The University of Massachusetts Amherst
University of Massachusetts Amherst

Search Google Appliance


Customized Stroke Rehabilitation & Software Development Tool-building

DS Tea
Tuesday, February 2, 2016 - 9:00pm

Ted Smith (PhD Student Advised by Professor Brun)
Title: Build it yourself! Homegrown tools in a large software company (Data Science Tea edition) Abstract: Developers sometimes take the initiative to build tools to solve problems they face. What motivates developers to build these tools? What is the value for a company? Are the tools built useful for anyone besides their creator? We conducted a qualitative study of tool building, adoption, and impact within Microsoft. Our paper presents our findings on the extrinsic and intrinsic factors linked to toolbuilding, the value of building tools, and the factors associated with tool spread. This talk presents these findings and delves into the qualitative methods used in uncovering them.

Hee Tae Jung (PhD Student Advised by Professor Grupen)
Title: Learning Task Space Customization Strategies for Individual Stroke Patients from Demonstration Abstract: The use of robots and serious games has become a popular trend in stroke rehabilitation research. However, despite the acknowledged value of customized service for individual patients, research on programming customized and adaptive therapy for individual patients has received little attention. The goal of the current study is to model the therapist's strategy to customize task space for the individual stroke patient. We propose to use a suit of kinematic parameters to learn a model that classifies the qualitative motor performance that a patient may exhibit while attaining the targets of appropriate difficulty. We further propose to apply the learned model to the estimated kinematic parameters within the entire potential task space. This results in the subspace that is customized for the individual patient. Specifically, we apply Lasso to train a classifier and Gaussian Processes to estimate unobserved kinematic parameters. The experiment results with two patients suggest that the proposed approach may achieve the goal from a single demonstration session and, when used along with robots or serious games, may reinforce the therapeutic goals set for the individual patient by the therapist.