東華大學圖書館 |

語系: 繁體中文

說明(常見問題)

回圖書館首頁

手機版館藏查詢

登入

回首頁

切換: 標籤 | MARC模式 | ISBD

Bayesian semiparametric regression f...

Su, Li.

FindBook

Google Book

Amazon

博客來

Bayesian semiparametric regression for censored and incomplete longitudinal data.

紀錄類型:	書目-語言資料,印刷品 : Monograph/item
正題名/作者:	Bayesian semiparametric regression for censored and incomplete longitudinal data./
作者:	Su, Li.
面頁冊數:	140 p.
附註:	Adviser: Joseph W. Hogan.
Contained By:	Dissertation Abstracts International68-07B.
標題:	Biology, Biostatistics. -
電子資源:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3272065
ISBN:	9780549119395

Bayesian semiparametric regression for censored and incomplete longitudinal data.
Su, Li.

Bayesian semiparametric regression for censored and incomplete longitudinal data. - 140 p.

Adviser: Joseph W. Hogan.

Thesis (Ph.D.)--Brown University, 2007.

Semiparametric regression by Bayesian penalized splines is a useful approach for characterizing the unknown functional relationships in regression analysis, when the prior information and substantive knowledge about the functionals are not available. Armed with the modeling flexibility from Bayesian penalized splines, we develop three types of Bayesian semiparametric regression models in various longitudinal study settings with censored and incomplete data. The proposed methods are motivated by scientific questions arising in HIV and AIDS research, and can be applied in different scientific research areas.

ISBN: 9780549119395Subjects--Topical Terms:

1018416
Biology, Biostatistics.

Bayesian semiparametric regression for censored and incomplete longitudinal data.
LDR:03340nam 2200301 a 45 001 962205
005 20110830
008 110831s2007 ||||||||||||||||| ||eng d
020 $a 9780549119395
035 $a (UMI)AAI3272065
035 $a AAI3272065
040 $a UMI $c UMI
100 1 $a Su, Li. $3 1285249
245 1 0 $a Bayesian semiparametric regression for censored and incomplete longitudinal data.
300 $a 140 p.
500 $a Adviser: Joseph W. Hogan.
500 $a Source: Dissertation Abstracts International, Volume: 68-07, Section: B, page: 4212.
502 $a Thesis (Ph.D.)--Brown University, 2007.
520 $a Semiparametric regression by Bayesian penalized splines is a useful approach for characterizing the unknown functional relationships in regression analysis, when the prior information and substantive knowledge about the functionals are not available. Armed with the modeling flexibility from Bayesian penalized splines, we develop three types of Bayesian semiparametric regression models in various longitudinal study settings with censored and incomplete data. The proposed methods are motivated by scientific questions arising in HIV and AIDS research, and can be applied in different scientific research areas.
520 $a We first propose a semiparametric regression model for longitudinal binary data that are irregularly spaced over time across individuals due to ignorable missingness. With flexible modeling by semiparametric formulations in both marginal mean and serial dependence structures, our likelihood-based approach can reduce the bias in marginal mean estimation induced by ignorable missingness, as shown by the simulation study.
520 $a The second part of this dissertation focuses on an important question in AIDS research, that is, the effect of Hepatitis C virus (HCV) coinfection on post-treatment HIV dynamics. Utilizing all available information in HIV cohort data and considering the biological background of HIV dynamics, we develop a joint model for interval-censored treatment initiation time and viral suppression time after treatment in addition to the longitudinal CD4 count process, where the evolution of the CD4 count process is captured by Bayesian penalized splines. The findings based on the data from HIV Epidemiology Research Study offer insights into the subject field and are consistent with the current knowledge on HCV coinfection.
520 $a Lastly, we propose a framework of mixtures of varying coefficient models to handle continuous-time informative dropout in longitudinal data. Following a mixture-modeling approach for dealing with informative dropout, we model the unknown functional relationship between model parameters in longitudinal outcome process and continuous dropout times by penalized splines. Using two examples on continuous and binary longitudinal data, we show that our framework can adjust marginal covariate effects for the potential selection bias induced by informative dropout. The strategies of sensitivity analysis regarding unverifiable model assumptions are also demonstrated by example.
590 $a School code: 0024.
650 4 $a Biology, Biostatistics. $3 1018416
690 $a 0308
710 2 $a Brown University. $3 766761
773 0 $t Dissertation Abstracts International $g 68-07B.
790 $a 0024
790 1 0 $a Hogan, Joseph W., $e advisor
791 $a Ph.D.
792 $a 2007
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3272065