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

Combining multiple biomarker models ...

Yuan, Zheng.

Linked to FindBook

Google Book

Amazon

博客來

Combining multiple biomarker models in logistic regression and survival analysis: Model combining and instability measure.

Record Type:	Language materials, printed : Monograph/item
Title/Author:	Combining multiple biomarker models in logistic regression and survival analysis: Model combining and instability measure./
Author:	Yuan, Zheng.
Description:	133 p.
Notes:	Adviser: Debashis Ghosh.
Contained By:	Dissertation Abstracts International68-02B.
Subject:	Biology, Biostatistics. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3253441

Combining multiple biomarker models in logistic regression and survival analysis: Model combining and instability measure.
Yuan, Zheng.

Combining multiple biomarker models in logistic regression and survival analysis: Model combining and instability measure. - 133 p.

Adviser: Debashis Ghosh.

Thesis (Ph.D.)--University of Michigan, 2007.

All methods are assessed using simulation studies and illustrated with an application to a dataset in prostate cancer.Subjects--Topical Terms:

1018416
Biology, Biostatistics.

Combining multiple biomarker models in logistic regression and survival analysis: Model combining and instability measure.
LDR:03086nam 2200301 a 45 001 974398
005 20110929
008 110929s2007 eng d
035 $a (UMI)AAI3253441
035 $a AAI3253441
040 $a UMI $c UMI
100 1 $a Yuan, Zheng. $3 1285278
245 1 0 $a Combining multiple biomarker models in logistic regression and survival analysis: Model combining and instability measure.
300 $a 133 p.
500 $a Adviser: Debashis Ghosh.
500 $a Source: Dissertation Abstracts International, Volume: 68-02, Section: B, page: 0700.
502 $a Thesis (Ph.D.)--University of Michigan, 2007.
520 $a All methods are assessed using simulation studies and illustrated with an application to a dataset in prostate cancer.
520 $a In medical research, there is great interest in developing methods for combining multiple biomarkers to predict clinical outcomes. We argue that the selection of markers should be considered. A major concern of selection methods is that uncer tainty in variable selection process is ignored, which could lead to poor prediction. An alternative approach is to combine predictions from multiple models, called "model combining". The primary interest of this dissertation is to combine the information from multiple biomarkers using model combining methods.
520 $a In Chapter 2, we consider binary outcomes. A model combining method, Adaptive Regression by Mixing with Screening (ARMS), is proposed for the classification of binary outcomes in logistic regression. It works by considering weighted combinations of logistic regression models; five different weighting schemes are developed. The weights and method are justified using decision theory and risk bound results.
520 $a Chapter 3 is concerned with censored survival outcomes. We propose a model combining method, Adaptive Regression by Mixing with Adaptive Screening (ARMAS), for survival data. It works by considering weighted combinations of Cox survival models. To reduce the set of models for combining in high-dimensional data, we propose an adaptive screening procedure in ARMAS based on the idea of adaptive model selection. In addition, we propose a modified version of ARMAS based on the imputed data from multiple imputations, called ARMAS-impute, which does not reply on the proportional hazards assumption.
520 $a Chapter 4 is devoted to the instability measure and shrinkage estimation for model combining in logistic regression. We propose an instability measure to capture the uncertainty of model selection procedures, called GDF Instability (GDFI), based on the concept of generalized degrees of freedom (GDF). A rule of thumb for the use of the GDFI is suggested. We also propose a shrinkage version of ARMS, ARMS-shrink, applying shrinkage estimation into the ARMS algorithm.
590 $a School code: 0127.
650 4 $a Biology, Biostatistics. $3 1018416
690 $a 0308
710 2 0 $a University of Michigan. $3 777416
773 0 $t Dissertation Abstracts International $g 68-02B.
790 $a 0127
790 1 0 $a Ghosh, Debashis, $e advisor
791 $a Ph.D.
792 $a 2007
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3253441