東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Modeling of silent events and evalua...

Li, Jiang.

Linked to FindBook

Google Book

Amazon

博客來

Modeling of silent events and evaluation of biomarkers in a prospective study with scheduled follow-up.

Record Type:	Language materials, printed : Monograph/item
Title/Author:	Modeling of silent events and evaluation of biomarkers in a prospective study with scheduled follow-up./
Author:	Li, Jiang.
Description:	188 p.
Notes:	Adviser: Michael LaValley.
Contained By:	Dissertation Abstracts International67-04B.
Subject:	Biology, Biostatistics. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3214960
ISBN:	9780542661679

Modeling of silent events and evaluation of biomarkers in a prospective study with scheduled follow-up.
Li, Jiang.

Modeling of silent events and evaluation of biomarkers in a prospective study with scheduled follow-up. - 188 p.

Adviser: Michael LaValley.

Thesis (Ph.D.)--Boston University, 2006.

The goal of this thesis is to estimate the probability that a 'silent' event happens in a given time period and to model and evaluate associated longitudinal biomarker measurements in a prospective study with regular follow-up visits. Subjects will be evaluated for biomarker values and for the event of interest at each visit. Events are 'silent', in that they are not apparent to the subjects, and there are no specific dates for their occurrence. All that is known about the timing of an event for a subject is that the event had not occurred at the previous visit, but it has occurred by the time of the current visit. I also assume that there are regularly scheduled follow-up visits for the study, but the actual visit times do not occur exactly according to the schedule, and are randomly distributed. First, I compare a number of models for interval-censored data and two discrete time survival models to determine which is preferable for such a study. It is shown that discrete time models can perform as well as the interval censoring models in certain situations. Joint models have been proposed to analyze both longitudinal and event data from the same study in a single analysis. In the second part of this thesis I propose a new joint model using a discrete time survival model for event times, with a biomarker assumed to follow a linear mixed effects model. The performance of this joint model is evaluated in simulation studies and shown to improve estimation and prediction compared to separate analyses of event times and biomarker trajectories. In the third part of my thesis I employ receiver operator characteristic (ROC) curves to characterize the predictive accuracy of biomarkers for discrete event times. Two methods for ROC analysis are developed and contrasted in simulated data.

ISBN: 9780542661679Subjects--Topical Terms:

1018416
Biology, Biostatistics.

Modeling of silent events and evaluation of biomarkers in a prospective study with scheduled follow-up.
LDR:02690nam 2200265 a 45 001 974289
005 20110929
008 110929s2006 eng d
020 $a 9780542661679
035 $a (UMI)AAI3214960
035 $a AAI3214960
040 $a UMI $c UMI
100 1 $a Li, Jiang. $3 1298231
245 1 0 $a Modeling of silent events and evaluation of biomarkers in a prospective study with scheduled follow-up.
300 $a 188 p.
500 $a Adviser: Michael LaValley.
500 $a Source: Dissertation Abstracts International, Volume: 67-04, Section: B, page: 1773.
502 $a Thesis (Ph.D.)--Boston University, 2006.
520 $a The goal of this thesis is to estimate the probability that a 'silent' event happens in a given time period and to model and evaluate associated longitudinal biomarker measurements in a prospective study with regular follow-up visits. Subjects will be evaluated for biomarker values and for the event of interest at each visit. Events are 'silent', in that they are not apparent to the subjects, and there are no specific dates for their occurrence. All that is known about the timing of an event for a subject is that the event had not occurred at the previous visit, but it has occurred by the time of the current visit. I also assume that there are regularly scheduled follow-up visits for the study, but the actual visit times do not occur exactly according to the schedule, and are randomly distributed. First, I compare a number of models for interval-censored data and two discrete time survival models to determine which is preferable for such a study. It is shown that discrete time models can perform as well as the interval censoring models in certain situations. Joint models have been proposed to analyze both longitudinal and event data from the same study in a single analysis. In the second part of this thesis I propose a new joint model using a discrete time survival model for event times, with a biomarker assumed to follow a linear mixed effects model. The performance of this joint model is evaluated in simulation studies and shown to improve estimation and prediction compared to separate analyses of event times and biomarker trajectories. In the third part of my thesis I employ receiver operator characteristic (ROC) curves to characterize the predictive accuracy of biomarkers for discrete event times. Two methods for ROC analysis are developed and contrasted in simulated data.
590 $a School code: 0017.
650 4 $a Biology, Biostatistics. $3 1018416
690 $a 0308
710 2 0 $a Boston University. $3 1017454
773 0 $t Dissertation Abstracts International $g 67-04B.
790 $a 0017
790 1 0 $a LaValley, Michael, $e advisor
791 $a Ph.D.
792 $a 2006
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3214960