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

Use of receiver operating characteri...

Wang, Jiping.

Linked to FindBook

Google Book

Amazon

博客來

Use of receiver operating characteristic curve in medical decision-making.

Record Type:	Language materials, printed : Monograph/item
Title/Author:	Use of receiver operating characteristic curve in medical decision-making./
Author:	Wang, Jiping.
Description:	90 p.
Notes:	Adviser: Howard Rockette.
Contained By:	Dissertation Abstracts International64-09B.
Subject:	Biology, Biostatistics. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3104774
ISBN:	9780496523412

Use of receiver operating characteristic curve in medical decision-making.
Wang, Jiping.

Use of receiver operating characteristic curve in medical decision-making. - 90 p.

Adviser: Howard Rockette.

Thesis (Ph.D.)--University of Pittsburgh, 2003.

Receiver operating characteristic (ROC) curve analysis has been widely applied to evaluate the accuracy of medical instruments, diagnostic procedures or screening tests. Recently, ROC analysis has also been used to evaluate the capability of algorithms based on statistical models to predict outcomes in the public health and medical areas. These ROC analysis procedures use the area under the ROC curve as an index of predictive capability and apply various inferential procedures to test the usefulness of the model. We have investigated this application of ROC methodology to logistic regression models where the parameters were estimated either by maximum likelihood estimation or the linear discriminate function. We demonstrate that the conventional method of assessing the adequacy of a predictive model results in an estimate of the area under the ROC curve (AUC) that is biased upward and the often used test based on the normal distribution to test predictive capability has an elevated type I error. Thus the conventional method results in an overly optimistic assessment of the model. We propose a less biased estimator of AUC, a valid test for the hypothesis AUC = A0 and a method of constructing confidence limits for AUC when the independent variables follow a multivariate normal distribution. Furthermore, the procedure appears robust for deviations from the multivariate normal distributions as long as at least one of the independent variables is normally distributed. A method to construct time dependent ROC based on Cox proportional hazard model was also proposed as a method of investigating changing predictive capability of a diagnostic marker over time.

ISBN: 9780496523412Subjects--Topical Terms:

1018416
Biology, Biostatistics.

Use of receiver operating characteristic curve in medical decision-making.
LDR:02558nam 2200277 a 45 001 963257
005 20110831
008 110831s2003 ||||||||||||||||| ||eng d
020 $a 9780496523412
035 $a (UMI)AAI3104774
035 $a AAI3104774
040 $a UMI $c UMI
100 1 $a Wang, Jiping. $3 1286320
245 1 0 $a Use of receiver operating characteristic curve in medical decision-making.
300 $a 90 p.
500 $a Adviser: Howard Rockette.
500 $a Source: Dissertation Abstracts International, Volume: 64-09, Section: B, page: 4145.
502 $a Thesis (Ph.D.)--University of Pittsburgh, 2003.
520 $a Receiver operating characteristic (ROC) curve analysis has been widely applied to evaluate the accuracy of medical instruments, diagnostic procedures or screening tests. Recently, ROC analysis has also been used to evaluate the capability of algorithms based on statistical models to predict outcomes in the public health and medical areas. These ROC analysis procedures use the area under the ROC curve as an index of predictive capability and apply various inferential procedures to test the usefulness of the model. We have investigated this application of ROC methodology to logistic regression models where the parameters were estimated either by maximum likelihood estimation or the linear discriminate function. We demonstrate that the conventional method of assessing the adequacy of a predictive model results in an estimate of the area under the ROC curve (AUC) that is biased upward and the often used test based on the normal distribution to test predictive capability has an elevated type I error. Thus the conventional method results in an overly optimistic assessment of the model. We propose a less biased estimator of AUC, a valid test for the hypothesis AUC = A0 and a method of constructing confidence limits for AUC when the independent variables follow a multivariate normal distribution. Furthermore, the procedure appears robust for deviations from the multivariate normal distributions as long as at least one of the independent variables is normally distributed. A method to construct time dependent ROC based on Cox proportional hazard model was also proposed as a method of investigating changing predictive capability of a diagnostic marker over time.
590 $a School code: 0178.
650 4 $a Biology, Biostatistics. $3 1018416
650 4 $a Health Sciences, Public Health. $3 1017659
690 $a 0308
690 $a 0573
710 2 $a University of Pittsburgh. $3 958527
773 0 $t Dissertation Abstracts International $g 64-09B.
790 $a 0178
790 1 0 $a Rockette, Howard, $e advisor
791 $a Ph.D.
792 $a 2003
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=3104774