Gelsy Torres-Oviedo, PhD - US grants

Technical description:

A major issue in rehabilitation robotics is the fact that motor learning acquired on devices like robots and treadmills is highly specific and does not improve patients? movements during ?real life? situations. How to manipulate the generalization of robotic-assisted motor learning? This is an interesting challenge for engineers and an important question for clinicians. To understand the poor transfer of robotic-assisted motor learning to natural movements the PI proposes to investigate what the device changes and how to manipulate it. This will be done though a novel application of factorization analyses that are conventionally used for image processing. The PI specifically proposes to quantify changes in muscle activity post-stroke when subjects learn a new walking pattern on a split-belt treadmill (research objective 1). This treadmill has two belts that can move at different speeds to create novel environmental conditions and induced motor learning. Then, the PI will identify what aspects of the muscle activation patterns learned on the split-belt treadmill generalize to natural walking, and whether the acquisition phase on the split-belt treadmill can be manipulated to increase the generalization effects (research objective 2). Anticipated results will advance the current knowledge of mechanisms available to post-stroke subjects for changing their aberrant muscle activity and the generalization of those changes to natural walking.

Broader significance and importance:

The potential benefit of this proposal is to advance our knowledge of the ?rules? for motor adaptation and generalization in the intact and post-stroke human motor system. This should allow us to understand how to use technology to train individuals, and ultimately rehabilitate their walking and movement control in general. The risks to the subjects participating in the proposed studies are minimal, and the benefit to society could be substantial. According to the Center for Disease Control and Prevention there are nearly 650,000 post-stroke survivors every year, making stroke the leading cause for long-term disability in the United States. This proposal has the potential to help develop effective and efficient rehabilitation treatments for post-stroke survivors. This would directly benefit patients by improving their movements and reducing their medical care costs, which impose a substantial economic burden to the individuals and society.

Broadening participation of underrepresented groups in engineering:

Importantly, the PI will use the research objectives in this proposal as means to expose students from underrepresented groups in engineering to solve real rehabilitation needs (educational objective 1), engage them in research that applies engineering tools to inform the field of physical rehabilitation (educational objective 2), and prepare them to continue graduate education (educational objective 3). These educational objectives will be fulfilled through the Pre-PhD Research Experience at Pitt (PREP) program developed by the PI. To implement her program the PI will recruit, mentor, and prepare undergraduate students from Hispanic serving Universities to pursue advanced degrees. She will also incorporate her research activities with the program INVESTING NOW, which prepares high schools students (89% from minority groups and 47% females) mostly from underrepresented groups in engineering to pursue engineering degrees. In sum, the PI?s educational objectives will prepare students from underrepresented groups in engineering to pursue advanced degrees, which will in turn redefine the boundaries of education and professional development available to underrepresented groups in science and engineering.

This research has been funded through the Broadening Participation Research Initiation Grants in Engineering solicitation, which is part of the Broadening Participation in Engineering Program of the Engineering Education and Centers Division.

? DESCRIPTION (provided by applicant): Step asymmetry post-stroke (i.e., limp) substantially affects the quality of life of stroke survivors because it impairs patients' mobility, thereby limiing their daily activities and increasing their dependency on others. Consequently, a primary interest for patients, clinicians, and researchers is to correct the step asymmetry in stroke survivors. Promising studies show that patients can re-learn to walk symmetrically if their step asymmetry is exaggerated with a split-belt treadmill that moves the legs at different speeds. While these results are encouraging, gait improvements are highly contextual and do not persist when walking over ground. To address this critical issue for gait rehabilitation, the PI is proposing a combination of computational and experimental approaches to identify key factors regulating the generalization of locomotor learning after stroke. The PI's central hypothesis is that inherent features from one's movement (e.g., kinematic errors and walking speed) regulate the generalization of locomotor learning. This hypothesis was formulated on the basis of the PI's preliminary data showing more generalization of treadmill learning to over ground walking when kinematic errors or walking speed during split-belt adaptation are similar to those naturally experienced. In the proposed computational approach, model inputs are errors that subjects experience during split-belt walking (for example, unexpected leg motions disturbing one's balance), model outputs are actions to correct these errors (for example, a larger step to prevent falling). The mathematical relationship between inputs and outputs is used to predict the effect of error size (Aim 1) and walking speed (Aim 2) on the generalization of learning in an individual basis. Once factors mediating the generalization of learning are identified, they can be harnessed to develop interventions that improve the gait of stroke survivors during real-life situations. PI qualifications: the PI is a prolific and creative bioengineer. Her first class trainng in physics, biomechanics, and neuroscience, in addition to her strong interest in rehabilitation make her the adequate individual for doing the proposed work. Her studies in human motor control are well recognized (>700 citations; h-index 11) in a relatively short, but highly productive academic career. Through this award she will receive mentorship from two extraordinary investigators with complementary expertise: Michael Boninger, MD, PhD. (clinical rehabilitation) and Reza Shadmehr, PhD (computational motor control). They will serve as primary co-mentors. In addition the PI will receive mentorship from an expert panel of collaborators including Dr. Subashan Perera (biostatistics), Dr. Steven Graham (neurology), Dr. Julie Fiez (neuropsychology) and Dr. Skidmore (post-stroke rehabilitation). Thus, this award will provide the mentorship and career development allowing the PI to become an independent researcher able to compete for R01-level funding to study gait deficits post-stroke through computational modeling.