[Mobilizeplans-starstudents] TODAY Mobilize Center Seminar, Ina Fiterau

[Diane Bush]

Our next Mobilize Center Seminar is scheduled for Thursday, April 6th, and features Ina Fiterau, one of our Mobilize Center Distinguished postdoctoral fellows.  She will be presenting “Leveraging Activity Data for the Prediction of Clinical Outcomes.”

Please also note that starting in April, our seminars will be held on a monthly basis (see Spring quarter schedule below). We look forward to seeing you next week and at future seminars!

TITLE:
Leveraging Activity Data for the Prediction of Clinical Outcomes

WHEN:
Thursday, April 6
noon - 1 pm

WHERE:
Y2E2 300, Stanford University

Abstract:
Wrist and hip-worn accelerometers have become standard tools for monitoring movement over extended time periods. They offer the possibility to determine long-term activity patterns that are instrumental for understanding obesity causes, prevention and control. Meaningful analysis of the real world data depends on the identification of intervals when the device is not worn. Lack of true knowledge of non-wear status, together with its similarity with sedentary behavior and sleep, makes this task difficult.

First, I will present an ensemble method for non-wear detection that leverages, in an unsupervised way, the predictions of existing wear/non-wear/sleep classification methods. Since field data are inevitably different from the data on which the predictors were trained or evaluated, their true accuracy is unknown. We derive accuracy estimators, which we obtain based on experts’ consistency of predictions by solving a linear system. We then use the accuracy estimates in a weighting scheme for the ensemble classification, in order to obtain superior predictions compared to the original classifiers.

In the second part of the talk I will describe ShortFuse, a deep learning method which builds representations from temporal features while incorporating the structured information into the deep learning models applied to the temporal features. We show that the inclusion of the structured covariates results in improved performance over the state-of-the-art for biomedical classification tasks. We also study the impact that individual covariates and sequences have on the performance of the learned representations, facilitating expert interpretation.

Upcoming Mobilize Center seminars:

May 4  --  Katherine Heller, Duke University<http://www2.stat.duke.edu/~kheller/>
June 1 – Peter Bailis, Stanford University<https://engineering.stanford.edu/people/peter-bailis>
June 8 – Ambuj Tewari, University of Michigan<https://lsa.umich.edu/stats/people/faculty/tewaria.html>

To keep up-to-date on upcoming speakers and the dates, visit Mobilize Events<http://mobilize.stanford.edu/events/>.


Diane Bush
Assistant to Professor Scott Delp
NMBL, Mobilize Center, OpenSim
Stanford University
dbush1 at stanford.edu<mailto:dbush1 at stanford.edu>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://simtk.org/pipermail/mobilizeplans-starstudents/attachments/20170406/1aada3a7/attachment.html>