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.

Long-Term Stability of Motor Cortical Activity: Implications for Brain Machine Interfaces and Optimal Feedback Control.

The human motor system is capable of remarkably precise control of movements--consider the skill of professional baseball pitchers or surgeons. This precise control relies upon stable representations of movements in the brain. Here, we investigated the stability of cortical activity at multiple spatial and temporal scales by recording local field potentials (LFPs) and action potentials (multiunit spikes, MSPs) while two monkeys controlled a cursor either with their hand or directly from the brain using a brain-machine interface. LFPs and some MSPs were remarkably stable over time periods ranging from 3 d to over 3 years; overall, LFPs were significantly more stable than spikes. We then assessed whether the stability of all neural activity, or just a subset of activity, was necessary to achieve stable behavior. We showed that projections of neural activity into the subspace relevant to the task (the "task-relevant space") were significantly more stable than were projections into the task-irrelevant (or "task-null") space. This provides cortical evidence in support of the minimum intervention principle, which proposes that optimal feedback control (OFC) allows the brain to tightly control only activity in the task-relevant space while allowing activity in the task-irrelevant space to vary substantially from trial to trial. We found that the brain appears capable of maintaining stable movement representations for extremely long periods of time, particularly so for neural activity in the task-relevant space, which agrees with OFC predictions. SIGNIFICANCE STATEMENT: It is unknown whether cortical signals are stable for more than a few weeks. Here, we demonstrate that motor cortical signals can exhibit high stability over several years. This result is particularly important to brain-machine interfaces because it could enable stable performance with infrequent recalibration. Although we can maintain movement accuracy over time, movement components that are unrelated to the goals of a task (such as elbow position during reaching) often vary from trial to trial. This is consistent with the minimum intervention principle of optimal feedback control. We provide evidence that the motor cortex acts according to this principle: cortical activity is more stable in the task-relevant space and more variable in the task-irrelevant space.

Degraded expression of learned feedforward control in movements released by startle.

Recent work has shown that preplanned motor programs can be rapidly released via fast conducting pathways using a startling acoustic stimulus. Our question was whether the startle-elicited response might also release a recently learned internal model, which draws on experience to predict and compensate for expected perturbations in a feedforward manner. Our initial investigation using adaptation to robotically produced forces showed some evidence of this, but the results were potentially confounded by co-contraction caused by startle. In this study, we eliminated this confound by asking subjects to make reaching movements in the presence of a visual distortion. Results show that a startle stimulus (1) decreased performance of the recently learned task and (2) reduced after-effect magnitude. Since the recall of learned control was reduced, but not eliminated during startle trials, we suggest that multiple neural centers (cortical and subcortical) are involved in such learning and adaptation. These findings have implications for motor training in areas such as piloting, teleoperation, sports, and rehabilitation.

Extracting kinetic information from human motor cortical signals.

Brain machine interfaces (BMIs) have the potential to provide intuitive control of neuroprostheses to restore grasp to patients with paralyzed or amputated upper limbs. For these neuroprostheses to function, the ability to accurately control grasp force is critical. Grasp force can be decoded from neuronal spikes in monkeys, and hand kinematics can be decoded using electrocorticogram (ECoG) signals recorded from the surface of the human motor cortex. We hypothesized that kinetic information about grasping could also be extracted from ECoG, and sought to decode continuously-graded grasp force. In this study, we decoded isometric pinch force with high accuracy from ECoG in 10 human subjects. The predicted signals explained from 22% to 88% (60 ± 6%, mean ± SE) of the variance in the actual force generated. We also decoded muscle activity in the finger flexors, with similar accuracy to force decoding. We found that high gamma band and time domain features of the ECoG signal were most informative about kinetics, similar to our previous findings with intracortical LFPs. In addition, we found that peak cortical representations of force applied by the index and little fingers were separated by only about 4mm. Thus, ECoG can be used to decode not only kinematics, but also kinetics of movement. This is an important step toward restoring intuitively-controlled grasp to impaired patients.

Post-stroke motor deficits are most evident at frequencies near 125 Hz in EMG multivariate probability distributions.

Muscle activity is widely measured to assess muscle condition in post-stroke patients. While many clinical researchers have relied on time-series analysis of muscle activity, the frequency domain could offer additional insight on motor impairment. Our previous work has characterized movement capabilities in stroke survivors across endpoint and joint kinematic variables while performing a self-directed motor exploration task. Our solution to managing such large volumes of data is to create personalized statistical profiles using multivariate probability distributions. In this study, we present frequency domain analysis of EMG distributions for chronic post-stroke survivors (N = 6) and healthy subjects (N = 5) to identify between group differences in muscle activity. Comparing probability density of muscle activity magnitudes, differences from healthy were most evident at 275 Hz. Unique aspects of each patient's deficits were most evident at 125 Hz. This is the first study to explore distributions of EMG in specific frequency bands for this patient population. Such identifiability could pinpoint specific motor deficits and track progress in neurologically impaired individuals that often have widely differing disease states.

Energetics during robot-assisted training predicts recovery in stroke.

Clinical investigators have asserted patients should be active participants in the therapy process in stroke rehabilitation. While robotics introduces new tools for measurement and treatment of motor impairments, it also presents challenges for evaluating how much a patient contributes to observed movements during training. Our approach employs established methods of inverse dynamics combined with measurements of human motion and interaction forces between the human and robot. Here, we investigated whether measures of patient active involvement predict the level of upper limb recovery due to robot-assisted therapy. Stroke survivors (n=11) completed "exploration" training with customizable forces that increased their velocities (i.e., negative damping). While our results showed a mild trend between mechanical work during training and expanded velocity capability (Pearson r = 0.57), we found significant correlations with the amount of positive work (i.e., propulsion; r = 0.77), but not negative work (i.e., braking; r = 0.41). This work supports robotic tools that encourage more positive work.

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.

Reducing Abnormal Muscle Coactivation After Stroke Using a Myoelectric-Computer Interface: A Pilot Study.

Background A significant factor in impaired movement caused by stroke is the inability to activate muscles independently. Although the pathophysiology behind this abnormal coactivation is not clear, reducing the coactivation could improve overall arm function. A myoelectric computer interface (MCI), which maps electromyographic signals to cursor movement, could be used as a treatment to help retrain muscle activation patterns. Objective To investigate the use of MCI training to reduce abnormal muscle coactivation in chronic stroke survivors. Methods A total of 5 healthy participants and 5 stroke survivors with hemiparesis participated in multiple sessions of MCI training. The level of arm impairment in stroke survivors was assessed using the upper-extremity portion of the Fugl-Meyer Motor Assessment (FMA-UE). Participants performed isometric activations of up to 5 muscles. Activation of each muscle was mapped to different directions of cursor movement. The MCI specifically targeted 1 pair of muscles in each participant for reduction of coactivation. Results Both healthy participants and stroke survivors learned to reduce abnormal coactivation of the targeted muscles with MCI training. Out of 5 stroke survivors, 3 exhibited objective reduction in arm impairment as well (improvement in FMA-UE of 3 points in each of these patients). Conclusions These results suggest that the MCI was an effective tool in directly retraining muscle activation patterns following stroke.

Distributions in the error space: goal-directed movements described in time and state-space representations.

Manipulation of error feedback has been of great interest to recent studies in motor control and rehabilitation. Typically, motor adaptation is shown as a change in performance with a single scalar metric for each trial, yet such an approach might overlook details about how error evolves through the movement. We believe that statistical distributions of movement error through the extent of the trajectory can reveal unique patterns of adaption and possibly reveal clues to how the motor system processes information about error. This paper describes different possible ordinate domains, focusing on representations in time and state-space, used to quantify reaching errors. We hypothesized that the domain with the lowest amount of variability would lead to a predictive model of reaching error with the highest accuracy. Here we showed that errors represented in a time domain demonstrate the least variance and allow for the highest predictive model of reaching errors. These predictive models will give rise to more specialized methods of robotic feedback and improve previous techniques of error augmentation.

Data sample size needed for prediction of movement distributions.

Human movement ability should be described not only by its typical behavior, but also by the wide variation in capabilities. This would mean that subjects that are encouraged to move throughout their workspace but otherwise free to move any way they like might reveal their unique movement tendencies. In this study, we investigate how much information (data) is needed to reliably construct a movement distribution that predicts an individual's movement tendencies. We analyzed the distributions of position, velocity and acceleration data derived during self-directed motor exploration by stroke survivors (n=10 from a previous study) and healthy individuals (n=5). We examined whether these simple kinematic variables differed in terms of the amount of data required. We found a trend of decreasing time needed for characterization with the order of kinematic variable, for position, velocity, and acceleration, respectively. In addition, we investigated whether data requirements differ between stroke survivors and healthy. Our results suggest that healthy individuals may require more data samples (time for characterization), though the trend was only significant for position data. Our results provide an important step towards using statistical distributions to describe movement tendencies. Our findings could serve as more comprehensive tools to track recovery in or design more focused training intervention in neurorehabiliation applications.

Direct classification of all American English phonemes using signals from functional speech motor cortex.

OBJECTIVE: Although brain-computer interfaces (BCIs) can be used in several different ways to restore communication, communicative BCI has not approached the rate or efficiency of natural human speech. Electrocorticography (ECoG) has precise spatiotemporal resolution that enables recording of brain activity distributed over a wide area of cortex, such as during speech production. In this study, we sought to decode elements of speech production using ECoG. APPROACH: We investigated words that contain the entire set of phonemes in the general American accent using ECoG with four subjects. Using a linear classifier, we evaluated the degree to which individual phonemes within each word could be correctly identified from cortical signal. MAIN RESULTS: We classified phonemes with up to 36% accuracy when classifying all phonemes and up to 63% accuracy for a single phoneme. Further, misclassified phonemes follow articulation organization described in phonology literature, aiding classification of whole words. Precise temporal alignment to phoneme onset was crucial for classification success. SIGNIFICANCE: We identified specific spatiotemporal features that aid classification, which could guide future applications. Word identification was equivalent to information transfer rates as high as 3.0 bits s(-1) (33.6 words min(-1)), supporting pursuit of speech articulation for BCI control.

Long-term, stable behavior of local field potentials during brain machine interface use.

Local field potentials (LFPs) have the potential to provide robust, long-lasting control signals for brain-machine interfaces (BMIs). Moreover, they have been hypothesized to be a stable signal source. Here we assess the long-term stability of LFPs and multi-unit spikes (MSPs) in two monkeys using both LFP-based and MSP-based, biomimetic BMIs to control a computer cursor. The monkeys demonstrated highly accurate performance using both the LFP- and MSP-based BMIs. This performance remained high for 11 and 6 months, respectively, without adapting or retraining. We evaluated the stability of the LFP features and MSPs themselves by building, in each session, linear decoders of the BMI-controlled cursor velocity using single features or single MSPs. We then used these single-feature decoders to decode BMI-controlled cursor velocity in the last session. Many of the LFP features and MSPs showed stably-high correlations with the cursor velocity over the entire study period. This implies that the monkeys were able to maintain a stable mapping between either motor cortical field potentials or multi-spike potentials and BMI-controlled outputs.

Long term, stable brain machine interface performance using local field potentials and multiunit spikes.

OBJECTIVE: Brain machine interfaces (BMIs) have the potential to restore movement to people with paralysis. However, a clinically-viable BMI must enable consistently accurate control over time spans ranging from years to decades, which has not yet been demonstrated. Most BMIs that use single-unit spikes as inputs will experience degraded performance over time without frequent decoder re-training. Two other signals, local field potentials (LFPs) and multi-unit spikes (MSPs), may offer greater reliability over long periods and better performance stability than single-unit spikes. Here, we demonstrate that LFPs can be used in a biomimetic BMI to control a computer cursor. APPROACH: We implanted two rhesus macaques with intracortical microelectrodes in primary motor cortex. We recorded LFP and MSP signals from the monkeys while they performed a continuous reaching task, moving a cursor to randomly-placed targets on a computer screen. We then used the LFP and MSP signals to construct biomimetic decoders for control of the cursor. MAIN RESULTS: Both monkeys achieved high-performance, continuous control that remained stable or improved over nearly 12 months using an LFP decoder that was not retrained or adapted. In parallel, the monkeys used MSPs to control a BMI without retraining or adaptation and had similar or better performance, and that predominantly remained stable over more than six months. In contrast to their stable online control, both LFP and MSP signals showed substantial variability when used offline to predict hand movements. SIGNIFICANCE: Our results suggest that the monkeys were able to stabilize the relationship between neural activity and cursor movement during online BMI control, despite variability in the relationship between neural activity and hand movements.

Myoelectric computer interfaces to reduce co-contraction after stroke.

A significant factor in impaired motor function caused by stroke is the inability to activate muscles independently. While the pathophysiology behind this co-contraction, sometimes called abnormal muscle synergy, is not clear, reducing the co-contraction could improve overall arm function. This pilot study describes the use of a myoelectric-computer interface (MCI) to retrain arm muscle activation and reduce co-contraction. We found that both healthy subjects and stroke survivors with hemiparesis learned to reduce co-contraction with MCI training. Three out of five stroke survivors experienced some improvement in arm function as well. These results suggest that MCIs could provide a novel, relatively inexpensive paradigm for stroke rehabilitation.

Control of a biomimetic brain machine interface with local field potentials: performance and stability of a static decoder over 200 days.

Brain-machine interfaces (BMIs) have the potential to restore lost function to individuals with severe motor impairments. An important design specification for BMIs to be clinically useful is the ability to achieve high performance over a period of months to years without requiring frequent recalibration. Here, we report the first successful implementation of a biomimetic BMI based on local field potentials (LFPs). A BMI decoder was built from a single recording session of a random-pursuit reaching task for each of two monkeys, and used to control cursor position in real time (online) over a span of 210 days. Performance using this BMI was similar to prior reports using BMIs based on single-unit spikes for 2D cursor control. During this ongoing study, target acquisition rates remained constant (in 1 monkey) or improved slightly (1 monkey) over a 7 month span, and performance metrics of cursor movement (path length and time-to-target) also remained constant or showed mild improvement as the monkeys gained practice. Based on these results, we expect that a stable, high-performance BMI based on LFP signals could serve as a viable alternative to single-unit based BMIs.

Startle stimuli reduce the internal model control in discrete movements.

A well known and major component of movement control is the feedforward component, also known as the internal model. This model predicts and compensates for expected forces seen during a movement, based on recent experience, so that a well-learned task such as reaching to a target can be executed in a smooth straight manner. It has recently been shown that the state of preparation of planned movements can be tested using a startling acoustic stimulus (SAS). SAS, presented 500, 250 or 0 ms before the expected "go" cue resulted in the early release of the movement trajectory associated with the after-effects of the force field training (i.e. the internal model). In a typical motor adaptation experiment with a robot-applied force field, we tested if a SAS stimulus influences the size of after-effects that are typically seen. We found that in all subjects the after-effect magnitudes were significantly reduced when movements were released by SAS, although this effect was not further modulated by the timing of SAS. Reduced after-effects reveal at least partial existence of learned preparatory control, and identify startle effects that could influence performance in tasks such as piloting, teleoperation, and sports.