DiagTest3Grp: An R Package for Analyzing Diagnostic Tests with Three Ordinal Groups

Medical researchers endeavor to identify potentially useful biomarkers to develop marker-based screening assays for disease diagnosis and prevention. Useful summary measures which properly evaluate the discriminative ability of diagnostic markers are critical for this purpose. Literature and existin...

Full description

Bibliographic Details
Main Authors: Jingqin Luo, Chengjie Xiong
Format: Article
Language:English
Published: Foundation for Open Access Statistics 2012-10-01
Series:Journal of Statistical Software
Subjects:
R
Online Access:http://www.jstatsoft.org/v51/i03/paper
id doaj-06bcfb3b04dd4c03878413c53f84e80a
record_format Article
spelling doaj-06bcfb3b04dd4c03878413c53f84e80a2020-11-24T22:09:30ZengFoundation for Open Access StatisticsJournal of Statistical Software1548-76602012-10-01513DiagTest3Grp: An R Package for Analyzing Diagnostic Tests with Three Ordinal GroupsJingqin LuoChengjie XiongMedical researchers endeavor to identify potentially useful biomarkers to develop marker-based screening assays for disease diagnosis and prevention. Useful summary measures which properly evaluate the discriminative ability of diagnostic markers are critical for this purpose. Literature and existing software, for example, R packages nicely cover summary measures for diagnostic markers used for the binary case (e.g., healthy vs. diseased). An intermediate population at an early disease stage usually exists between the healthy and the fully diseased population in many disease processes. Supporting utilities for three-group diagnostic tests are highly desired and important for identifying patients at the early disease stage for timely treatments. However, application packages which provide summary measures for three ordinal groups are currently lacking. This paper focuses on two summary measures of diagnostic accuracy—volume under the receiver operating characteristic surface and the extended Youden index, with three diagnostic groups. We provide the R package DiagTest3Grp to estimate, under both parametric and nonparametric assumptions, the two summary measures and the associated variances, as well as the optimal cut-points for disease diagnosis. An omnibus test for multiple markers and a Wald test for two markers, on independent or paired samples, are incorporated to compare diagnostic accuracy across biomarkers. Sample size calculation under the normality assumption can be performed in the R package to design future diagnostic studies. A real world application evaluating the diagnostic accuracy of neuropsychological markers for Alzheimer’s disease is used to guide readers through step-by-step implementation of DiagTest3Grp to demonstrate its utility.http://www.jstatsoft.org/v51/i03/paperdiagnostic testthree ordinal groupsvolume under the ROC surfaceYouden indexRDiagTest3Grp
collection DOAJ
language English
format Article
sources DOAJ
author Jingqin Luo
Chengjie Xiong
spellingShingle Jingqin Luo
Chengjie Xiong
DiagTest3Grp: An R Package for Analyzing Diagnostic Tests with Three Ordinal Groups
Journal of Statistical Software
diagnostic test
three ordinal groups
volume under the ROC surface
Youden index
R
DiagTest3Grp
author_facet Jingqin Luo
Chengjie Xiong
author_sort Jingqin Luo
title DiagTest3Grp: An R Package for Analyzing Diagnostic Tests with Three Ordinal Groups
title_short DiagTest3Grp: An R Package for Analyzing Diagnostic Tests with Three Ordinal Groups
title_full DiagTest3Grp: An R Package for Analyzing Diagnostic Tests with Three Ordinal Groups
title_fullStr DiagTest3Grp: An R Package for Analyzing Diagnostic Tests with Three Ordinal Groups
title_full_unstemmed DiagTest3Grp: An R Package for Analyzing Diagnostic Tests with Three Ordinal Groups
title_sort diagtest3grp: an r package for analyzing diagnostic tests with three ordinal groups
publisher Foundation for Open Access Statistics
series Journal of Statistical Software
issn 1548-7660
publishDate 2012-10-01
description Medical researchers endeavor to identify potentially useful biomarkers to develop marker-based screening assays for disease diagnosis and prevention. Useful summary measures which properly evaluate the discriminative ability of diagnostic markers are critical for this purpose. Literature and existing software, for example, R packages nicely cover summary measures for diagnostic markers used for the binary case (e.g., healthy vs. diseased). An intermediate population at an early disease stage usually exists between the healthy and the fully diseased population in many disease processes. Supporting utilities for three-group diagnostic tests are highly desired and important for identifying patients at the early disease stage for timely treatments. However, application packages which provide summary measures for three ordinal groups are currently lacking. This paper focuses on two summary measures of diagnostic accuracy—volume under the receiver operating characteristic surface and the extended Youden index, with three diagnostic groups. We provide the R package DiagTest3Grp to estimate, under both parametric and nonparametric assumptions, the two summary measures and the associated variances, as well as the optimal cut-points for disease diagnosis. An omnibus test for multiple markers and a Wald test for two markers, on independent or paired samples, are incorporated to compare diagnostic accuracy across biomarkers. Sample size calculation under the normality assumption can be performed in the R package to design future diagnostic studies. A real world application evaluating the diagnostic accuracy of neuropsychological markers for Alzheimer’s disease is used to guide readers through step-by-step implementation of DiagTest3Grp to demonstrate its utility.
topic diagnostic test
three ordinal groups
volume under the ROC surface
Youden index
R
DiagTest3Grp
url http://www.jstatsoft.org/v51/i03/paper
work_keys_str_mv AT jingqinluo diagtest3grpanrpackageforanalyzingdiagnostictestswiththreeordinalgroups
AT chengjiexiong diagtest3grpanrpackageforanalyzingdiagnostictestswiththreeordinalgroups
_version_ 1725811463879131136