10 MCI subjects and 18 cognitively healthy controls. Then, three network parameters were calculated: average node degree (ND), cluster coefficient (CC), and characteristic path length (PL). Our results showed that information flux values were lower for AD and MCI subjects, compared to controls. Additionally, significantly lower ND and higher PL values were obtained in AD group, compared with MCI and controls in alpha frequency band. These findings support the idea of disconnection syndrome in AD and revealed less efficient brain organization as the disease progresses.

The aim of this study was to characterize the non-stationarity level of resting-state EEG in patients with dementia due to Alzheimer’s disease (AD), subjects with mild cognitive impairment (MCI) and healthy controls. A frequency-dependent implementation of the Kullback-Leibler divergence was used to characterize non-stationarity patterns. The results showed a statistically significant increase in non-stationarity for AD patients with respect to controls in the 1–70 Hz frequency range, as well as a less pronounced increase for MCI subjects with respect to controls. These results suggest that EEG activity during short time windows consists of more structured oscillations than that of AD patients or MCI subjects.

The aim of this study was to analyze electroencephalographic (EEG) background activity in Alzheimer’s disease (AD) and mild cognitive impairment (MCI) by means of multiscale spectral entropy (MSSE). To achieve this goal, five minutes of EEG activity were acquired from 18 cognitive healthy controls, 10 MCI subjects and 32 AD patients. Our results showed statistically significant differences (p-values < 0.05, Kruskal-Wallis test) in MSSE values for all scale factors. Additionally, a 3D receiver operating characteristic (ROC) curve was used to assess the discrimination ability of MSSE among the 3 groups, showing a good three-way discrimination power (volume under the surface of 0.6). These results suggest that MSSE can be a useful measure to characterize neural alterations in AD, even at early stages.

The signals delivered by different sensory modalities provide us with complementary information about the environment. A key component of interacting with the world is how to direct ones’ sensors so as to extract task-relevant information in order to optimize subsequent perceptual decisions. This process is often referred to as active sensing. Importantly, the processing of multisensory information acquired actively from multiple sensory modalities requires the interaction of multiple brain areas over time. Here we investigated the neural underpinnings of active visual-haptic integration during performance of a two-alternative forced choice (2AFC) reaction time (RT) task. We asked human subjects to discriminate the amplitude of two texture stimuli (a) using only visual (V) information, (b) using only haptic (H) information and (c) combining the two sensory cues (VH), while electroencephalograms (EEG) were recorded. To quantify multivariate interactions between EEG signals and active sensory experience in the three sensory conditions, we employed a novel information-theoretic methodology. This approach provides a principled way to quantify the contribution of each one of the sensory modalities to the perception of the stimulus and assess whether the respective neural representations may interact to form a percept of the stimulus and ultimately drive perceptual decisions. Application of this method to our data identified (a) an EEG component (comprising frontal and occipital electrodes) carrying behavioral information that is common to the two sensory inputs and (b) another EEG component (mainly motor) reflecting a synergistic representational interaction between the two sensory inputs. We suggest that the proposed approach can be used to elucidate the neural mechanisms underlying cross-modal interactions in active multisensory processing and decision-making.

The scope of this mini review is to show as neuromotor control often applies mathematical principles and properties improving efficiency of movements. We reported four examples: harmony of human walking is based on the fractal property of self-similarity because based on the golden ratio, anatomy of human hand follows Fibonacci’s sequence for optimizing grasping, the 2/3 power-law at which curvilinear movements obey, and the possibility of brain to predict the movement of falling objects applying an implicit knowledge of the Newton’s gravitational law.

Grip Strength reduction is a marker of the age-related functional decline. The present work developed an exergame to continuously and transparently monitor handgrip strength and endurance in older persons. The game was tested on ten young adults reporting positive results, both in terms of usability and correlation with clinical standards, thus representing a promising, non-intrusive tool to help detecting early signs of frailty in community-dwelling older persons.

In the present research we provide an overview of new solutions and problems of wearable and virtual reality devices for neurorehabilitation. A lot of commercial devices currently help physiotherapist to provide rehabilitation for reaching, grasping, walking and balance recovery, adapting cognitive load and providing challenging exercises. The present review highlighting the positive step made by the translational research and the unsolved problems. In particular, a critical step over to increase usability should be made in the future because wearable devices and virtual reality should represent a promising option to ensure a prolonged home rehabilitation.

A brief history of the evolution of gait analysis in the field of clinical decision making process is provided. There are also provided some indications about future incarnations of this topic. An example of simple, but clinically relevant, data integration is given among the possible new solutions, evidencing the effective current limits of the technique. These limits represent the real future challenge in the field of movement analysis.

Aim of this study was to develop an anatomical calibration procedure for wearable Magnetic and Inertial Measurement Units to assess Reach-To-Grasp kinematics in the stroke unit. A calliper hosting a MIMU was used to measure the direction of axes identified by pairs of selected Anatomical Landmarks and the inter-ALs distance along these axes. The upper limb was modelled as a two-link open kinematic chain and forward kinematics was used to estimate the position of the end-effector on six subjects wearing three MIMUs on the thorax, upper arm and forearm while performing fifteen RTG cycles. The proposed procedure showed a low bias and good levels of agreement with respect to reference positional data.

The central nervous system controls body functions and initiates actions through an integrated system of multiple feedback loops. System identification can be a valuable tool to assess the system dynamics. The challenge is to assess the functioning of an intact feedback system in vivo, where cause and effect are intermingled. In this paper we show the errors which can occur when estimating system dynamics using standard (open-loop) system identification algorithms in the presence of feedback loops. Furthermore, we show how these issues can be addressed by using closed-loop algorithms. A closed-loop system identification approach is essential to assess the separate feedback loops of the central nervous system in an intact functional system.

This paper focuses on the pathophysiology of fixed dystonia (i.e., sustained muscle contractions resulting in abnormal postures) in Complex Regional Pain Syndrome from an engineering point of view. Although the mechanisms are still elusive, the evidence implicating involvement of aberrant muscle force feedback is compelling. A neuromuscular model with aberrant muscle force feedback successfully mimicked fixed dystonia while results of several experiments point to involvement of muscle force feedback.

The correlation of the lower-leg muscle contraction to the ankle impedance of unimpaired subjects is studied. Each subject participated in 5 experimental trials, each with a different co-contraction level: 0%, 10%, 20%, 30%, and 40% of their maximum voluntary contraction (MVC). A linear model is developed to relate the muscle contraction and the ankle impedance. Next, an ANOVA test is used to verify the significance of the parameters. Low correlation is found on the inversion-eversion degree-of-freedom of the ankle, suggesting non-linear models might be more effective in describing this relationship.

Short segment and nonlinear parameter varying methods are two promising approaches to the identification of the time-varying properties of joint stiffness during functional activities. This paper describes each method, demonstrates their application, and discusses their strengths and weaknesses.

Neurological disorders are typically followed by the secondary changes in neuromuscular properties. Characterizing the neuromuscular abnormalities and separating their tendon-muscular (intrinsic) and reflexive components are critical for prescribing an appropriate and effective treatment, since each component needs different types of treatments. Our studies demonstrate that the non-parametric parallel-cascade system-identification approach was quite successful in determining the neuromuscular abnormalities, separating the intrinsic and reflex components, tracking the development of their abnormalities, or the improvements of these impairments due to the administration of different types of interventions.

In this study an upper limb biomechanical model which includes strain-adjustable ligaments of the acromioclavicular joint was developed. The model including 7° of freedom is able to evaluate the movements of the shoulder, elbow and wrist. The ligaments strain can be adjusted in order to simulate different types of Rockwood acromioclavicular dislocation.Movements recorded from three healthy subjects are used as preliminary assessment of the proposed model.

This paper provides an overview of the development process of wearable device applications using two case studies. KneeHapp is a smart textile bandage that measures the performance of different rehabilitation exercises performed by patients after a knee injury and HipRApp is a bandage that tracks the recovery of patients after a hip surgery based on gait analysis. We summarize the common phases in the development of wearable device applications and discuss what kind of computations would be suitable for deployment into wearable devices with limited resources.

Locomotion mode recognition plays an important role in the control of lower-limb exoskeletons and prostheses. In such applications, the accurate and timely classification of the locomotion mode, using the minimum number of sensors, is still a challenge. In this paper we present an algorithm to recognize four different locomotion modes (namely stand, walk, stair ascent, and stair descent) and all the possible transitions among them, based on wearable sensors. The algorithm grounds on an event-based and mode-dependent strategy, which is able to recognize the locomotion mode during the swing phase. Tests conducted with three healthy subjects showed an average recognition accuracy of 98.8 ± 0.4% in steady locomotion conditions. Transitions between different modes were also accurately detected during the swing phase. Further studies will be conducted to validate the algorithm and test it in real-time applications with wearable robots.

Motor rehabilitation and assistance post-stroke are becoming a major concern for healthcare services with an increasingly aging population. Wearable robots can be a technological solution to support gait rehabilitation and to provide assistance to enable users to carry out activities of daily living independently. To address the need for long-term assistance for stroke survivors suffering from drop foot, this paper proposes a low-cost, assistive ankle joint exoskeleton for gait assistance. The proposed exoskeleton is designed to provide ankle foot support thus enabling normal walking gait. Baseline gait reading was recorded from two force sensors attached to a custom-built shoe insole of the exoskeleton. From our experiments, the average maximum force during heel-strike (63.95 N) and toe-off (54.84 N) were found, in addition to the average period of a gait cycle (1.45 s). The timing and force data were used to control the actuation of tendons of the exoskeleton to prevent the foot from preemptively hitting the ground during swing phase.

Extraction of muscle synergies from electromyography (EMG) recordings relies on the analysis of multi-trial muscle activation data. To identify the underlying modular structure, dimensionality reduction algorithms are usually applied to the EMG signals. This process requires a rigid alignment of muscle activity across trials that is typically achieved by the normalization of the length of each trial. However, this time-normalization ignores important temporal variability that is present on single trials as result of neuromechanical processes or task demands. To overcome this limitation, we propose a novel method that simultaneously aligns muscle activity data and extracts spatial and temporal muscle synergies. This approach relies on an unsupervised learning algorithm that extends our previously developed space-by-time decomposition to incorporate the identification of linear time warps for individual trials. We apply the proposed method to high-dimensional spatiotemporal EMG data recorded during performance of whole-body reaching movements and show that it identifies meaningful spatial and temporal structure in muscle activity despite differences in trial lengths. We suggest that this algorithm is a useful tool to identify muscle synergies in a variety of natural self-paced motor behaviors.

Many diseases are characterized by limitations in mobility, including a wide range of musculoskeletal and neurological conditions. Reduced mobility impacts a patients ability to perform activities of daily living, which in turn reduces health-related quality of life. Mobility can be assessed by collecting patient-reported outcome scores from standardized questionnaires and by directly measuring physical activity parameters from wearable accelerometer data. In this work, we explored the relationship between subjectively and objectively measured mobility by training machine learning models to predict patient responses based on features derived from real-world acceleration data. Our method achieved up to 82% accuracy using a random forest classifier and set the basis to develop novel data-driven digital biomarkers for objective, quantitative and more frequent evaluation of patients’ mobility.

Stimulation in various sites and with various means has been used as a therapeutic tool for many years. Even if the outcome may not always be as satisfactory as expected, the relatively good tolerability and safety of stimulation therapy have maintained the field evolving. In a broad spectrum, non-invasive stimulation therapy can be divided into non-invasive brain stimulation (NIBS) and non-invasive non-brain stimulation (NInBS). I will revise the novel forms of NIBS and NInBS that have demonstrated beneficial effects on neurological symptoms in the past 3 years, and comment on the tendencies that can be detected with therapeutic application of stimulation in neurological disorders.

Repetitive transcranial magnetic stimulation (rTMS) is a non-invasive neuromodulation technique able to influence cortical excitability. In stroke patients, both excitatory and inhibitory protocols have been used to favour functional recovery of the upper limb function. In more than 30 clinical trials, inhibitory rTMS of the unaffected (contralesional) primary motor cortex (M1) has been shown effective to improve function of the affected hand in mild stroke patients. Excitatory rTMS of the affected (ipsilesional) M1 has been less investigated. Nonetheless, its safety and the possible clinical efficacy is supported by several data. However, there is considerable heterogeneity across studies in the stimulation protocols, patient populations, lesion site, outcome measures, stroke aetiology and duration. That means that the studies are not readily comparable, the reproducibility of the results can be limited, and an analysis of the real clinical significance is very difficult. For these reasons, the need for a multicentre clinical trial is advocated by many authors.

Transcranial direct current stimulation (tDCS) has been applied recently over the primary motor cortex and has recently been shown to neuroplasticity of the endogenous pain modulation (EPM) system. The aim of the current study is to present early results related to motor cortex and spinal neuroplasticity control of the EPM in healthy volunteers by measuring pressure pain threshold and cold pain intensity as outcome measures before and after M1 cortex or suboccipital DCS. Healthy volunteers (aged 18–40 years), in a randomized, double-blind, placebo-controlled, clinical trial, were assigned to four DCS groups: sham-M1 cortex DCS, active-M1 cortex DCS, sham-suboccipital (SODCS) and active-SODCS. Data collected to date suggest that both M1 and spinal-DCS modulate thermal noxious stimuli. However, the demonstration of EPM neuroplasticity requires careful attention to the test and conditioning paradigm.

The use of mechanical energies delivered in form of vibration therapy has raised more and more importance in neurorehabilitation, in view of the ability of vibration therapy to interact with neuromuscular system. The aim of this lecture will be to discuss the main experimental and clinical findings related to the application of segmental (or focal) vibration in neurorehabilitation.

The Individual Alpha Peak Frequency (IAF) is the individual dominant electroencephalogram (EEG) frequency in the range of n to m (n = 8 and m = 12). IAF is related to various cognitive functions such as attention and working memory; and can be affected by biological, psychological and social aspects. In this paper, a Neurofeedback (NF) protocol is presented, which takes into consideration these three aspects. The main purpose is to create an Individual Alpha Peak Frequency (IAPF) dataset for a NF system in order to predict the number of NF sessions for a cognitive skills improvement. Two studies were performed using this protocol with 10 students divided in experimental and control groups, where an advance in the IAPF (Frequency and Absolute Power) can be observed in the first group.

The possibility to prescribe home-based rehabilitation activity after stroke strongly increases the amount of exercises to perform, thus helping the maintenance of relearned skills, the completion of the rehabilitation program, the practice of physical and mental concentration. Even more important is the monitoring of the patient activity at home, as it is provided by the Remote Monitoring Validation Engineering System (ReMoVES) platform [1]. The present work refers to the implementation and integration in ReMoVES platform of a digital and web-based version of Albert’s [2] and Line Bisection [3] tests devoted to visuo-spatial neglect evaluation and its remote monitoring. A statistical analysis devoted to validating test-retest reliability is proposed. Concurrent correlation between digital and traditional administration of the tests is presented, in order to evaluate the validity of the remote monitoring of the home-administration through ReMoVES platform.

Depression is a common sequela after stroke and has severe implications on a patient’s life. Post-stroke depression has been linked to cognitive impairment, but the mechanisms that lead to this deficit are not well understood. We tested 18 chronic stroke patients with depression in a psychophysical task to evaluate their attentional processing under varying cognitive loads. We found that the level of depression had no effect on the unconscious, bottom-up components of attentional processing but did influence the top-down ones. These results support the notion that depression might act like an additional cognitive load, impeding the conscious processes and responses although the information has been unconsciously processed.

Mild cognitive impairment is a borderline condition between normal aging and first stages of dementia. Programs based on multidimensional trainings appear to have a role in reducing cognitive decline.A 8 weeks tele-rehabilitation program, administered through an ad-hoc designed web-application, was developed. The design of the web-application was based on feedback received from specialists, patients and caregivers. The program comprised three trainings: cognitive, physical and social.Thirty subjects (age 73 ± 4 yrs) with mild cognitive impairment were included in the study to investigate the feasibility of the proposed program. 25 participants completed the program and showed a high adherence to the proposed activities (85% and 83% for cognitive and physical activity, respectively). Good usability and high level of appreciation (mean appreciation = 3.7 ± 0.8, scale range 1–5) were reported together with physical and memory benefits. The proposed program seems thus suitable to provide multidimensional trainings to subjects living with mild cognitive impairment.

VR is a useful tool for the improvement and customization of the classical clinical settings. The aim of this paper is to present an innovative virtual-based cognitive rehabilitation protocol. We also present the results of a patient, SB, who showed an improvement in the results of the test in both the target and non-target cognitive domains.

An important challenge of stroke rehabilitation consists on the application of efficient therapies during the initial stage of recovery. These strategies should focus on the activation of the local neural loop by reducing those parts of the movement inducing fatigue. Therefore, reducing the recruitment of fast-twitch muscle fibers is an important point while choosing an appropriate therapy. This paper is focus on the evaluation of superficial electromyographic signals during supported and non-supported elbow flexion task to prove their differences in terms of fast-twitch muscle fiber recruitment. To do that, the median frequency extracted from different upper limb muscles of 21 healthy subjects is evaluated showing significant differences supporting authors’ hypothesis.

Speed control is a crucial factor of motor skill. Although rehabilitation for damaged central mechanisms underpinning the appropriate temporal tuning during bimanual coordination is an important target, the detailed knowledge about the methodology of the therapy is quite limited. This study aimed to clarify the effect of slowing the pace of bimanual rhythmic steering task using a new steering device on the homologous muscle coupling exemplified by the EMG-EMG coherence. At the slow pace, homologous coherence in control subjects at a frequency band of 65–115 Hz in forearm flexors increased relative to the natural speed, whereas small increment in subjects after stroke, suggesting activated central coupling mechanisms dominate the distal muscles for speed adaptation. Relationship between the capability of this mechanism and recovery after stroke is of interesting topic in future studies and will be beneficial for physical rehabilitation.

Understanding motor learning mechanisms could help find new approaches in motor neurorehabilitation. We used the classical Serial Reaction Time Task (SRTT) to study motor learning in healthy subjects. The aim was to investigate its neural basis by asking whether neural interactions, i.e., functional connectivity (FC), during a task-free resting-state influence implicit sequence learning and consolidation. Contrary to what was expected, neither online implicit motor learning nor offline consolidation of the sequence was found. Instead, a rapid and transient effect on response times appeared every time a repeated sequence was followed by random trials. We argue that this effect, commonly observed with the SRTT, is a form of motor adaptation rather than learning. A significant positive correlation was found between this adaptation and global FC of the basal ganglia during the resting-state period preceding the task. This suggests that strategies geared towards enhancing resting-state neural interactions of key nodes for motor skills may be increase the efficacy of physical therapy.

Musculoskeletal pain is the most frequent health complaint reported by workers in Europe. Neurofeedback has been proposed to be an alternative to the current treatment of pain, however, the extent to which musculoskeletal pain alters the electroencephalographic (EEG) signal is still not known. The current study aims at identifying the signal characteristics provoked by musculoskeletal pain during movement. Healthy volunteers and patients diagnosed with Lateral Epicondylalgia performed wrist extension movements while EEG signals were collected. The power of the EEG signal was calculated and differences between healthy volunteers and patients were assessed. EEG activity of pain patients differed significantly from that observed in heathy volunteers within the alpha and beta band. This alteration is movement related and is particularly visible in frontal channel locations. The results of the current study are currently being implemented for the development of a neurofeedback protocol to treat musculoskeletal pain.

In this work, we analyze the suppression of the power in mu and beta bands used in Motor-Imagery Brain Computer Interface systems (MI BCI) when using artificial frames. We compared the suppression effect between real and artificial frames. Experimental results in a single subject example show that artificial frames capture the same effect observed in real frames at a similar level. This interesting result supports the use of artificial frames during the BCI training process, which should reduce the number of real frames and hence reduce the calibration time in practical applications.

Dementia, and specially Alzheimer’s disease (AD) and Mild Cognitive Impairment (MCI) are one of the most important diseases suffered by elderly population. Music therapy is one of the most widely used non-pharmacological treatment in the field of cognitive impairments, given that music influences their mood, behavior, the decrease of anxiety, as well as facilitating reminiscence, emotional expressions and movement. In this work we present HAIDA, a multi-platform support system for Musical Therapy oriented to cognitive impairment, which includes not only therapy tools but also non-invasive biometric analysis, speech, activity and hand activity. At this moment the system is on use and recording the first sets of data. Results obtained using HAIDA will be presented in a near future after the analysis.

In this paper we propose a system based on multimodal biofeedback for improving postural stability.The core elements of the system are: a sensor unit based on accelerometers and rate-gyroscopes; a coding unit able to A/D convert the signals from the sensor unit and generate three different biofeedback restitutions; a multimodal biofeedback system embedding: (a) a sound system for audio biofeedback including 4 speakers, (b) a belt with four vibrotactile actuators for tactile biofeedback, (c) a Video Display terminal for visual biofeedback. The study reports results from an application of the sound biofeedback to 15 subjects in a dedicated protocol.

The abuse of the smartphone technology is today causing a new syndrome, called the text neck. This syndrome damages the neuro-musculoskeletal apparatus. It is recorded in subjects of different ages and in particular among young people. A Medical App for the self-care in the text neck prevention has been proposed. This App returns: (a) a biofeedback on correct/incorrect posture; (b) a temporal accumulation function that indicates how long a subject is using the device and (c) a set of programmed postural exercises. The App was instrumentally validated in diminishing the neck tilt (related to the text neck), using wearable technologies, on a group of 15 young subjects.

This study aimed to evaluate the corticomuscular changes associated with motor training. We quantified functional and corticomuscular changes during a four-days motor training. Results revealed better functional performance and increased magnitude of corticomuscular coherence in the 8–13 Hz frequency band over days. These results are supposed to reflect a better integration of sensorimotor integration.

Connectograms are novel tools for intuitive brain connectivity visualization and could be useful in the context of personalized medicine and patient-tailored rehabilitation. In this study, we investigated the motor network structural connectivity pattern of a patient with cervical dystonia (CD) before and after treatment with Botulin Neurotoxin (BoNT) and compared it with an healthy control (HC) subject. Results showed abnormal inter-hemispheric connectivity in CD patient both before and after BoNT, suggesting the importance of associating rehabilitation to BoNT in CD treatment.

Trunk control is considered an important predictor for functional outcome after stroke. Therefore incorporating exercises to increase trunk control is of great importance during stroke rehabilitation. The purpose of this study is to investigate the effect of trunk training on trunk control, standing balance and gait. Results showed that trunk training is able to improve the total score of Trunk Impairment Scale (Z = 16.21, p < 0.0001), Berg Balance Scale (Z = 4.50, p < 0.0001) and Timed Up and Go test (Z = 3.72, p < 0.0001). Consequently, we recommend to incorporate trunk training in standard care after stroke.

In the majority of stroke patients, trunk function is impaired and characterized by diminished sitting balance, trunk coordination and muscle strength. However, there is no true consensus how these impairments are translated during walking after sub-acute stroke. Results showed that the hemiplegic gait pattern was characterized with increased thoracic flexion, pelvic hike and decreased rotational amplitude which more specifically resulted in more external rotation.

The assessment of stroke patients is a long lasting process in which physicians estimate the impairment import of the patients through clinical scales and physiological exams. This is a crucial phase for defining the patient treatment as well as monitoring the recovery process. In order to support the whole set of operations involved in the assessment, the authors proposed a simple evaluation tool that easily allows physicians to collect performance, muscles activity and upper-limb kinematic data from patients and thus extract clinically useful information. Preliminary results obtained form 3 healthy subjects and 2 patients demonstrated the suitability of the proposed system for collecting relevant information in just one short session.

Over the last decades, Virtual Reality (VR) emerged as a potential tool for developing new rehabilitation treatments in neurological patients. However, despite the increasing number of studies, a clear comprehension about the impact of immersive VR-treatment on balance and posture is still scarce. In this study, we aimed to investigate the postural alterations in post-stroke patients and elderly people during the exposure to an immersive moving virtual scenario. Ten patients with sub-acute stroke and ten healthy subjects took part in this study. Both groups were immersed in a CAVE system on a baropodometric platform. The VR task consisted in ten trials, which differed in term of speed and movement direction. Results showed that the sway path length (representative of the body sway amplitudes) was trial-dependent for both groups. Therefore, the kind of virtual visual stimulation influenced the body sway response and it could be considered for designing a rehabilitation protocol.

The aims of the current paper was analysis the effects of wearing high heel on lower limb coordination. Wearing high heels shoes change gait biomechanics. Nine youthful healthy females participated in this study. Kinematic data were captured at 200 Hz using 6-camera motion analysis system (S Infrared, Vicon camera, Oxford metrics, Oxford, UK). Results show no significant differences in knee phase plane and CRP (except 45–62% of trial time) were found between bare-foot and high-heel conditions but significant differences were found in hip phase plane. Also, peak values of the lower joint CRP in conditions are different. Peak value was found at 55 and 60% in high-heel and bare-foot respectively.

Can intense learning induce neuronal fatigue in well rested subjects? To answer this question, we recorded high density EEG during extended practice in a visuo-motor adaptation. We found that prolonged learning leads to a progressive slowing of electroencephalographic (EEG) activity over the cortical areas engaged by the task itself. Extended daytime training resulted in task-specific and local sEEG power increase in the low frequency range over areas that were engaged during training. This was accompanied by increased performance errors in the test that shared the task characteristics. Finally, an afternoon nap but not an equivalent period of quiet wake renormalized the local EEG changes induced by intense training and restored both performance and the ability to learn.

While the concept of force is solidly grounded in Newtonian mechanics, it is not known if it is also represented in a consistent way by our brains as they control interactions of the hand with external objects. For example, a force of 10 N applied against different springs will cause different amounts of displacement. Are we able to represent 10 N in a way that is independent of the effects of applying such force to different objects? Here, we developed a simple method to address this question by engaging subjects in a task whose success depends critically upon the ability to exert a fixed force against different simulated springs. Our preliminary findings indicate that while this task is difficult, subjects learn after some training to exert the same force against different springs and in different directions.

Direct in-vivo measurement of individual muscle force requires placement of a force sensing element in series with the musculo-tendon structures, which is infeasible due to the invasive nature of such a procedure. Current non-invasive methods for measuring muscle forces use measurements of muscle activity to estimate force. These estimations are, however, prone to inaccuracies, as muscle activity is not always measurable for the complete set of muscles acting around the joints of interest.In this paper, we present a novel estimator that integrates a forward dynamics estimation approach with knowledge of the optimal contraction strategy to obtain accurate estimates of individual muscle force when measurement of muscle activity is not available for all muscles. We show that the system improves accuracy over standard estimation methods.

Neuromuscular diseases require a continuous assessment of physical performances by physiotherapists at clinics. However, ICT-based solutions to help the operators in this pivotal task are lacking. Here, we describe a dedicated ICT platform, specifically designed for neuromuscular diseases, able to collect data from validated assessments, including the PUL, the 6MWT, the MFM and the NSAA. The help of physiotherapists was fundamental in its development and the first related feedbacks from the operators on its usage were extremely positive, making PhysioTest a promising tool for present and future use in this specific domain.

The progressive miniaturization of electronic devices and their exponential increase in processing, storage and transmission capabilities, is leading to unprecedented scenarios as the Internet of Things (IoT). Although most of the recent biomedical research applications of pervasive technology focused on remote monitoring and diagnostics, Smart Objects may enable novel quantitative approaches in rehabilitation too. In this paper, we present some preliminary results regarding a pediatric rehabilitation protocol based on Smart Objects and biofeedback, administered to a small group of hemiplegic children. Despite the very limited number of treatments, all children enjoyed participating in the study, and the preliminary results represent an interesting starting point to fuel the scientific and clinical discussion towards what could be defined “Pediatric Rehabilitation 2.0”.

In this study we investigate the effects of motor adaptation and motor learning in persons with Parkinson’s disease during an home-based gait training program built on auditory biofeedback (BF). For this purpose, we assessed the motor response produced immediately after auditory BF messages and analysed it at multiple time-scales, within and between all training sessions. The findings indicate that motor adaptation is possible in a home-based training context for persons with PD using a wearable BF system.

Computed gait analysis is a valid tool able to assess human walking. The joint angle variation during the cycle consists in a function. The derivation process, producing angular velocity and acceleration, may reveal and quantify the small angle variations. This approach may support physician’s interpretation of the joint movements. We would like to propose a calculation, combining the data from angular velocity and angular acceleration, to obtain a graph showing some features of joint behavior. We compared the results among a healthy subject, a coxarthrosis patient and a stroke patient. The built graph clearly shows the direction changes of the joint movements and the differences between the two sides and quantifies the acceleration/deceleration of a joint. The proposed approach is a mathematical artifice, which amplifies the information contained in the usual angle graphs and possibly opens the doors to more specific rehabilitation approaches.

This work presents a novel accelerometer-based platform, designed to answer to the increasing request from the clinical practice of new tools to objectively quantify movement quality and patient performances during the execution of functional rehabilitation tasks. Inertial data from a first cohort of stroke patients performing upper and lower limbs rehabilitation exercises have been collected using the new platform. The aim of this preliminary study is to correlate well-known metrics obtained from the data, with subjective scores given by therapists. The goal is to figure out if some of the metrics has the potential to reflect the visual inspection of trained and skilled therapists. Preliminary results show that energy expenditure, smoothness and acceleration variability appear to match therapists’ evaluations. Additionally, energy expenditure could have the potential to reveal rehabilitation progresses. Future analyses on a larger dataset are already planned for stronger evidence of these early results.

The aim of the present study was to investigate whether anodal transcranial direct-current stimulation (tDCS) combined with attention to the paretic hand can improve hand performance in stoke patients. Eight chronic stroke patients participated in the double-blind, crossover, sham-controlled study. There were three intervention conditions: (1) anodal tDCS was applied over the hand motor cortex (M1) in the affected hemisphere when the patients paid attention to their paretic hands; (2) anodal tDCS was applied without the patients’ attention to the hand; (3) attention was paid to the hand without real tDCS. The task performance of the upper-limb motor function one week after the interventions was significantly better in the combined tDCS and attention condition than in the tDCS only condition. We conclude that tDCS over the M1 in conjunction with attention to the paretic hand in stroke patients improves the performance of the hand.

Although recent advance of Brain Machine Interface (BMI), it was less clear what is a critical brain activation in relation to imaging one’s own body movement. By comparing the electromyography (EMG) signals related to primal motor cortical areas together with the somatosensory area (parietal cortex) by electroencephalography (EEG), we found brain activity that precedes by some 200 ms in motor imagery task in the parietal cortex in comparison with the voluntary task of their lower ankle uplift. Such components would be a direct precursor to the motor commands in the context of the skeleton muscle model.

Paired Associated Stimulation (PAS) protocol has been used to enhance motor cortico-spinal pathway excitability. We explored the consequences of using PAS protocol synchronized with a dynamic task that causes fatigue in the main muscle responsible for the task execution, Abductor Pollicis Brevis (APB). Results showed an excitability depression in APB and an enhancement in the control muscle, Abductor Digiti Minimi (ADM). This suggests that ADM became the target of synaptic plasticity processes, as a consequence of APB fatigue. We conclude that fatigue can modulate the directionality of plastic changes and therefore, the target muscle.

Trigeminal input to noradrenergic Locus Coeruleus (LC) neurons is important for the maintenance of arousal and may boost cognitive performance. Since it has been shown that LC enhances discrimination of acoustic stimuli and the associated electroencephalographic (EEG) waves, such as event related potentials (ERPs) P300, we set a protocol aimed to verify whether trigeminal nerve stimulation (TNS) may be utilized for improving acoustic discrimination of different tones. In two separate experiments, we evaluated P300 ERPs elicited by an acoustic oddball paradigm performed before and after TNS or sham-TNS, respectively. The ultimate purpose of this evaluation is to verify whether it is possible to exploit TNS as a treatment in subjects with disorders of consciousness.

The development of systems for supporting neuro-rehabilitation is of primary importance, due to the high number of people in need of rehabilitation and the limited effectiveness of most of the current developed systems. Our research work aims at developing more engaging interaction modalities for neuro rehabilitation systems, through virtual reality, music based on harp therapy and fragrance feedback modalities and which are also fun and motivational for the patients. The proposed interaction modalities consist of a set of virtual immersive environments which includes an olfactory feedback, where odours are used to increase the sense of presence and the attention of the patients during the execution of the exercises. While the patient performs the rehabilitation exercise, the harp therapist plays the harp accordingly to the patient emotional condition. The system shows a virtual scenario, including virtual objects and/or 360 $$^{\circ }$$ videos used to increase his sense of presence in the scenario. Odours are associated with virtual scenarios.

This work presents preliminary results on the association between central and peripheral expressions of muscle fatigue induced by unresisted repetitive movements. We tested cortico-spinal excitability and intra-cortical inhibition right at the end of 30 s of maximal rate finger tapping (ft) or after 10 s of rest; the contractile properties of the muscle were also tested. This procedure was repeated 12 times. In half of the repetitions, the evaluation was done during induced muscle isquemia. In all cases ft rate decreased during the 30 s of task. Isquemia produced a slowing of muscle contractile properties in all cases, after ft as well as after 10 s of rest post-ft. Intracortical inhibition increased immediately after ft, but recovered after 10 s, regardless the presence of muscle isquemia. Our results suggest that the increment of inhibition in M1 after fatiguing repetitive movements is central in origin, and not an adaptation to the slowed contractility of the muscle.

CIMT (Constraint Induced movement Therapy) is based on two basic principles. Forced used of the affected extremity by restraining the unaffected extremity and massed practice of the affected extremity. We have introduced special day program with CIMT for people after brain injuries on our department.Method: 16 patients with hemiparesis due to brain injury were involved in program with lower limbs (Group A) and 16 patients in program with upper limbs (Group B). They had good cognitive functions to understand and follow the tasks. They had activities using CIMT principles for 6 h a day for four weeks (from Monday to Friday). The evaluation of the effect was done using standardised functional.Results: The results are very promising. 14 patients in each group finished the programme, almost all of them were much better after the programme, 85% were better even after three and six months after finishing it.

Linear mixed effects models were used to describe the dynamics of M1, spinal and muscle excitability during index finger tapping at the maximal possible rate, for 3 min. Our results show that tapping rate and amplitude decreased, following a triphasic pattern that seems to evolve parallel to changes in excitability measured by transcranial magnetic stimulation and electrical stimulation along the cortico-muscular axis.

This paper presents a protocol appropriate for monitoring plastic changes following bio-robotic interventions for rehabilitation of gait in individuals suffering from spinal cord injury. The protocol includes methods for monitoring descending motor and ascending sensory pathways, intrinsic spinal circuits and the level of functionality of a spinal cord injured individual. It is expected that the usage of the presented protocol and the considerations presented in this paper, would lead to improved understanding of the plastic changes across the neuroaxis in relation to bio-robotic based rehabilitation, and ultimately lead to a better understanding of the relation between the plastic changes and the therapeutic effects thereof.

Transcranial magnetic stimulation (TMS) is one of non-invasive brain stimulation methods, and is often utilized for rehabilitation combined with repetitive facilitative exercise. The brain activity evoked by TMS was analyzed as functional Near Infrared Spectroscopy (fNIRS) signal from frontal robe. We found that prefrontal blood flow increased by TMS to the motor cortex. In addition, the prefrontal activity was enhanced by observing a video content of an exercise behavior of others, considered to activate mirror neurons. These results suggest the possibility of improving the effect of rehabilitation using TMS by visual priming.

We describe the data processing employed to match EMG signals with the categorized behavioral movements obtained from video recordings in awake rats while reaching and grasping pellets. We want to determine whether or not, the forelimb muscles follow a temporal synchronized pattern of activity during reaching and grasping. Data processing included raw data filtering and obtaining the first derivate. We found consistency in this procedure between attempts, indicating the utility for studying the brain to spinal cord function underlying volitional and skilled movements in the rat.

Sensorimotor Brain-Computer Interface (BCI) systems can be beneficial for post-stroke motor recovery. A successful trial on subacute stroke patients carried out at Fondazione Santa Lucia demonstrated clinical and neurophysiological benefits derived from BCI-supported motor imagery (MI) training of the upper limb. A further translational effort led to the implementation of the Promotœr, an all-in-one BCI-supported MI training station dedicated to patients with upper limb motor impairment due to central nervous system injury of different etiology. The BCI training is delivered in add-on to standard rehabilitation therapy. We present here the results on 12 chronic stroke patients who underwent clinical, neurophysiological and neuropsychological evaluation before and after such training. Results are promising in terms of feasibility of a BCI training in the context of a real rehabilitation program and in terms of clinical benefits observed in the patients.

Lifestyle and risk factors affect the development of diseases, cognitive decline and loss of autonomy during ageing. In the Italian Smart Health 2.0 project, we developed the SmartAging program offering innovative ICT solutions, grounded upon the Italian Telecom platform (“Nuvola Italian Home Doctor”), to contrast cognitive decline and support “active” ageing. Specifically, the SmartAging program includes tools for daily telemonitoring at subjects’ home of: (i) eating behaviour, (ii) physical exercise, (iii) tobacco consumption, (iv) cultural, social and relational activities, (v) cognitive functions, (vi) weight, temperature, and blood pressure. Furthermore, a Point of Care device and a Self-Assessment Health Station (i.e. health medical equipment), in proximity to the subjects’ home, allowed periodic measures of blood biomarkers and physiological indices. To demonstrate the feasibility of this program, we performed a clinical proof of concept study for 16 weeks on 10 elderly subjects (5 being oncological long-term survivors). All subjects expressed a need for education and assistance concerning a healthy lifestyle. Most of them expressed a preference for ICT-based home programs to improve the lifestyle over institutional programs. All expressed satisfaction with the SmartAging program in terms of simplicity, clarity, adequacy and reported a positive impact on their quality of life. These results suggest that the SmartAging program fulfills all the requirements to be considered an innovative technological approach for the lifestyle education and telemonitoring of seniors, to prevent dementing disorders. Some concepts of the SmartAging program have been further developing in the H2020-MSCA-ITN-ETN BBDiag project.

An associative brain-computer-interface (BCI) that correlates a peripherally generated afferent volley with the peak negativity (PN) of the movement related cortical potential (MRCP) induces plastic changes in the human motor cortex. The aim of the current study was to compare the effectiveness of this intervention when the MRCP PN time is pre-determined from a training data set (BCIoffline), or detected online (BCIonline). Ten healthy participants completed both interventions in randomized order. The mean peak-to-peak motor evoked potential (MEP) amplitudes were significantly larger 30 min after (277 ± 72 µV) the BCI interventions compared to pre-intervention MEPs (233 ± 64 µV) regardless of intervention type and stimulation intensity (p = 0.029). These results provide further strong support for the associative nature of the associative BCI but also suggest that they likely differ to the associative long-term potentiation protocol they were modelled on in the exact sites of plasticity.

Hybrid brain-machine interfaces (BMIs) combining brain and muscle activity are a promising therapeutic alternative for rehabilitation of stroke patients with severe paralysis. In this study, we compare different approaches utilizing electroencephalographic (EEG) and electromyographic (EMG) activity to detect movement attempts of stroke patients with complete hand paralysis. Data of 20 patients with a chronic stroke involving the motor cortex were analyzed, and the performance of EEG-based, EMG-based or hybrid classifiers were simulated offline. We show that the combination of EEG and EMG improves the accuracy of movement detection, but that muscles unrelated to the task can also provide high accuracies, reflecting compensatory mechanisms. This result underscores the importance of appropriate designs of hybrid BMIs to maximize their rehabilitative potential.

Various experimental strategies using brain-machine interfaces, both for the upper limb and the lower limb, have managed to generate plastic changes in the nervous system aiming to rehabilitate diseases involving movement restrictions. Due to the variability in the results of previous studies and the lack of experiments with associative facilitation, more interventions are necessary with functional tasks, involving different lower limb muscles, trying to rehabilitate patients with these pathologies. In this study, we present data of a stroke patient in which an intervention with a cycling task was studied. The intervention consists in a brain machine interface, which has been integrated driving a functional electrical stimulation device. Increases in corticospinal excitability of both Tibialis anterior muscles of the patient were observed after the intervention. Such results could imply that the brain machine interface would be behind these changes, which would help to rehabilitate these patients.

Rehabilitation procedures require an objective evaluation of motor recovery to optimize their functional outcomes. Here, we propose an innovative framework for the identification of neural biomarkers of motor recovery in the forearm muscles of tetraplegic and stroke patients. The system incorporates a custom-made ergometer and high-density surface EMG recordings on the flexor and extensor finger/wrist muscles of the forearm. Blind source separation was used to identify motor unit activity during isometric maximal voluntary contractions. Preliminary results on two patients show a significant increase in the average discharge rate of the identified motor units after some weeks of rehabilitation training. Moreover, the relative change of this parameter matched with the motor recovery exhibited by the patients. This study demonstrates that the identification of individual motor units from the forearm muscles can provide a reliable neural biomarker of motor recovery.

The development of usable and accurate brain-computer interface (BCI) systems enables the transfer of this technology to clinical routine. When working with electroencephalographic signals (EEG), an important factor to optimize the signal to noise ratio of the signal is to choose the appropriate spatial filters. Specific aims of this study were (a) to compare classification performances of two commonly used filters, two bipolar filters (longitudinal and transversal) and the combination of both bipolar filters obtained by pooling EEG features together, (b) to compare the number of physical electrodes needed as consequence of the spatial filter choice. Bipolar filters showed classification performances comparable to those provided by the most commonly used filters, despite requiring a significantly lower number of electrodes. Longitudinal bipolar filters showed the best accuracy to number of electrodes ratio; thus, its usage is suggested for applications such as upper limb motor rehabilitation in post-stroke patients.

A novel FES rowing system is presented that facilitates a near normal rowing style. The system is presently based on the Concept2 ergometer. He we present an error analysis the C2 performance monitor which suggests that a more accurate monitor will be required for medal placement in competitive rowing events.

Functional electrical stimulation (FES) can be used to support walking and cycling in spinal cord injured individuals. In the current contribution, we present a new method that, for the first time, enables FES-supported swimming in paraplegics. The proposed setup includes a waterproof stimulator, cables, and electrodes. In preliminary experiments, flexion and extension movements of the knee could be generated to support the propulsion during crawling. A synchronization of the voluntary arm and leg movements showed to have a stabilizing effect on the body position of the swimmer in the water, especially on the roll angle. To enable such a synchronized swimming, an electrotactile feedback algorithm was developed that informs the swimmer about the leg movement by a stimulation of sensory unimpaired regions at the back of the subject at a sensory level. The new setup and methods are currently being tested during the STIMSWIM pilot study with paraplegics. In first preliminary results, an improvement of swimming velocity compared to non FES-assisted swimming was observed in the two first subjects.

In this study, a novel approach is proposed to improve stance phase support symmetry recovery in acute post-stroke individuals, based on knee joint control. Functional electrical stimulation (FES) is delivered to the quadriceps and hamstrings muscles of the paretic limb based on the online estimation of knee angle and support phase. This paper is a proof-of-concept article, which introduces the global approach and some preliminary results obtained on two participants with post-stroke hemiplegia.

We describe here our progress in developing a rat model for cortically controlled functional electrical stimulation (FES). In these experiments we are using neural activity recorded from motor cortex to drive stimulation of paralyzed muscles in order to restore voluntary motor function following spinal cord injury (SCI). We first evaluated the ability of different cortical interfaces to estimate intended movements and muscle activation, determining whether intracortical interfaces were necessary. We found that spiking activity from intracortical electrodes predicted movements with reasonable accuracy, whereas less invasive epidural electrodes performed considerably worse. We are currently evaluating the utility of these approaches to restore function following SCI as well as their ability to rehabilitate function with repeated training.

Cycling induced by Functional Electrical Stimulation (FES) is a promising rehabilitative approach for stroke patients, mainly in the post-acute phase. This systematic review summarizes the current evidence about its effectiveness. Seven randomized controlled trials involving 221 subjects were found. Overall, a slight trend but no significant differences were achieved in favor of FES-cycling in walking speed and muscle strength. Further high-quality studies are advocated to derive an evidence-based conclusion about the effectiveness of FES-cycling in post-acute stroke patients.

Sitting Pivot Transfers (SPT) are key factors for enabling independence of individuals with paraplegia. However, repetitive instances of SPT may lead to overload of upper limbs, possibly causing pain and injury. In this work, we explore the use of electrical stimulation during the lift pivot phase in order to reduce the overload on upper limbs. First, an experimental set-up was developed to investigate the use of multiple interface modalities for controlling the applied electrical stimuli. Then, the set-up was used in a study involving 5 subjects with paraplegia. The results indicate the load reduction on upper limbs.

Recent studies demonstrated that stroke patients with large lesion to the cortico-spinal tract (CST) show secondary degradation of ipsilesional white matter tracts and poor motor recovery (POOR) in the subacute period. This suggests that white matter preservation might be an interesting target for this group of patients and might help improve motor outcome. Non-invasive brain stimulation (NIBS) can boost motor function in stroke patients, but we lack information on its impact on white matter microstructure. Therefore, we conducted a randomized controlled trial comparing the effects of cathodal transcranial direct current stimulation (ca-tDCS), continuous theta burst stimulation (cTBS), or sham stimulation on white matter degradation and motor improvement, as measured with Diffusion Tensor Imaging (DTI) and standardised motor assessments. Results showed preserved structural connectivity after ca-tDCS compared with the other experimental groups in POOR patients. This correlated with better clinical recovery.

Motor deficit is a prominent feature among stroke survivors. Robot-assisted therapies have been proposed as a strategy to boost rehabilitation, by allowing therapy to be provided in a more reproducible and intense manner, while quantitatively monitoring patient’s improvement. However, those approaches have so far not shown superiority over conventional treatments. One potential solution to reach better outcomes would be to personalize the treatment. In this regard, a better understanding of the mechanisms underlying motor recovery is pivotal to tailor therapy to each patient. Here, we explored the cortical changes occurring during robotic training. We recorded resting-state fMRI before and after the treatment in three sub-acute post-stroke survivors, and we investigated the functional connectivity between motor regions. We observed a cortical reorganization following training, consistent with motor improvements.

Stroke is a devastating neurological condition, often causing severe functional and cognitive deficits, sharply diminishing the patient’s quality of life. Among others, robot-assisted rehabilitation has been widely proposed to enhance the rehabilitation outcome. However, clinical scores and robotic parameters often used to inform rehabilitative-decision process are unfit to fully describe the neural reorganization that occur after a brain insult. The lack of reliable, simple, and sensitive neural biomarkers has potentially limited the clinical translation of these advanced rehabilitative technologies. Here, we show that EEG-topographic measures can be extracted as robust and sensitive biomarkers of stroke recovery to inform robotic therapies.

Supporting mental training practice with non-invasive wearable technologies is a hot topic in applied research on the promotion of well-being and self-enhancement. To test the potential for strengthening stress management skills of an intensive neurofeedback-supported mindfulness protocol, we devised a longitudinal controlled trial. Statistical comparison of training-induces changes of subjective and psychophysiological measures of stress highlighted both a significant reduction of subjectively perceived stress and a consistent increase of cardiovascular correlates of vagal tone in the experimental group with respect to the active control group.

According to the recent studies on the psychophysiology of touch, a haptic effector designed in a neuromorphic way was projected, designing an electronic card as to be able to deliver variable signals over time and in intensity. The two-dimensional arrays of micro-actuators were made either with planar geometry or with three-dimensional, semi-spherical or “dome” geometry. Subsequently, on both the behavioral and the electrophysiological level the haptic sensation received by the effector was evaluated on 6 subjects and compared to real stimulations of different grains (Paper). During the various stimulations the subject was in a state of Resting State (RS). Each stimulation had a frequency range ranging from 2 to 500 Hz on 2 and 5 s. Analysis of behavioral responses and the alpha rhythm in RS showed significant differences for low frequencies vs Paper. RS highlighted differences in ROIs on the various frequency distributions, especially low frequencies in Frontal ROI. This pilot study indicates that the best frequencies for a haptic simulation are between a range from 20 Hz to 166 Hz.

Epilepsy is often associated with modifications in autonomic nervous system, which usually precede the onset of seizures of several minutes. Identifying those changes is pivotal to predict the onset of seizure and to set up an early intervention. The aim of this study was to develop a patient-specific approach to predict seizures using electrocardiogram. Time- and frequency- domain features as well as recurrence quantification analysis variables, were extracted from the RR series. A machine learning approach based on support vector machine was then applied to predict seizures. The dataset consisted of 12 patients with 38 different types of seizures. An average sensibility of 83.8% and specificity of 72.8% were obtained. The results of the proposed approach show that it is feasible to predict seizure in advance, considering patient-specific, and possibly seizure-specific, characteristics.

Here, we tested that healthy elderly (Nold) and Alzheimer’s disease (AD) individuals can be discriminated with a moderate accuracy using resting state eyes-closed electroencephalographic (rsEEG) markers. Eyes-closed rsEEG data were collected in 100 Nold and 120 AD subjects. eLORETA freeware estimated the source current density (SCD) and functional connectivity (lagged linear connectivity, LLC) in frontal, central, parietal, occipital, temporal, and limbic regions. Delta (2–4 Hz), theta (4–8 Hz), alpha 1 (8–10.5 Hz), and alpha 2 (10.5–13 Hz) were the frequency bands of interest. Univariate (i.e., single rsEEG marker with receiver operating characteristic, ROC, curve) and multivariate (i.e., multiple rsEEG markers with artificial neural networks, ANNs) classifiers were used. The best accuracy was of 76% with univariate classifiers and 77% with multiple classifiers. The present results suggest that both univariate and multivariate rsEEG classifiers allowed a moderate classification rate between Nold and AD individuals. Interestingly, the accuracy based on multiple rsEEG markers as inputs to ANNs was similar to that obtained with a single rsEEG marker, unveiling their information redundancy for classification purposes. In future AD studies, multiple rsEEG markers should also include other classes of independent linear (i.e. directed transfer function) and nonlinear (i.e. entropy) variables to improve the classification.

Patients with age-related mild cognitive impairments could be better assisted in hospitals and nursing homes where innovative systems provide personalized cognitive rehabilitation or physical and recreational activities.The Apulian project named RESCAP aimed to the quantitative evaluation of the EEG event-related responses to visual and audio stimulations induced in a virtual environment and then designed innovative domotic technologies to implement into real-life solutions. In particular, two protocols for cognitive and emotional evaluation were implemented, and quantitative results were obtained by means of pattern analysis and recognition processed with statistical and machine learning algorithms methodologies.

The assessment of the motor performance is an essential component of rehabilitation of humans with a disability. The wearable system is a preferable solution compared to the laboratory-based instrumentation for the clinical use. We integrated a system which combines array electrodes over the muscles of interest connected to the 24-channel digital amplifier and the shoe insole with five robust pressure transducers and one inertial measurement unit per leg. The system sends signals wirelessly to a computer. The graphical interface is simple for usage and simple to interpret. The EMG patterns (muscle activities) and gait events are synchronized by the software. The EMG patterns are determined by cubic spline interpolation of data from 24-pads over the muscle. The gait events are estimated by fuzzy logics applied on the thresholds from signals coming from the insole. The system was tested and accepted by the therapist as a useful and easy clinical tool.

The effects of electrode size, interelectrode distance and grid size are discussed for HDsEMG electrode arrays.The issue of electrode-skin noise is addressed. Three HDsEMG applications are described. The widening gap between technological progress in sEMG and education/training of clinical operators is discussed.

Fatigue is a disabling symptom that constrains motor performance. It depends on the physiological capabilities that establish the level of fatigability for a given task and the physiological and psychological factors that contribute to perceptions of fatigue [1]. The relative influence of the potential factors that can contribute to fatigue depends on the details of the task being performed and the characteristics of the performer. This presentation reviews our current knowledge of the potential mechanisms that can limit the performance of the neuromuscular system during sustained activity.

Achieving the functional grasp by electrical stimulation using surface electrodes is a demanding task. The innervations of muscles come via ulnar, radial and median nerves. The anatomy of nerve branches connecting various muscles in the forearm differs significantly between individuals. We hypothesize that the anatomical differences between the paretic and nonparetic arms are minimal. Based on this assumption we developed a method where the differences of muscle activities (EMG) between the healthy and paretic arms recorded by the 24-contact electrode within an array define the target zones to be stimulated on the affected forearm. We used special electrode where magnetic contacts allow simple change of the stimulation pads. The examiner positions the magnetic contact on the pads where the EMG differences are maximal. The stimulator delivers asynchronous stimulation to the selected pads. We proved that the method is working in stroke patients by measuring joint angles and the grasping force.

We systematically assessed the impact of motor unit action potential (MUAP) compensation on muscle excitation estimation from high-density EMG (hdEMG). For this purpose, we used experimentally recorded hdEMG signals from biceps brachii to estimate the MUAPs of 200 MUs at different elbow angles. We then used these MUAPs to generate synthetic EMG signals with known muscle shortening and muscle excitation profiles. Both constant and force varying isometric muscle contractions were analyzed. Novel metric for muscle excitation estimation, so called global activity index (GAI) was introduced and compared to cumulative motor unit spike train (CST), identified by Convolution Kernel Compensation (CKC) method and to spatial average of root-mean-square (RMS) value of hdEMG signals. The results of all three metrics were compared to the CST of all the simulated motor units by calculating normalized RMS error (NRMSE). Processing costs of investigated methods were assessed by measuring the processing time on a personal computer with CORE i7 processor. The results demonstrate that the GAI significantly outperforms the RMS metric and is comparable to the CKC CST at significantly lower computational costs.

Impaired muscular activation causes pain, disability and loss of quality of life. To preserve and to restore movement performance is a challenge, and technologies enabling more effective treatment have gained high importance. Surface-Electromyography (sEMG) technologies allow the pain-free assessment of muscular activation, but they have rarely been taken out from the laboratory into everyday life, yet. This crucial step can only be taken if sEMG achieves high acceptance by the end-users. The challenge is to identify, adjust or develop sEMG tools, signal processing strategies, and application procedures enabling sEMG to meet the expectations of physicians, therapists, and patients. Two examples are given showing how integration of biomechanical knowledge improves interpretation of sEMG in dynamic contractions and helps to gain clinical relevant information.

The contemporary myoelectric prostheses are advanced mechatronic systems, but human-machine interfacing for robust control of these devices is still an open challenge. We present a novel method for the recognition of user intention based on pattern classification which is inspired by the natural coordination of multiple muscles during hand and wrist motions. The coordinated muscle activation produces a characteristic distribution of the amplitude features of the electromyography signals, and the novel method establishes the class boundaries to capture this natural distribution. The method has been tested in healthy subjects operating a prosthesis during a challenging functional task (bottle grasping, turning and releasing). The novel approach outperformed the commonly used benchmark (linear discriminant analysis), while using shorter training and fewer features. Further developments can, therefore, lead to a method that is suitable for practical implementation and allows robust and efficient control.

When manipulating objects, our sensorimotor system integrates information from multiple sensory streams. An important stream is the force information that is sensed by two modalities in our body – kinesthetic and tactile. In this study, we examined the integration between kinesthetic and tactile information during adaptation to a force perturbation. We exposed participants who performed reaching movements toward targets to a velocity-dependent force field. For one of the groups, we added artificial tactile stimulation by stretching the skin of their fingertips in the same direction of the applied force. We found no difference between the kinematics of two groups, but the skin-stretch affected their manipulation and grip forces. These results suggest that tactile and kinesthetic information may have distinct mechanisms of integration with kinematic states during force field adaptation.

Prior work has shown that independent motor memories of opposing dynamics can be learned when the movements are preceded by unique lead-in movements, each associated with a different direction of dynamics. Here we examine generalization effects using visual lead-in movements. Specifically, we test how variations in lead-in kinematics, in terms of duration, speed and distance, effect the expression of the learned motor memory. We show that the motor system is more strongly affected by changes in the duration of the movement, whereas longer movement distances have no effect.

Dysfunctional muscle activations are observed after stroke and in other neurological disorders. We propose an approach to rehabilitation exercises based on conditioning the availability of visual feedback upon the attainment of an established pattern of muscle activation. As a proof of concept, we tested this approach on unimpaired participants who were asked to perform reaching arm movements while holding a robotic manipulandum. A cursor provided participants with visual feedback of their hand position. The cursor visibility was set in proportion to the RMS EMG activity of the triceps, varying from transparent to maximally visible within a range corresponding to the desired muscle activity level. The great majority of the subjects increased significantly the activity of the triceps as they practiced the reaching task however this returned to baseline at the end of training. We expect that changes can be retained after more intense training if they have functional relevance.

It is important to evaluate and maintain driver’s sense of agency (SoA), because poor SoA of assisted driver may result in slow and inaccurate response in case decisions are required from the driver. This study investigated the relationship between SoA and alpha-band power of EEG in a simulated driving environment.

Several daily life tasks require us to coordinate the motion and the force produced by both hands, using our position sense as well as sense of effort. However, the interaction between these in bimanual tasks is not extensively investigated. In this work we focused on bimanual tasks where subjects were require either (1) to place the hands in a same position (identical configuration in joint space) under different loading conditions, or (2) to produce an equal isometric force with the two hands with different arms configurations. The hand motion and force required for accomplishing the task were constrained to the vertical direction. In the first task, the difference between the hands positions was greater when the two hands were loaded with different weights. Conversely, in the second task, when subjects were asked to exert small equal forces with both hands, the systematic difference between left and right hand force was not influenced by symmetric or asymmetric arm configurations, but by the absolute position of the left arm, regardless of the right arm position.

Our group recently developed a novel fMRI-compatible wrist robot, the MR-SoftWrist, to study the neural processes that underlie force-field adaptation. Here we present our first fMRI pilot study on four healthy subjects. Our study validates the MR-SoftWrist as a tool for investigating motor adaptation of wrist pointing movements during fMRI and establishes neural activations associated with active motor control and off-line neural processing in three dynamic conditions.

Non-invasive motor cortex stimulation may result in long-term potentiation (LTP)-like plasticity of corticospinal excitability, and this may be useful to support neurorehabilitation after lesion, such as in stroke. However, the reported plasticity effects show large interindividual variability, and even intraindividual reliability is moderate at best. One possible strategy to improve size effect and consistency is to couple pulsed transcranial magnetic stimulation (TMS) to the endogenous brain state. We show here that instantaneous brain states measured with EEG have significant impact on TMS-induced corticospinal excitability. Consistent stimulation on the negative peak of the ongoing µ-rhythm results in LTP-like plasticity in 21/23 subjects, while stimulation at the positive peak had no effect. Findings raise the intriguing possibility that real-time information of instantaneous brain state can be utilized to control efficacy of plasticity induction in humans, and this may be utilized in clinical settings to support therapeutic reorganization of brain networks.

Brain-computer interfaces have increasingly found applications within the rehabilitation of lost motor function in stroke patients. Most studies have targeted upper limb muscles and used sensorimotor rhythms as the control signal. In a series of studies, we have introduced an associative BCI modeled on known theories of memory and learning that implements the movement related cortical potential (MRCP) as a way to control an external device that provides afferent generated feedback to the user’s brain at the time of the peak negative phase of the MRCP. In its application to lower limb muscles it demonstrates significant plasticity induction that requires no user training. In the current study, we tested if this associative BCI is effective when targeting upper limb muscles. Further, we explored if there is a difference when the MRCP is generated as part of a simple (wrist extension) versus a complex (reach and grasp) movement.

Artificial activation of peripheral afferent fibers, with the resulting sensory feedback timed to arrive at the peak negativity of the movement-related cortical potential, induces significant increases in the excitability of cortical projections to the target muscle in healthy individuals and chronic stroke patients. In the currently ongoing study, we applied this associative brain-computer interface paradigm to sub-acute stroke patients. Compared to a sham group, where the peripheral electrical stimulation intensity was below the activation threshold of the sensory afferents, the associative intervention group displayed substantial increases in corticospinal excitability to the target muscle (tibialis anterior).

Different directions of transcranial magnetic stimuli can selectively activate independent sets of local interneuron circuits in the human motor cortex. The differences in the activity of these circuits during motor initiation of voluntary planned movements (i.e., the last tens of ms before a movement is released) have not been studied so far. Here we test if different excitability modulation effects can be observed after repeatedly associating postero-anterior (PA) or antero-posterior (AP) transcranial magnetic stimulation (TMS) with cortical states of motor initiation. Repeatedly stimulating the motor cortex with TMS resulted in a significant reduction (p – 0.02) of MEP amplitudes induced with PA-TMS and AP-TMS. No significant differences in MEPs were observed with different TMS directions. This research line is expected to provide relevant knowledge regarding the possible functional roles of cortical circuits in preparation for voluntary movements and also to improve brain state-dependent stimulation protocols aimed to guide targeted plastic changes.

This study presents the modulation of the cortical excitability by the association of the voluntary motor-related cortical patterns with the assistance provided by a hybrid robotic system during the execution of reaching movement. The cortical excitability was assessed before and after the intervention measuring the peak-to-peak amplitude of the Motor Evoked Potentials (MEPs) induced through transcranial magnetic stimulation pulses. Eight healthy subjects participated in the experiments. Results showed an overall and distributed increase in the cortical excitability as a result of the proposed intervention.

High-frequency repetitive transcranial magnetic stimulation (rTMS) of the left dorsolateral prefrontal cortex (DLPFC) shows therapeutic potential in pharmaco-resistant patients with major depression. However, clinical efficacy is limited by high inter-individual variability and low response rates. One possible strategy to improve the effect size and consistency may be brain state dependent brain stimulation, i.e. coupling of TMS pulses to the endogenous brain states as reflected by the instantaneous oscillatory brain activity. Here we present findings from a proof-of-principle study of alpha-oscillation synchronized brain stimulation of the frontal cortex in patients with major depression (BOSSFRONT). Repetitive stimulation consistently on the negative peak of ongoing alpha activity in left DLPFC, but not brain state independent intermittent theta-burst stimulation (iTBS), resulted in suppression of resting-state alpha activity in left DLPFC and an increase in TMS-induced beta activity. Findings show that alpha-synchronized rTMS of left DLPFC is both feasible and safe, and suggest that it interferes with frontal brain networks important in the pathophysiology of major depression.

Brain-computer interfaces (BCI) has being used to treat and assist neurologic patients to make the activities of daily living through electroencephalography (EEG) signals. Real movement or imagery movement produce changes in power of sensorimotor rhythms, called event-related (de)synchronization for the supression or for the increase of oscillatory activities. Also, changes between functional connectivity of cortical areas have been reported, evaluated by different recording techniques. Using the phase-locking-value index (PLV), we compared the changes in functional connectivity with the changes in power of surface EEG of ten healthy subjects during a BCI task using motor imagery. Modulation of functional connectivity related to task and to the discriminate band frequencies were studied, what might have implications for improving control of BCI systems.

Pathological tremor is the most prevalent movement disorder and affects daily living activities. Research on tremor suppression over the past five years suggest that transcutaneous stimulation below the motor threshold (also known as afferent or sensory stimulation) has a powerful inhibitory effect on the descending supraspinal tremorogenic input by means of spinal interneurons. The aim of this study was to review the most recent and promising strategies to suppress tremor using afferent electrical stimulation in two main pathologies: Essential Tremor (ET) and Parkinson’s Disease (PD). Five studies were retrieved from PubMed database. Most of these studies reported at least 40% of tremor reduction during afferent stimulation and one article reported a suppression effect five minutes after stopping the stimulation. More research on how to maximize the suppression lasting effect as a therapeutic tool is required.

We present the feasibility of a complete rehabilitation system based on brain computer interface (BCI) triggered functional electrical stimulation (FES) and avatar mirroring to improve the function of the paretic limbs. The system was tested on two chronic stroke patients with 25 BCI training sessions over 13 weeks. The Upper-Extremity Fugl-Meyer Assessment (FMA-UE) and the Modified Ashworth Scale (MAS) were improved in both patients, however, one patient showed more improvement than the other, for whom the longer time since the stroke (31 years vs. 4 years), no residual motor function in the paretic wrist, and poor BCI performance may have limited recovery. Randomized controlled studies are necessary to show the effectiveness of the BCI-FES-avatar system with a larger sample size and matched parameters between test and control groups.

Brain-machine interfaces (bci) translate brain activity into control signals of external devices, such as robots, prostheses or computers. A well-established bci paradigm uses signal power modulations of fast rhythmic brain activity. Such power modulations are linked to a broad variety of sensorimotor, cognitive and perceptual tasks, and feedback for the user can be provided by different sensory modalities, so we decided to investigate whether different sensory modalities of feedback might differently modulate the electroencephalography (eeg) during a bci task. Ten healthy volunteers performed bci motor imagery session while controlling a hand exoskeleton. Participants received feedback with different sensory modalities: visual, somatosensory (using a hand exoskeleton) or auditory. As expected, we found that cortical oscillations of eeg in beta frequencies were modulated by movements. Our main finding was that modulation of beta band in eeg was strongly increased by somatosensory feedback using the exoskeleton, a finding with important implications for design and implementation of bci experiments.

Pedaling wheelchair can be effective in rehabilitation training of lower limbs of paralyzed subjects by combination with functional electrical stimulation (FES). Since FES cycling training with cycling ergometer suggested that different effects were obtained under different cycling speeds, control of cycling speed is required for mobile FES cycling system in its use at rehabilitation. In this paper, closed-loop control of low cycling speed (0.2 m/s) of the mobile FES cycling was examined with 3 neurologically intact subjects, because a preliminary test of closed-loop control of low speed cycling showed backward propulsion. Two stimulation timing patterns based on EMG signals measured under different cycling speeds were tested in the closed-loop control. The stimulation timing pattern based on muscle activation timing measured under the cycling speed near the target speed was suggested to be effective.

The lesions induced by unilateral strokes perturb the complex and critical interhemispheric balance. While a high asymmetry measured in the acute phase is known to be a predictor for poor motor recovery, the evolution of this imbalance along motor recovery has not been studied. Here, we evaluated the evolution of the cortical power asymmetry during a robot-assisted motor task along a rehabilitation intervention. Preliminary results suggest that a reduction of the brain asymmetry towards values exhibited by healthy controls is associated with higher motor recovery.

This paper proposes the methodology to design a closed-loop system to control presence in a virtual rehabilitation setup, as a believed equivalence to therapeutic engagement and an improvement strategy of motor function recovery. This proposal starts with an analogy of adherence, defined by acceptance, participation and fulfillment (these last two indicators of engagement) with presence, measurable by means of subjective questionnaires and real-time biometrics (psychophysiology, biomechanics and task-performance). These objective measurements shape the feedback of a control system composed by a subject-state classifier and a controller based in soft computing to assure higher levels of presence via control rules that modify a virtual reality (VR) structure and code. This also describes the design of a previous open-loop setup that defines the signals that had better reflected emotions and presence states on task performance, after adjust VR variables that are classified by layers in a games design approach.

There is an increasing interest in the use of Gait Analysis (GA) in Parkinson’s Disease (PD), however no one has previously investigated what are the principal trends on the main applications of quantitative GA in studies involving this neurological disorder. We performed a systematically literature search for articles published through 2013 to present using three electronic databases. We retrieved 76 articles that met the inclusion criteria and identified four main research areas which refers to GA for: pathophysiological mechanisms underlying PD; assessment tool for treatment outcomes; automatic recognition of PD symptoms and algorithms for classification between PD patients and healthy subjects.

Intra-subject variability of the oscillatory activity in EEG signals limits the personal-adaptability of brain-computer interfaces for neurorehabilitation. The main object of this paper is to construct a fused classification model which is robust to the individual differences in the optimal frequency bands for classifying the spectral features into the dual or single tasks. The proposed decision fusion model results in the higher classification accuracy of 6%, compared to the averaged test accuracy of single classifiers using the best performing band as spectral features. Our study expands the usage of EEG spectral features for neuro-rehabilitation systems without selecting a specific frequency range depending on subject, task or environment.

Online myoelectric control involves two types of adaptation: computational adaptation, in which the controller learns to associate muscle patterns with performed forces; and behavioural adaptation, where the users learn the new interface, and adapt their motor control strategies based on the errors they observe. In order to study the behavioural motor learning during online myoelectric control, twelve able-bodied participants performed single and 2-finger presses through force and myoelectric control. Myoelectric control was obtained with linear ridge regression, and was based on a training set only containing single finger presses. The distance between muscle patterns of force and EMG control trials indicated that motor learning leads to changes in neural drive, even on the trained presses. This suggests that motor learning is an integral part of myoelectric control, where the ability of the user to learn the EMG-to-force mapping impacts the overall performance of the myoelectric controller.

We recently introduced a novel method for assessment of motor unit (MU) identification accuracy from high-density electromyogram (hdEMG), demonstrating that previously introduced Pulse-to-Noise Ratio (PNR) can be used to reliably assess the accuracy of a single MU firing identification. In particular, we can use previously identified MU firings as witnesses of accuracy when identifying a new MU firing. In this study, we systematically tested the number of witnesses needed in this process. For this purpose, we used experimentally recorded MU action potentials from biceps brachii to simulate isometric hdEMG signals with known MU firing patterns. Afterwards, we used Convolution Kernel Compensation (CKC) method to identify MU spike trains. By using leave-one-pulse-out paradigm, MU spike train and corresponding PNR were recalculated using 10, 20, 50 and 100 randomly selected true positive (TP) plus, optionally, one false positive (FP) firing in MU filter estimation. Regardless the number of TPs, the PNR dropped significantly whenever a FP was added to MU filter estimation, demonstrating that even 10 correctly identified MU firings may be used to assess the accuracy of every additional MU firing. This important result extends the recently introduced methodology for quality assurance in MU identification from hdEMG.

We analyzed the accuracy of different muscle excitation estimation techniques on a set of high-density surface electromyograms (hdEMG), recorded during dynamic contractions of biceps brachii muscle. The recorded hdEMG signals were decomposed by Convolution Kernel Compensation (CKC) technique and motor unit action potentials (MUAPs) of 250 identified MUs were identified at different elbow angles. The identified MUAPs were then used to generate synthetic hdEMG signals with known muscle excitation patterns. Cyclostationary CKC (csCKC) technique was then used to identify MU spike trains from synthetic hdEMG signals in isometric and dynamic conditions and cumulative MU spike trains (CST) were calculated using 5, 10 and 15 identified MUs, respectively. The normalized root-mean-square error (NRMSE) was calculated between the simulated excitation pattern and CTSs and the results were compared to CST comprising all the simulated MUs and to root-mean-square (RMS) value of hdEMG signals, averaged over all the hdEMG channels. On average, the RMS metric resulted in 14% error in muscle excitation estimation. The csCKC-based CSTs with 5, 10 and 15 MUs outperformed the RMS metric by ~4%, ~5% and 6%, respectively.

Notwithstanding recent advances in the detection and processing techniques, a new critical issue is emerging in surface electromyography: the spatial localisation of muscle activity. This issue has direct impact on the estimation of muscle activity, normalisation, cross-talk and myoelectric fatigue, and therefore on the correct interpretation of surface electromyograms (EMGs). Inferences prompting from surface EMGs must therefore be drawn circumspectly. This report briefly introduces the spatial localisation issue and highlights its implication for the use and interpretation of surface EMGs.

Electromyograms (EMGs) and ultrasound (US) images provide complementary information on muscle function. The integration of these two techniques may provide key insights into the electromechanical properties of skeletal muscles as well as into physiological adaptations due to ageing, pathologies, injuries or rehabilitation. From a technological point of view, the simultaneous acquisition of US and surface EMG from the same muscle region is challenging as the two detection systems may interfere with each other. This paper will describe these methodological issues and how they have been addressed to ensure high quality detection of both surface EMGs and US images. Finally, an overview on current and possible, future applications will be provided.

Multi-channel sEMG techniques open new perspectives in the non-invasive assessment of neuromuscular system but currently, can be applied only during isometric or dynamic tasks because of the encumbrance of existing sEMG acquisition systems and their sensitivity to movement artefacts. The aim of this work is to describe a new, wireless, wearable, and modular multi-channel sEMG acquisition system for the study of the neuromuscular system during highly-dynamic tasks.

Estimating the contributions of individual muscles during limb movements is crucial to understand motor system organization. In pathological conditions, identifying the roles of each individual muscles may provide a basis for devising personalized treatments. In a previous study we demonstrated how arm and muscle geometry can be estimated from isometric force data and used to reliably estimate isometric endpoint forces in various arm configurations. Here we use a Hill-type muscle model to predict muscle torques and equivalent endpoint forces during planar arm movements in real-time. In conjunction with a planar robot manipulandum, the model is then used to modify the directions of action of individual muscles or muscle groups.

A recent research line investigates the motor control adaptation after the perturbation of the muscle pulling directions, simulated with myoelectric control of an isometric reaching task in virtual reality. Such perturbations rely on an estimation of the mapping between the electromyographic (EMG) signal of recorded muscles and the end-point forces. However, the consistency of this mapping across sessions performed in different days is not known. In this study we tested the consistency of the EMG-to-force mapping recorded on multiple sessions in different days. We also tested the consistency of the EMG-to-force mapping after subject posture changes and accidental EMG electrodes detachment. A good consistency was identified when all the EMG recordings are normalized to a unique value, even if incorrect electrode positioning remains one of the major issues for reliable use of myoelectric control.