|
Activity Number:
|
81
- Contributed Poster Presentations: Section on Statistics in Epidemiology
|
|
Type:
|
Contributed
|
|
Date/Time:
|
Monday, August 3, 2020 : 10:00 AM to 2:00 PM
|
|
Sponsor:
|
Section on Statistics in Epidemiology
|
|
Abstract #312499
|
|
|
Title:
|
Non-Parametric Bayesian Built Environment Heterogenous Effect Estimation
|
|
Author(s):
|
Adam Peterson* and Brisa Sanchez
|
|
Companies:
|
University of Michigan and Drexel University
|
|
Keywords:
|
Bayesian;
Non-Parametric;
Network;
Built Environment;
Urban
|
|
Abstract:
|
The presence of specific urban environment resources such as food vendors and recreation centers may impact diet and physical activity, thereby affecting the subsequent risk of important health conditions like cardiovascular disease. The relevant distance or spatial scale, at which theses resources impact subjects has been an area of active research for some time now. However, most approaches have focused on identifying a population level spatial scale even though it is understood that these resources likely affect different members of a population differently. In order to identify this heterogenous effect we adapt the common functional regression framework by placing a dirichlet process prior on the regression coefficients. We demonstrate this modeling approach using data from the California Department of Education and U.S. census for outcome and socio-economic covariate measures. We use a proprietary source for the built environment data.
|
Authors who are presenting talks have a * after their name.
