A computational model of use-dependent motor recovery following a stroke: optimizing corticospinal activations via reinforcement learning can explain residual capacity and other strength recovery dynamics

Insights into the dependence of post-stroke motor recovery on the initial corticospinal tract connectivity from a computational model

There is a consensus that motor recovery post-stroke primarily depends on the degree of the initial connectivity of the ipsilesional corticospinal tract (CST). Indeed, if the residual CST connectivity is sufficient to convey motor commands, the neuromotor system continues to use the CST predominantly, and motor function recovers up to 80%. In contrast, if the residual CST connectivity is insufficient, hand/arm dexterity barely recovers, even as the phases of stroke progress. Instead, the functional upregulation of the reticulospinal tract (RST) often occurs. In this study, we construct a computational model that reproduces the dependence of post-stroke motor recovery on the initial CST connectivity. The model emulates biologically plausible evolutions of primary motor descending tracts, based on activity-dependent or use-dependent plasticity and the preferential use of more strongly connected neural circuits. The model replicates several elements of the empirical evidence presented by the Fugl-Meyer Assessment (FMA) subscores, which evaluate the capabilities for out-of-synergy and in-synergy movements. These capabilities presumably change differently depending on the degree of the initial CST connectivity post-stroke, providing insights into the interactive dynamics of the primary descending motor tracts. We discuss findings derived from the proposed model in relation to the well-known proportional recovery rule. This modeling study aims to present a way to differentiate individuals who can achieve 70 to 80% recovery in the chronic phase from those who cannot by examining the interactive evolution of out-of-synergy and in-synergy movement capabilities during the subacute phase, as assessed by the FMA.

Using a Webcam to Assess Upper Extremity Proprioception: Experimental Validation and Application to Persons Post Stroke

Many medical conditions impair proprioception but there are few easy-to-deploy technologies for assessing proprioceptive deficits. Here, we developed a method-called "OpenPoint"-to quantify upper extremity (UE) proprioception using only a webcam as the sensor. OpenPoint automates a classic neurological test: the ability of a person to use one hand to point to a finger on their other hand with vision obscured. Proprioception ability is quantified with pointing error in the frontal plane measured by a deep-learning-based, computer vision library (MediaPipe). In a first experiment with 40 unimpaired adults, pointing error significantly increased when we replaced the target hand with a fake hand, verifying that this task depends on the availability of proprioceptive information from the target hand, and that we can reliably detect this dependence with computer vision. In a second experiment, we quantified UE proprioceptive ability in 16 post-stroke participants. Individuals post stroke exhibited increased pointing error (p < 0.001) that was correlated with finger proprioceptive error measured with an independent, robotic assessment (r = 0.62, p = 0.02). These results validate a novel method to assess UE proprioception ability using affordable computer technology, which provides a potential means to democratize quantitative proprioception testing in clinical and telemedicine environments.

An ANN models cortical-subcortical interaction during post-stroke recovery of finger dexterity

Objective.Finger dexterity, and finger individuation in particular, is crucial for human movement, and disruptions due to brain injury can significantly impact quality of life. Understanding the neurological mechanisms responsible for recovery is vital for effective neurorehabilitation. This study explores the role of two key pathways in finger individuation: the corticospinal (CS) tract from the primary motor cortex and premotor areas, and the subcortical reticulospinal (RS) tract from the brainstem. We aimed to investigate how the cortical-reticular network reorganizes to aid recovery of finger dexterity following lesions in these areas.Approach.To provide a potential biologically plausible answer to this question, we developed an artificial neural network (ANN) to model the interaction between a premotor planning layer, a cortical layer with excitatory and inhibitory CS outputs, and RS outputs controlling finger movements. The ANN was trained to simulate normal finger individuation and strength. A simulated stroke was then applied to the CS area, RS area, or both, and the recovery of finger dexterity was analyzed.Main results.In the intact model, the ANN demonstrated a near-linear relationship between the forces of instructed and uninstructed fingers, resembling human individuation patterns. Post-stroke simulations revealed that lesions in both CS and RS regions led to increased unintended force in uninstructed fingers, immediate weakening of instructed fingers, improved control during early recovery, and increased neural plasticity. Lesions in the CS region alone significantly impaired individuation, while RS lesions affected strength and to a lesser extent, individuation. The model also predicted the impact of stroke severity on finger individuation, highlighting the combined effects of CS and RS lesions.Significance.This model provides insights into the interactive role of cortical and subcortical regions in finger individuation. It suggests that recovery mechanisms involve reorganization of these networks, which may inform neurorehabilitation strategies.

Self-organizing recruitment of compensatory areas maximizes residual motor performance post-stroke

Whereas the orderly recruitment of compensatory motor cortical areas after stroke depends on the size of the motor cortex lesion affecting arm and hand movements, the mechanisms underlying this reorganization are unknown. Here, we hypothesized that the recruitment of compensatory areas results from the motor system's goal to optimize performance given the anatomical constraints before and after the lesion. This optimization is achieved through two complementary plastic processes: a homeostatic regulation process, which maximizes information transfer in sensory-motor networks, and a reinforcement learning process, which minimizes movement error and effort. To test this hypothesis, we developed a neuro-musculoskeletal model that controls a 7-muscle planar arm via a cortical network that includes a primary motor cortex and a premotor cortex that directly project to spinal motor neurons, and a contra-lesional primary motor cortex that projects to spinal motor neurons via the reticular formation. Synapses in the cortical areas are updated via reinforcement learning and the activity of spinal motor neurons is adjusted through homeostatic regulation. The model replicated neural, muscular, and behavioral outcomes in both non-lesioned and lesioned brains. With increasing lesion sizes, the model demonstrated systematic recruitment of the remaining primary motor cortex, premotor cortex, and contra-lesional cortex. The premotor cortex acted as a reserve area for fine motor control recovery, while the contra-lesional cortex helped avoid paralysis at the cost of poor joint control. Plasticity in spinal motor neurons enabled force generation after large cortical lesions despite weak corticospinal inputs. Compensatory activity in the premotor and contra-lesional motor cortex was more prominent in the early recovery period, gradually decreasing as the network minimized effort. Thus, the orderly recruitment of compensatory areas following strokes of varying sizes results from biologically plausible local plastic processes that maximize performance, whether the brain is intact or lesioned.

NSF DARE-Transforming modeling in neurorehabilitation: Four threads for catalyzing progress

We present an overview of the Conference on Transformative Opportunities for Modeling in Neurorehabilitation held in March 2023. It was supported by the Disability and Rehabilitation Engineering (DARE) program from the National Science Foundation's Engineering Biology and Health Cluster. The conference brought together experts and trainees from around the world to discuss critical questions, challenges, and opportunities at the intersection of computational modeling and neurorehabilitation to understand, optimize, and improve clinical translation of neurorehabilitation. We organized the conference around four key, relevant, and promising Focus Areas for modeling: Adaptation & Plasticity, Personalization, Human-Device Interactions, and Modeling 'In-the-Wild'. We identified four common threads across the Focus Areas that, if addressed, can catalyze progress in the short, medium, and long terms. These were: (i) the need to capture and curate appropriate and useful data necessary to develop, validate, and deploy useful computational models (ii) the need to create multi-scale models that span the personalization spectrum from individuals to populations, and from cellular to behavioral levels (iii) the need for algorithms that extract as much information from available data, while requiring as little data as possible from each client (iv) the insistence on leveraging readily available sensors and data systems to push model-driven treatments from the lab, and into the clinic, home, workplace, and community. The conference archive can be found at (dare2023.usc.edu). These topics are also extended by three perspective papers prepared by trainees and junior faculty, clinician researchers, and federal funding agency representatives who attended the conference.

Punishment Leads to Greater Sensorimotor Learning But Less Movement Variability Compared to Reward

When a musician practices a new song, hitting a correct note sounds pleasant while striking an incorrect note sounds unpleasant. Such reward and punishment feedback has been shown to differentially influence the ability to learn a new motor skill. Recent work has suggested that punishment leads to greater movement variability, which causes greater exploration and faster learning. To further test this idea, we collected 102 participants over two experiments. Unlike previous work, in Experiment 1 we found that punishment did not lead to faster learning compared to reward (n = 68), but did lead to a greater extent of learning. Surprisingly, we also found evidence to suggest that punishment led to less movement variability, which was related to the extent of learning. We then designed a second experiment that did not involve adaptation, allowing us to further isolate the influence of punishment feedback on movement variability. In Experiment 2, we again found that punishment led to significantly less movement variability compared to reward (n = 34). Collectively our results suggest that punishment feedback leads to less movement variability. Future work should investigate whether punishment feedback leads to a greater knowledge of movement variability and or increases the sensitivity of updating motor actions.

NSF DARE-transforming modeling in neurorehabilitation: perspectives and opportunities from US funding agencies

In recognition of the importance and timeliness of computational models for accelerating progress in neurorehabilitation, the U.S. National Science Foundation (NSF) and the National Institutes of Health (NIH) sponsored a conference in March 2023 at the University of Southern California that drew global participation from engineers, scientists, clinicians, and trainees. This commentary highlights promising applications of computational models to understand neurorehabilitation ("Using computational models to understand complex mechanisms in neurorehabilitation" section), improve rehabilitation care in the context of digital twin frameworks ("Using computational models to improve delivery and implementation of rehabilitation care" section), and empower future interdisciplinary workforces to deliver higher-quality clinical care using computational models ("Using computational models in neurorehabilitation requires an interdisciplinary workforce" section). The authors describe near-term gaps and opportunities, all of which encourage interdisciplinary team science. Four major opportunities were identified including (1) deciphering the relationship between engineering figures of merit-a term commonly used by engineers to objectively quantify the performance of a device, system, method, or material relative to existing state of the art-and clinical outcome measures, (2) validating computational models from engineering and patient perspectives, (3) creating and curating datasets that are made publicly accessible, and (4) developing new transdisciplinary frameworks, theories, and models that incorporate the complexities of the nervous and musculoskeletal systems. This commentary summarizes U.S. funding opportunities by two Federal agencies that support computational research in neurorehabilitation. The NSF has funding programs that support high-risk/high-reward research proposals on computational methods in neurorehabilitation informed by theory- and data-driven approaches. The NIH supports the development of new interventions and therapies for a wide range of nervous system injuries and impairments informed by the field of computational modeling. The conference materials can be found at https://dare2023.usc.edu/ .

MCP extensors respond faster than flexors in individuals with severe-to-moderate stroke-caused impairment: Evidence of uncoupled neural pathways

Damage in the corticospinal system following stroke produces imbalance between flexors and extensors in the upper extremity, eventually leading to flexion-favored postures. The substitution of alternative tracts for the damaged corticospinal tract is known to excessively activate flexors of the fingers while the fingers are voluntarily being extended. Here, we questioned whether the cortical source or/and neural pathways of the flexors and extensors of the fingers are coupled and what factor of impairment influences finger movement. In this study, a total of seven male participants with severe-to-moderate impairment by a hemiplegic stroke conducted flexion and extension at the metacarpophalangeal (MCP) joints in response to auditory tones. We measured activation and de-activation delays of the flexor and extensor of the MCP joints on the paretic side, and force generation. All participants generated greater torque in the direction of flexion (p = 0.017). Regarding co-contraction, coupled activation of the extensor is also made during flexion in the similar way to coupled activation of the flexor made during extension. As opposite to our expectation, we observed that during extension, the extensor showed marginally significantly faster activation (p = 0.66) while it showed faster de-activation (p = 0.038), in comparison to activation and de-activation of the flexor during flexion. But movement smoothness was not affected by those factors. Our results imply that the cortical source and neural pathway for the extensors of the MCP joints are not coupled with those for the flexors of the MCP joints.

Targeting neuroplasticity to improve motor recovery after stroke: an artificial neural network model

After a neurological injury, people develop abnormal patterns of neural activity that limit motor recovery. Traditional rehabilitation, which concentrates on practicing impaired skills, is seldom fully effective. New targeted neuroplasticity protocols interact with the central nervous system to induce beneficial plasticity in key sites and thereby enable wider beneficial plasticity. They can complement traditional therapy and enhance recovery. However, their development and validation is difficult because many different targeted neuroplasticity protocols are conceivable, and evaluating even one of them is lengthy, laborious, and expensive. Computational models can address this problem by triaging numerous candidate protocols rapidly and effectively. Animal and human empirical testing can then concentrate on the most promising ones. Here, we simulate a neural network of corticospinal neurons that control motoneurons eliciting unilateral finger extension. We use this network to (i) study the mechanisms and patterns of cortical reorganization after a stroke; and (ii) identify and parameterize a targeted neuroplasticity protocol that improves recovery of extension torque. After a simulated stroke, standard training produced abnormal bilateral cortical activation and suboptimal torque recovery. To enhance recovery, we interdigitated standard training with trials in which the network was given feedback only from a targeted population of sub-optimized neurons. Targeting neurons in secondary motor areas on ∼20% of the total trials restored lateralized cortical activation and improved recovery of extension torque. The results illuminate mechanisms underlying suboptimal cortical activity post-stroke; they enable the identification and parameterization of the most promising targeted neuroplasticity protocols. By providing initial guidance, computational models could facilitate and accelerate the realization of new therapies that improve motor recovery.

Adaptive Cognitive Training with Reinforcement Learning

Computer-assisted cognitive training can help patients affected by several illnesses alleviate their cognitive deficits or healthy people improve their mental performance. In most computer-based systems, training sessions consist of graded exercises, which should ideally be able to gradually improve the trainee’s cognitive functions. Indeed, adapting the difficulty of the exercises to how individuals perform in their execution is crucial to improve the effectiveness of cognitive training activities. In this article, we propose the use of reinforcement learning (RL) to learn how to automatically adapt the difficulty of computerized exercises for cognitive training. In our approach, trainees’ performance in performed exercises is used as a reward to learn a policy that changes over time the values of the parameters that determine exercise difficulty. We illustrate a method to be initially used to learn difficulty-variation policies tailored for specific categories of trainees, and then to refine these policies for single individuals. We present the results of two user studies that provide evidence for the effectiveness of our method: a first study, in which a student category policy obtained via RL was found to have better effects on the cognitive function than a standard baseline training that adopts a mechanism to vary the difficulty proposed by neuropsychologists, and a second study, demonstrating that adding an RL-based individual customization further improves the training process.

Characteristics of the severely impaired hand in survivors of stroke with chronic impairments

Background: Diminished sensorimotor control of the hand is one of the most common outcomes following stroke. This hand impairment substantially impacts overall function and quality of life; standard therapy often results in limited improvement. Mechanisms of dysfunction of the severely impaired post-stroke hand are still incompletely understood, thereby impeding the development of new targeted treatments.Objective: To identify and determine potential relationships among the mechanisms responsible for hand impairment following strokeMethods: This cohort study observed stroke survivors (n = 95) with severe, chronic hand impairment (Chedoke-McMaster Hand score = 2-3). Custom instrumentation created precise perturbations and measured kinematic responses. Muscle activation was recorded through electromyography. Strength, spasticity, muscle relaxation time, and muscle coactivation were quantified.Results: Maximum grip strength in the paretic hand was only 12% of that achieved by the nonparetic hand, and only 6 of 95 participants were able to produce any net extension force. Despite force deficits, spastic reflex response of the finger flexor evoked by imposed stretch averaged 90.1 ± 26.8% of maximum voluntary activation, relaxation time averaged 3.8 ± 0.8 seconds, and coactivation during voluntary extension exceeded 30% of maximum contraction, thereby resulting in substantial net flexion. Surprisingly, these hypertonicity measures were not significantly correlated with each other.Conclusions: Survivors of severe, chronic hemiparetic stroke experience profound weakness of both flexion and extension that arises from increased involuntary antagonist activation and decreased voluntary activation. The lack of correlation amongst hypertonicity measures suggests that these phenomena may arise from multiple, potentially independent mechanisms that could require different treatments.

Reinforcement learning for intelligent healthcare applications: A survey

Discovering new treatments and personalizing existing ones is one of the major goals of modern clinical research. In the last decade, Artificial Intelligence (AI) has enabled the realization of advanced intelligent systems able to learn about clinical treatments and discover new medical knowledge from the huge amount of data collected. Reinforcement Learning (RL), which is a branch of Machine Learning (ML), has received significant attention in the medical community since it has the potentiality to support the development of personalized treatments in accordance with the more general precision medicine vision. This report presents a review of the role of RL in healthcare by investigating past work, and highlighting any limitations and possible future contributions.

A Review of Sensory Feedback in Upper-Limb Prostheses From the Perspective of Human Motor Control

This manuscript reviews historical and recent studies that focus on supplementary sensory feedback for use in upper limb prostheses. It shows that the inability of many studies to speak to the issue of meaningful performance improvements in real-life scenarios is caused by the complexity of the interactions of supplementary sensory feedback with other types of feedback along with other portions of the motor control process. To do this, the present manuscript frames the question of supplementary feedback from the perspective of computational motor control, providing a brief review of the main advances in that field over the last 20 years. It then separates the studies on the closed-loop prosthesis control into distinct categories, which are defined by relating the impact of feedback to the relevant components of the motor control framework, and reviews the work that has been done over the last 50+ years in each of those categories. It ends with a discussion of the studies, along with suggestions for experimental construction and connections with other areas of research, such as machine learning.

Key components of mechanical work predict outcomes in robotic stroke therapy

Background

Clinical practice typically emphasizes active involvement during therapy. However, traditional approaches can offer only general guidance on the form of involvement that would be most helpful to recovery. Beyond assisting movement, robots allow comprehensive methods for measuring practice behaviors, including the energetic input of the learner. Using data from our previous study of robot-assisted therapy, we examined how separate components of mechanical work contribute to predicting training outcomes.

Methods

Stroke survivors (n = 11) completed six sessions in two-weeks of upper extremity motor exploration (self-directed movement practice) training with customized forces, while a control group (n = 11) trained without assistance. We employed multiple regression analysis to predict patient outcomes with computed mechanical work as independent variables, including separate features for elbow versus shoulder joints, positive (concentric) and negative (eccentric), flexion and extension.

Results

Our analysis showed that increases in total mechanical work during therapy were positively correlated with our final outcome metric, velocity range. Further analysis revealed that greater amounts of negative work at the shoulder and positive work at the elbow as the most important predictors of recovery (using cross-validated regression, R² = 52%). However, the work features were likely mutually correlated, suggesting a prediction model that first removed shared variance (using PCA, R² = 65–85%).

Conclusions

These results support robotic training for stroke survivors that increases energetic activity in eccentric shoulder and concentric elbow actions.

Trial registration

ClinicalTrials.gov, Identifier: NCT02570256. Registered 7 October 2015 – Retrospectively registered,

Breaking Proportional Recovery After Stroke

People with hemiparesis after stroke appear to recover 70% to 80% of the difference between their baseline and the maximum upper extremity Fugl-Meyer (UEFM) score, a phenomenon called proportional recovery (PR). Two recent commentaries explained that PR should be expected because of mathematical coupling between the baseline and change score. Here we ask, If mathematical coupling encourages PR, why do a fraction of stroke patients (the "nonfitters") not exhibit PR? At the neuroanatomical level of analysis, this question was answered by Byblow et al-nonfitters lack corticospinal tract (CST) integrity at baseline-but here we address the mathematical and behavioral causes. We first derive a new interpretation of the slope of PR: It is the average probability of scoring across remaining scale items at follow-up. PR therefore breaks when enough test items are discretely more difficult for a patient at follow-up, flattening the slope of recovery. For the UEFM, we show that nonfitters are most unlikely to recover the ability to score on the test items related to wrist/hand dexterity, shoulder flexion without bending the elbow, and finger-to-nose movement, supporting the finding that nonfitters lack CST integrity. However, we also show that a subset of nonfitters respond better to robotic movement training in the chronic phase of stroke. These persons are just able to move the arm out of the flexion synergy and pick up small blocks, both markers of CST integrity. Nonfitters therefore raise interesting questions about CST function and the basis for response to intensive movement training.

Perspectives and Challenges in Robotic Neurorehabilitation

The development of robotic devices for rehabilitation is a fast-growing field. Nowadays, thanks to novel technologies that have improved robots’ capabilities and offered more cost-effective solutions, robotic devices are increasingly being employed during clinical practice, with the goal of boosting patients’ recovery. Robotic rehabilitation is also widely used in the context of neurological disorders, where it is often provided in a variety of different fashions, depending on the specific function to be restored. Indeed, the effect of robot-aided neurorehabilitation can be maximized when used in combination with a proper training regimen (based on motor control paradigms) or with non-invasive brain machine interfaces. Therapy-induced changes in neural activity and behavioral performance, which may suggest underlying changes in neural plasticity, can be quantified by multimodal assessments of both sensorimotor performance and brain/muscular activity pre/post or during intervention. Here, we provide an overview of the most common robotic devices for upper and lower limb rehabilitation and we describe the aforementioned neurorehabilitation scenarios. We also review assessment techniques for the evaluation of robotic therapy. Additional exploitation of these research areas will highlight the crucial contribution of rehabilitation robotics for promoting recovery and answering questions about reorganization of brain functions in response to disease.

The effect of a novel gait retraining device on lower limb kinematics and muscle activation in healthy adults

The Re-Link Trainer (RLT) is a modified walking frame with a linkage system designed to apply a non-individualized kinematic constraint to normalize gait trajectory of the left limb. The premise behind the RLT is that a user's lower limb is constrained into a physiologically normal gait pattern, ideally generating symmetry across gait cycle parameters and kinematics. This pilot study investigated adaptations in the natural gait pattern of healthy adults when using the RLT compared to normal overground walking. Bilateral lower limb kinematic and electromyography data were collected while participants walked overground at a self-selected speed, followed by walking in the RLT. A series of 2-way analyses of variance examined between-limb and between-condition differences. Peak hip extension and knee flexion were reduced bilaterally when walking in the RLT. Left peak hip extension occurred earlier in the gait cycle when using the RLT, but later for the right limb. Peak hip flexion was significantly increased and occurred earlier for the constrained limb, while peak plantarflexion was significantly reduced. Peak knee flexion and plantarflexion in the right limb occurred later when using the RLT. Significant bilateral reductions in peak electromyography amplitude were evident when walking in the RLT, along with a significant shift in when peak muscle activity was occurring. These findings suggest that the RLT does impose a significant constraint, but generates asymmetries in lower limb kinematics and muscle activity patterns. The large interindividual variation suggests users may utilize differing motor strategies to adapt their gait pattern to the imposed constraint.

The negotiated equilibrium model of spinal cord function

The belief that the spinal cord is hardwired is no longer tenable. Like the rest of the CNS, the spinal cord changes during growth and ageing, when new motor behaviours are acquired, and in response to trauma and disease. This paper describes a new model of spinal cord function that reconciles its recently appreciated plasticity with its long-recognized reliability as the final common pathway for behaviour. According to this model, the substrate of each motor behaviour comprises brain and spinal plasticity: the plasticity in the brain induces and maintains the plasticity in the spinal cord. Each time a behaviour occurs, the spinal cord provides the brain with performance information that guides changes in the substrate of the behaviour. All the behaviours in the repertoire undergo this process concurrently; each repeatedly induces plasticity to preserve its key features despite the plasticity induced by other behaviours. The aggregate process is a negotiation among the behaviours: they negotiate the properties of the spinal neurons and synapses that they all use. The ongoing negotiation maintains the spinal cord in an equilibrium - a negotiated equilibrium - that serves all the behaviours. This new model of spinal cord function is supported by laboratory and clinical data, makes predictions borne out by experiment, and underlies a new approach to restoring function to people with neuromuscular disorders. Further studies are needed to test its generality, to determine whether it may apply to other CNS areas such as the cerebral cortex, and to develop its therapeutic implications.

How do strength and coordination recovery interact after stroke? A computational model for informing robotic training

Robotic devices can train strength, coordination, or a combination of both. If a robotic device focuses on coordination, what happens to strength recovery, and vice versa? Understanding this interaction could help optimize robotic training. We developed a computational neurorehabilitation model to gain insight into the interaction between strength and coordination recovery after stroke. In the model, the motor system recovers by optimizing the activity of residual corticospinal cells (focally connected, excitatory and inhibitory) and reticulospinal cells (diffusely connected and excitatory) to achieve a motor task. To do this, the model employs a reinforcement learning algorithm that uses stochastic search based on a reward signal produced by task execution. We simulated two tasks that require strength and coordination: a finger movement task and a bilateral wheelchair propulsion task. We varied the reward signal to value strength versus coordination, determined by a weighting factor. The model predicted a nonlinear relationship between strength and coordination recovery consistent with clinical data obtained for each task. The model also predicted that stroke can cause a competition between strength and coordination recovery, due to a scarcity of focal and inhibitory cells. These results provide a rationale for implementing robotic movement therapy that can adaptively alter the combination of force and coordination training to target desired components of motor recovery.

Simulation of variable impedance as an intervention for upper extremity motor exploration

Current methods in robot-assisted therapy are limited in providing predictions of the effectiveness of interventions. Our approach focuses on how robotic interaction can impact the distribution of movements expressed in the arm. Using data from a previous study with stroke survivors (n=10), we performed simulations to examine how changes in hand endpoint impedance would alter exploratory motion. We present methods for designing a custom training intervention, by relating the desired change in acceleration covariance in planar motion with a corresponding change in inertia matrix. We first characterized motor exploration in terms of overall covariance in acceleration, and secondly as covariance that varies with position in the workspace. Using a forward dynamics simulation of the hand endpoint impedance, we found that the variable change in endpoint inertia resulted in better recovery of acceleration covariance compared to the fixed change in inertia method. These results could significantly impact rehabilitation firstly in terms of design principles for altering coordination patterns through direct assistance. Furthermore, our work might serve to improve therapy by facilitating access to repeated practice of independent joint motion.

Robot Training With Vector Fields Based on Stroke Survivors' Individual Movement Statistics

The wide variation in upper extremity motor impairments among stroke survivors necessitates more intelligent methods of customized therapy. However, current strategies for characterizing individual motor impairments are limited by the use of traditional clinical assessments (e.g., Fugl-Meyer) and simple engineering metrics (e.g., goal-directed performance). Our overall approach is to statistically identify the range of volitional movement capabilities, and then apply a robot-applied force vector field intervention that encourages under-expressed movements. We investigated whether explorative training with such customized force fields would improve stroke survivors' (n = 11) movement patterns in comparison to a control group that trained without forces (n = 11). Force and control groups increased Fugl-Meyer UE scores (average of 1.0 and 1.1, respectively), which is not considered clinically meaningful. Interestingly, participants from both groups demonstrated dramatic increases in their range of velocity during exploration following only six days of training (average increase of 166.4% and 153.7% for the Force and Control group, respectively). While both groups showed evidence of improvement, we also found evidence that customized forces affected learning in a systematic way. When customized forces were active, we observed broader distributions of velocity that were not present in the controls. Second, we found that these changes led to specific changes in unassisted motion. In addition, while the shape of movement distributions changed significantly for both groups, detailed analysis of the velocity distributions revealed that customized forces promoted a greater proportion of favorable changes. Taken together, these results provide encouraging evidence that patient-specific force fields based on individuals' movement statistics can be used to create new movement patterns and shape them in a customized manner. To the best of our knowledge, this paper is the first to directly link engineering assessments of stroke survivors' exploration movement behaviors to the design of customized robot therapy.

Computational neurorehabilitation: modeling plasticity and learning to predict recovery

Despite progress in using computational approaches to inform medicine and neuroscience in the last 30 years, there have been few attempts to model the mechanisms underlying sensorimotor rehabilitation. We argue that a fundamental understanding of neurologic recovery, and as a result accurate predictions at the individual level, will be facilitated by developing computational models of the salient neural processes, including plasticity and learning systems of the brain, and integrating them into a context specific to rehabilitation. Here, we therefore discuss Computational Neurorehabilitation, a newly emerging field aimed at modeling plasticity and motor learning to understand and improve movement recovery of individuals with neurologic impairment. We first explain how the emergence of robotics and wearable sensors for rehabilitation is providing data that make development and testing of such models increasingly feasible. We then review key aspects of plasticity and motor learning that such models will incorporate. We proceed by discussing how computational neurorehabilitation models relate to the current benchmark in rehabilitation modeling - regression-based, prognostic modeling. We then critically discuss the first computational neurorehabilitation models, which have primarily focused on modeling rehabilitation of the upper extremity after stroke, and show how even simple models have produced novel ideas for future investigation. Finally, we conclude with key directions for future research, anticipating that soon we will see the emergence of mechanistic models of motor recovery that are informed by clinical imaging results and driven by the actual movement content of rehabilitation therapy as well as wearable sensor-based records of daily activity.

Robotic Rehabilitator of the Rodent Upper Extremity: A System and Method for Assessing and Training Forelimb Force Production after Neurological Injury

Rodent models of spinal cord injury are critical for the development of treatments for upper limb motor impairment in humans, but there are few methods for measuring forelimb strength of rodents, an important outcome measure. We developed a novel robotic device--the Robotic Rehabilitator of the Rodent Upper Extremity (RUE)--that requires rats to voluntarily reach for and pull a bar to retrieve a food reward; the resistance of the bar can be programmed. We used RUE to train forelimb strength of 16 rats three times per week for 23 weeks before and 38 weeks after a mild (100 kdyne) unilateral contusion at the cervical level 5 (C5). We measured maximum force produced when RUE movement was unexpectedly blocked. We compared this blocked pulling force (BPF) to weekly measures of forelimb strength obtained with a previous, well-established method: the grip strength meter (GSM). Before injury, BPF was 2.6 times higher (BPF, 444.6 ± 19.1 g; GSM, 168.4 ± 3.1 g) and 4.9 times more variable (p < 0.001) than pulling force measured with the GSM; the two measurement methods were uncorrelated (R(2) = 0.03; p = 0.84). After injury, there was a significant decrease in BPF of 134.35 g ± 14.71 g (p < 0.001). Together, our findings document BPF as a repeatable measure of forelimb force production, sensitive to a mild spinal cord injury, which comes closer to measuring maximum force than the GSM and thus may provide a useful measure for quantifying the effects of treatment in rodent models of SCI.

The Virtual Brain: Modeling Biological Correlates of Recovery after Chronic Stroke

There currently remains considerable variability in stroke survivor recovery. To address this, developing individualized treatment has become an important goal in stroke treatment. As a first step, it is necessary to determine brain dynamics associated with stroke and recovery. While recent methods have made strides in this direction, we still lack physiological biomarkers. The Virtual Brain (TVB) is a novel application for modeling brain dynamics that simulates an individual’s brain activity by integrating their own neuroimaging data with local biophysical models. Here, we give a detailed description of the TVB modeling process and explore model parameters associated with stroke. In order to establish a parallel between this new type of modeling and those currently in use, in this work we establish an association between a specific TVB parameter (long-range coupling) that increases after stroke with metrics derived from graph analysis. We used TVB to simulate the individual BOLD signals for 20 patients with stroke and 10 healthy controls. We performed graph analysis on their structural connectivity matrices calculating degree centrality, betweenness centrality, and global efficiency. Linear regression analysis demonstrated that long-range coupling is negatively correlated with global efficiency (P = 0.038), but is not correlated with degree centrality or betweenness centrality. Our results suggest that the larger influence of local dynamics seen through the long-range coupling parameter is closely associated with a decreased efficiency of the system. We thus propose that the increase in the long-range parameter in TVB (indicating a bias toward local over global dynamics) is deleterious because it reduces communication as suggested by the decrease in efficiency. The new model platform TVB hence provides a novel perspective to understanding biophysical parameters responsible for global brain dynamics after stroke, allowing the design of focused therapeutic interventions.

Robot-assisted motor training: assistance decreases exploration during reinforcement learning

Reinforcement learning (RL) is a form of motor learning that robotic therapy devices could potentially manipulate to promote neurorehabilitation. We developed a system that requires trainees to use RL to learn a predefined target movement. The system provides higher rewards for movements that are more similar to the target movement. We also developed a novel algorithm that rewards trainees of different abilities with comparable reward sizes. This algorithm measures a trainee's performance relative to their best performance, rather than relative to an absolute target performance, to determine reward. We hypothesized this algorithm would permit subjects who cannot normally achieve high reward levels to do so while still learning. In an experiment with 21 unimpaired human subjects, we found that all subjects quickly learned to make a first target movement with and without the reward equalization. However, artificially increasing reward decreased the subjects' tendency to engage in exploration and therefore slowed learning, particularly when we changed the target movement. An anti-slacking watchdog algorithm further slowed learning. These results suggest that robotic algorithms that assist trainees in achieving rewards or in preventing slacking might, over time, discourage the exploration needed for reinforcement learning.

Machine-Based, Self-guided Home Therapy for Individuals With Severe Arm Impairment After Stroke: A Randomized Controlled Trial

BACKGROUND

Few therapeutic options exist for the millions of persons living with severe arm impairment after stroke to increase their dose of arm rehabilitation. This study compared self-guided, high-repetition home therapy with a mechanical device (the resonating arm exerciser [RAE]) to conventional therapy in patients with chronic stroke and explored RAE use for patients with subacute stroke.

METHODS

A total of 16 participants with severe upper-extremity impairment (mean Fugl-Meyer [FM] score = 21.4 ± 8.8 out of 66) >6 months poststroke were randomized to 3 weeks of exercise with the RAE or conventional exercises. The primary outcome measure was FM score 1 month posttherapy. Secondary outcome measures included Motor Activity Log, Visual Analog Pain Scale, and Ashworth Spasticity Scale. After a 1-month break, individuals in the conventional group also received a 3-week course of RAE therapy.

RESULTS

The change in FM score was significant in both the RAE and conventional groups after training (2.6 ± 1.4 and 3.4 ± 2.4, P = .008 and .016, respectively). These improvements were not significant at 1 month. Exercise with the RAE led to significantly greater improvements in distal FM score than conventional therapy at the 1-month follow-up (P = .02). In a separate cohort of patients with subacute stroke, the RAE was found feasible for exercise.

DISCUSSION

In those with severe arm impairment after chronic stroke, home-based training with the RAE was feasible and significantly reduced impairment without increasing pain or spasticity. Gains with the RAE were comparable to those found with conventional training and also included distal arm improvement.

Time-sensitive reorganization of the somatosensory cortex poststroke depends on interaction between Hebbian and homeoplasticity: a simulation study

Together with Hebbian plasticity, homeoplasticity presumably plays a significant, yet unclear, role in recovery postlesion. Here, we undertake a simulation study addressing the role of homeoplasticity and rehabilitation timing poststroke. We first hypothesize that homeoplasticity is essential for recovery and second that rehabilitation training delivered too early, before homeoplasticity has compensated for activity disturbances postlesion, is less effective for recovery than training delivered after a delay. We developed a neural network model of the sensory cortex driven by muscle spindle inputs arising from a six-muscle arm. All synapses underwent Hebbian plasticity, while homeoplasticity adjusted cell excitability to maintain a desired firing distribution. After initial training, the network was lesioned, leading to areas of hyper- and hypoactivity due to the loss of lateral synaptic connections. The network was then retrained through rehabilitative arm movements. We found that network recovery was unsuccessful in the absence of homeoplasticity, as measured by reestablishment of lesion-affected inputs. We also found that a delay preceding rehabilitation led to faster network recovery during the rehabilitation training than no delay. Our simulation results thus suggest that homeoplastic restoration of prelesion activity patterns is essential to functional network recovery via Hebbian plasticity.

Perspectives for computational modeling of cell replacement for neurological disorders

Mathematical modeling of anatomically-constrained neural networks has provided significant insights regarding the response of networks to neurological disorders or injury. A logical extension of these models is to incorporate treatment regimens to investigate network responses to intervention. The addition of nascent neurons from stem cell precursors into damaged or diseased tissue has been used as a successful therapeutic tool in recent decades. Interestingly, models have been developed to examine the incorporation of new neurons into intact adult structures, particularly the dentate granule neurons of the hippocampus. These studies suggest that the unique properties of maturing neurons, can impact circuit behavior in unanticipated ways. In this perspective, we review the current status of models used to examine damaged CNS structures with particular focus on cortical damage due to stroke. Secondly, we suggest that computational modeling of cell replacement therapies can be made feasible by implementing approaches taken by current models of adult neurogenesis. The development of these models is critical for generating hypotheses regarding transplant therapies and improving outcomes by tailoring transplants to desired effects.

Neuromotor recovery from stroke: computational models at central, functional, and muscle synergy level

Computational models of neuromotor recovery after a stroke might help to unveil the underlying physiological mechanisms and might suggest how to make recovery faster and more effective. At least in principle, these models could serve: (i) To provide testable hypotheses on the nature of recovery; (ii) To predict the recovery of individual patients; (iii) To design patient-specific “optimal” therapy, by setting the treatment variables for maximizing the amount of recovery or for achieving a better generalization of the learned abilities across different tasks. Here we review the state of the art of computational models for neuromotor recovery through exercise, and their implications for treatment. We show that to properly account for the computational mechanisms of neuromotor recovery, multiple levels of description need to be taken into account. The review specifically covers models of recovery at central, functional and muscle synergy level.

Technologies and combination therapies for enhancing movement training for people with a disability

There has been a dramatic increase over the last decade in research on technologies for enhancing movement training and exercise for people with a disability. This paper reviews some of the recent developments in this area, using examples from a National Science Foundation initiated study of mobility research projects in Europe to illustrate important themes and key directions for future research. This paper also reviews several recent studies aimed at combining movement training with plasticity or regeneration therapies, again drawing in part from European research examples. Such combination therapies will likely involve complex interactions with motor training that must be understood in order to achieve the goal of eliminating severe motor impairment.