東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back to Search results for [ null ]

Switch To: Labeled | MARC Mode | ISBD

Methods for confounding adjustment i...

The University of Chicago.

Linked to FindBook

Google Book

Amazon

博客來

Methods for confounding adjustment in time series data: Applications to short term effects of air pollution on respiratory health.

Record Type:	Language materials, printed : Monograph/item
Title/Author:	Methods for confounding adjustment in time series data: Applications to short term effects of air pollution on respiratory health./
Author:	Zibman, Chava.
Description:	104 p.
Notes:	Adviser: Vanja Dukic.
Contained By:	Dissertation Abstracts International69-07B.
Subject:	Health Sciences, Epidemiology. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3322663
ISBN:	9780549744559

Methods for confounding adjustment in time series data: Applications to short term effects of air pollution on respiratory health.
Zibman, Chava.

Methods for confounding adjustment in time series data: Applications to short term effects of air pollution on respiratory health. - 104 p.

Adviser: Vanja Dukic.

Thesis (Ph.D.)--The University of Chicago, 2008.

This work explores methods for confounding adjustment in time series models. First, we will offer an analysis of daily time-series of asthma-prescriptions counts for Chicago's Medicaid population (aggregated at the ZIP-code level), as a function of air pollution during four summers. We employ a Bayesian hierarchical semi-parametric Poisson regression model, a generalized additive model where adjustment for confounding is implemented via a time-varying smooth function, which accounts for the slowly varying unobserved confounders. This function is represented via a pre-specified number of spline bases. We specify the prior distribution for the variance of basis coefficients ("smoothness parameter") to control the amount of confounding adjustment. In this manner, we are also able to account for the uncertainty regarding how much confounding adjustment is performed.

ISBN: 9780549744559Subjects--Topical Terms:

1019544
Health Sciences, Epidemiology.

Methods for confounding adjustment in time series data: Applications to short term effects of air pollution on respiratory health.
LDR:02932nam 2200289 a 45 001 854343
005 20100702
008 100702s2008 ||||||||||||||||| ||eng d
020 $a 9780549744559
035 $a (UMI)AAI3322663
035 $a AAI3322663
040 $a UMI $c UMI
100 1 $a Zibman, Chava. $3 1020691
245 1 0 $a Methods for confounding adjustment in time series data: Applications to short term effects of air pollution on respiratory health.
300 $a 104 p.
500 $a Adviser: Vanja Dukic.
500 $a Source: Dissertation Abstracts International, Volume: 69-07, Section: B, page: 4249.
502 $a Thesis (Ph.D.)--The University of Chicago, 2008.
520 $a This work explores methods for confounding adjustment in time series models. First, we will offer an analysis of daily time-series of asthma-prescriptions counts for Chicago's Medicaid population (aggregated at the ZIP-code level), as a function of air pollution during four summers. We employ a Bayesian hierarchical semi-parametric Poisson regression model, a generalized additive model where adjustment for confounding is implemented via a time-varying smooth function, which accounts for the slowly varying unobserved confounders. This function is represented via a pre-specified number of spline bases. We specify the prior distribution for the variance of basis coefficients ("smoothness parameter") to control the amount of confounding adjustment. In this manner, we are also able to account for the uncertainty regarding how much confounding adjustment is performed.
520 $a Next, we develop a data-driven method for choosing a subset of trigonometric basis functions to be used for representation of the non-parametric (confounder) component. We seek to identify those bases which explain much of the variation in the response, but not the variation that is already explained by the predictor of interest. Our approach involves looking at the spectral representations of the observed variables to select the set of candidate trigonometric functions to be used to adjust for confounding. In this way, we can account for higher frequency confounding using fewer basis functions than what might be required using splines. High-frequency confounding techniques may be of particular relevance to air-pollution studies. This approach is studied in a series of simulation studies, where we compare several selection criteria for choosing the number of bases. We conclude by applying this method again to the asthma data. Analyses of the pollution and asthma data consistently fail to show any significant effects of ozone at least at the relatively low levels of pollution observed in Chicago.
590 $a School code: 0330.
650 4 $a Health Sciences, Epidemiology. $3 1019544
650 4 $a Statistics. $3 517247
690 $a 0463
690 $a 0766
710 2 $a The University of Chicago. $3 1017389
773 0 $t Dissertation Abstracts International $g 69-07B.
790 $a 0330
790 1 0 $a Dukic, Vanja, $e advisor
791 $a Ph.D.
792 $a 2008
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3322663