Nicolas Schweighofer - US grants

[unreadable] DESCRIPTION (provided by applicant): Despite tremendous progress in psychology and neuroscience, physical therapists treating patients with stroke still rely on unspecific guidelines to determine task practice schedules for functional motor skill reacquisition. The present proposal is novel in defining specific rehabilitation practice protocols derived from psychological learning theory, with particular emphasis on the "micro-scheduling" of the practice trials to enhance learning. Previous research in word-learning and motor skill learning shows that practice schedules 1) in which the task presentations are variable, or 2) in which the intervals between presentations of individual tasks are expanded, significantly enhance long-term retention and generalization of learning. In this pilot study, the PIs use a manual grasping task, and propose to demonstrate with participants who are recovering from stroke-hemiparesis that massed, random, and expanded variable practice schedules affect differentially the acquisition performance and the learning of motor skills. Further, using a stratified experimental design, they aim to show an interaction between practice schedule and impairment severity such that the best practice schedule depends on stroke severity. The hypothesis is that subjects with mild deficits benefit most from expanded variable practice, whereas subjects with moderate deficits benefit most from massed practice schedules. [unreadable] [unreadable]

The goal of "Skill Acquisition Through Interactive Avatars" is to create a computer-based humanoid simulation that can teach humans how to move. The hardware requirements for such as system in its final stage would be an inexpensive state-of-the-art personal computer equipped with a camera system. The avatar will be able to demonstrate movements to its user, monitor the execution of these movements by the user, and suggest corrections in case of inadequate performance. In order to be effective, the avatar will take into account neuroscientific knowledge about the organization of human motor control and human motivation during learning. This research will be useful in a large number of applications, including rehabilitation of movement-impaired patients (e.g., stroke-patients), sport and exercise education, dance instruction, childcare and special needs education, and interactive entertainment industry. Additionally, the technology developed for this project has the potential to pioneer new algorithms for autonomous robot control using "teaching from demonstration", to contribute to the development of automated surveillance systems for human environments, to the generation of humanoid computer simulations, and also to gaining new insights into biological motor control and the functioning of the nervous system.
As intellectual merits, it will be necessary to gain new understanding of how to recognize and classify human movement from real-time motion capture, how to create skilled movement, and how to teach humans effectively. The basic research problems of understanding human movement, both in terms of movement perception as well as movement generation, are central to advancing information technology in human-computer as well as human-robot interaction, i.e., the creation of autonomous artificial perception and movement systems. Research on intrinsic motivation in motor learning will advance important insights into how to help humans with learning disabilities, but also how to create machines that are motivating to interact with.
As broader impact, the research of this project will make important steps towards creating interactive environments that can assist people in their professional and private lives. Such technology will soon become a major component of our world, starting with clinical, entertainment, and business scenarios, and finally finding its way into private households. Understanding how to create and teach skilled movement in a user friendly way will be useful in building autonomous robot systems that can assist humans, entertain humans, replace humans in hazardous environments, rescue humans, or simply become a companion

This proposal takes an innovative stand in proposing that computational neuroscience can guide the design of effective and motivating adaptive training schedules for motor tasks. Although the goals of learning are generalization and long-term retention, current performance is a poor predictor of these learning goals. In this proposal, the general hypothesis is that adaptive scheduling of multiple motor tasks, based on long-term memory predictions, can enhance learning and that these long-term predictions are most effective when derived from neurally-based computational models of the motor memory system. The two specific research objectives of the proposed work are 1) to determine the mechanisms of multiple motor adaptation in humans, using a combined computational and behavioral approach, and 2) to investigate methods for tailoring training schedules to individual learners using multiple motor adaptation tasks.

The proposed research is in line with two of the 14 grand challenges for the 21st century, identified by the U.S. National Academy of Engineering (NAE): 'reverse-engineering the brain' and 'advancing personalized learning.' The work proposed considers these challenges as related and that 'advancing personalized learning' must be based on an understanding of the learning and motivational systems of the brain. Although there have been a few attempts to generate learning programs along these lines, this type of research is still in its infancy and is mostly based on descriptive models of learning and memory. Here, the PI will reverse-engineer the motor memory system with computational models that are both neurally and behaviorally valid and relevant. This has the potential to be useful in a large number of applications, including rehabilitation of movement-impaired patients (e.g., stroke patients), sport and exercise education, dance instruction, and special needs education.

DESCRIPTION (provided by applicant): Because each patient post stroke has unique impairments and function, it is important to depart from a "one size fits all" approach to rehabilitation. Although there is now evidence that motor therapy can improve function and use of the more affected limb for patients with moderate to mild impairments, change in use in the months following therapy is variable: for some patients there is an increase in use, but for others a decrease in use. Our long-term goal is to determine prospectively the dose of therapy that leads to further improvements of use after therapy for individual patients while keeping cost at reasonable levels - we call this dose critical dose. Our objective here is to investigate long-term predictions of use as a function of the dose of therapy and of the patient's neurological and behavioral characteristics. Our general hypothesis is that, for a subset of patients, there is a threshold level for arm and hand function to be achieved after therapy, such that if therapy brings function above this threshold, spontaneous use and function will reinforce each other in a virtuous circle. We formulated our hypothesis based on computational models that demonstrate such a threshold and account for existing data. We will address our general hypothesis and accomplish our objective with two aims. Aim 1. Determine the effect of a distributed dose of therapy on immediate and long-term gains in upper extremity use. Aim 2: Develop a means to compute the critical dose for individual patients. With the first aim, we will test our general hypothesis and generate relevant clinical data of function and use. The data will then be used in the second aim to develop a predictive model, based on the Extended Kalman Filter, of long-term arm and hand use as a function of the dose of therapy as well as behavioral and neurological data. Our proposed work is significant because such a predictive model of stroke recovery, once subjected under future funding to clinical trials, can be used to influence policy regarding the necessary dosage of effective treatments at a reasonable cost for the growing number of persons who have survived stroke. This work will also make an important neuro- scientific contribution as we will model, and test behaviorally, the causal and adaptive linkages between the decisions to use the affected arm and recovery of motor function. PUBLIC HEALTH RELEVANCE: Stroke is the leading cause of disability in the US, and about 65% of stroke survivors experience mostly unilateral long-term upper extremity functional limitations. Improving use of the more affected arm is important because difficulty in using this arm in daily tasks has been associated with reduced quality of life. Although there is now definite evidence that motor training can improve recovery of function and use for patients with moderate to mild impairments, in some cases, the gains in arm use due to therapy are small and may not be sustained. We propose here a novel evidence-based method to allow therapists to determine in advance of treatment and for individual patients, the dose of therapy that maximizes the efficacy of treatment while keeping cost at reasonable levels for each patient.

This project addresses automated, adaptive-rehabilitation and training assessment for victims of neurological disorders that impact motor functions.

As the population ages and survival rates post-stroke increase, there is a need to improve the effectiveness and efficiency of motor therapy. This I-Corps team's goal is to improve arm and hand function following neurological disorders that affect the motor system, such as stroke, Parkinson's disease, and traumatic brain injury (TBI). This I-Corps team has developed a new device, MOTION-REACH, that provides automated and high-intensity arm motor training based on continuous accurate measurements of hand path and arm/trunk kinematics acquired from a 3D camera. This allows for objective assessment of reaching performance, minimization of compensatory movements, and adaptive training. Adaptive training with MOTION-REACH is intuitive and can be performed by patients alone or in a semi-supervised clinical environment. MOTION-REACH also provides fast, high-precision, and validated assessments. Session-by-session reports and overall reports for the patient, the therapist, and the health insurance provider are generated and saved on a database in the cloud.

The MOTION-REACH device will expand therapist time availability because the device can be used by an unsupervised patient and does not require 1-to-1 supervision by the therapist, as it automatically monitors and prompts patients to perform movements correctly. Customer segments will include hospital-based and private physical therapy clinics. The device can be sold into the private patient marketplace, to be used in-home by private-pay patients under the remote guidance of a physical therapist. The value proposition for the therapist derives from the ability to have patients performing highly effective therapeutic exercises without direct therapist supervision. The value proposition for the patient and the patient's medical insurance provider is that optimal and sustainable improvement is achieved in fewer therapy sessions than with traditional stroke recovery and other therapy techniques. This team's long-term goal is to develop personalized neurorehabilitation, similar to the successful drug dosing control in clinical pharmacology.

Project Summary The large variability in lesions, impairment, and responsiveness to training following stroke has hindered the development of principled and cost-effective approaches to neuro-rehabilitation of the upper extremity. Our long-term goal is to develop predictive personalized neurorehabilitation therapy based on large data sets. This proposal is based on a unique opportunity to design and execute a large neuro-rehabilitation cohort study at a relatively low cost. Building on our established US-French collaborations, with interdisciplinary expertise in neurorehabilitation, brain imaging, dynamical systems, and statistical learning, we will predict recovery and individualize therapy with the following novel three-pronged approach. In Aim 1, we will develop a database of clinical and neural patient characteristics, treatments, and outcomes from 500 patients post-stroke receiving upper extremity rehabilitation therapy with the ARMEO Spring device (a gravity compensating exoskeleton) in routine clinical care. Inclusion criteria will be as broad as possible to include patients with a large variety of brain lesions, as assessed by state- of-the-art magnetic resonance imaging (MRI) and functional MRI scans. In aim 2, using the database, we will predict long-term changes in upper extremity outcomes as a function of patient's characteristics and treatment using dynamical models that link motor learning to recovery. The final models will expand and combine previous computational models of motor learning at small time scales with models of recovery at long time scales, and will include mixed effects to accurately predict long-term recovery for individual patients. In aim 3, based on these predictions, we will perform a feasibility study aimed at individualizing upper extremity rehabilitation to maximize recovery. Given a new patient, characterized by a number of baseline characteristics that predict recovery, we will select the schedule of treatment that was the most effective for similar patients in the database. The recovery models and scheduling methods developed in this proposal will provide the basis for future clinical software that suggests timing, dosage, and content of therapy from early clinical data, kinematic performance, and routine scans. Such an approach will transform neurorehabilitation programs because the clinician, patient, and insurance company will be able to determine effective treatments while reducing costs.

This project aims to improve motor learning by customizing the practice schedule for each learner. The investigators propose a novel algorithm that will generate practice schedules. The schedules will depend on the learner’s unique attributes, data from other learners, and the possible limits on the total amount of training. Using an online motor learning task, the investigators will test the algorithm with adults across the lifespan recruited from the community. In future applications, the algorithm has the strong potential to improve learning in sports, technical training, and surgical technique training. This work is also relevant for treating motor symptoms in conditions such as stroke, spinal cord injury, traumatic brain injury, and Parkinson’s disease. The proposed research will provide educational opportunities for students from high school to Ph.D. across disciplines such as artificial intelligence, brain science, and psychology.<br/> <br/>The investigators propose a novel, theoretically sound, and self-improving algorithm to personalize motor adaptation training. The algorithm will select the daily dose and schedule of training that maximizes the long-term performance predicted by a dynamical model of motor memory, given the learner’s unique characteristics, data from other learners, and constraints on both total and daily doses of practice. The investigators will compare the predictive abilities of models with different memory time scales via cross-validation. The investigators will then pilot the training algorithm with college students (ages 18-30) who will learn an online motor adaptation task over 3 days, followed by a 1-month post-training retention test. Then, the investigators will test the efficacy of personalized learning by deploying the online task to the community. Sex, age, baseline movement variance, genetic factors (BDNF, APOE genes), time of day, and spatial memory covariates will be incorporated into the model to improve predictions. Because the algorithm is self-improving, the investigators will compare the performance in the 1-month post-training test of each new sub-group of 30 participants to that of the preceding sub-group. Furthermore, to test the efficacy of the adaptive schedule relative to a “one-size-fits-all” schedule, the investigators will compare the performance of the last group to that of an additional sub-group of matched participants (n=30) who will receive three days of equally-dosed practice.<br/><br/>This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.