[Mobilizeplans-starstudents] Nov 3, Mobilize Center Seminar
dbush1 at stanford.edu
Mon Oct 31 08:30:57 PDT 2016
The next Mobilize Center Seminar is scheduled for Thursday, November 3, and will feature two speakers who are both Distinguished Mobilize Center postdoctoral fellows: Lukasz Kidzinski and Jessica Selinger. Lukasz will be speaking on Data-Driven Musculoskeletal Models: A Case Study for Modelling Disease Progression from Sparsely Sampled Observations and Jessica will present on Energy Optimization during Human Movement: from Mechanisms to Models to Mobile Health. We look forward to seeing you!
Data-Driven Musculoskeletal Models: A Case Study for Modelling Disease Progression from Sparsely Sampled Observations
Thursday, November 3
noon - 12:30 pm
Y2E2 300, Stanford University
In clinical practice, modeling of treatment outcomes is often based on visits that are sparsely scattered in time. Statistical tools, such as sparse functional principal components, can help extract information from these sparse observations, yet they allow only for univariate predictions of a key indicator. Understanding treatment outcomes often requires analyzing the progression of multiple covariates or more complex objects (e.g., functions or images). Moreover, predictions of a complete object of interest can facilitate communication and interpretability of results.
We illustrate the problem using a dataset of 6066 visits of 2898 patients with cerebral palsy who visited Gillette Children's Hospital as part of routine care. For each visit eleven kinematic joint angles (e.g., knee and hip flexion) per leg were observed during gait. We sought to predict the natural progression of these kinematics, as it may improve predictive power of existing models of treatment outcomes, by providing a better baseline for progression of the disease.
The model we developed will allow us to improve existing models of treatment outcome by incorporating the effect of natural progression in the prediction of long term effects of a therapy. Analysis of entire curves instead of derivative features, together with appropriate visualization techniques using musculoskeletal simulation software such as OpenSim, facilitates qualitative interpretation and communication of results.
This is a joint work with: Apoorva Rajagopal, Jennifer Hicks, Michael H Schwartz, Trevor Hastie, Scott Delp
Lukasz is a postdoctoral researcher in the Mobilize Center at Stanford. His research involves development of statistical methods incorporating biomechanical domain knowledge. In 2014, he received a Ph.D. in mathematical statistics from the Université Libre de Bruxelles, developing spectral methods for functional time series. He obtained two master degrees, in mathematics and computer science, at the University of Warsaw. He is a recipient of a Swiss NSF grant for the project on learning analytics.
Energy Optimization during Human Movement: from Mechanisms to Models to Mobile Health
Thursday, November 3
12:30 - 1:00 pm
Y2E2 300, Stanford University
Perhaps the most fundamental principle underlying the control of locomotion is that humans and other animals move in ways that minimize their energetic cost. That is, we often prefer to move in ways that burn as few calories as possible. Although the principle of energy optimization has been established for decades, little is known about how these preferences are formed and integrated into the control of movement or how they might be altered to encourage more active lifestyles.
In this talk, I will first present work from my PhD demonstrating that humans can continuously optimize their gait to minimize energy expenditure—over very short timescales and in response to remarkably small energetic incentives. To do so, I developed a paradigm where lower limb robotic exoskeletons were used to alter the energetic consequences of walking. I will discuss how experience with these novel energetic gradients, gained through variability in gait, affects optimization and present a simple reinforcement-learning model that captures much of the locomotor behaviour. Next, I will demonstrate how these insights are guiding bio-feedback strategies to improve locomotor learning and the control of assistive robotic technologies. Finally, I will discuss how this work led me to the Mobilize Center and the potential I see for mobile health technology to help shift, or re-weight, our objectives during movement—away from energy conservation toward healthier and more active lifestyles.
Jessica C. Selinger recently joined the Mobilize Center as a Postdoctoral Fellow this September. Prior to that, she completed her PhD in the Department of Biomedical Physiology & Kinesiology at Simon Fraser University. Jessica also holds a MSc in Biomechanics and BSc in Life Sciences from Queen’s University, Canada. Her research interests are focused on understanding the fundamental principles that underlie the neuromechanics of locomotion, as well as the application of these principles to technology that can improve human mobility.
Please see Mobilize Events<http://mobilize.stanford.edu/events/> for a list of upcoming speakers.
Assistant to Professor Scott Delp
NMBL, Mobilize Center, OpenSim
dbush1 at stanford.edu<mailto:dbush1 at stanford.edu>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the Mobilizeplans-starstudents