ROC CURVES WITH MULTIPLE MARKER MEASUREMENTS

Authors
Citation
Pa. Murtaugh, ROC CURVES WITH MULTIPLE MARKER MEASUREMENTS, Biometrics, 51(4), 1995, pp. 1514-1522
Citations number
10
Categorie Soggetti
Statistic & Probability","Statistic & Probability
Journal title
ISSN journal
0006341X
Volume
51
Issue
4
Year of publication
1995
Pages
1514 - 1522
Database
ISI
SICI code
0006-341X(1995)51:4<1514:RCWMMM>2.0.ZU;2-L
Abstract
Properties of receiver operating characteristic (ROC) curves are explo red for markers that are measured repeatedly, through space or time, f or each subject. The true underlying response, positive or negative, o f each subject is assumed to be constant across marker measurements, a nd is determined from assessment of some ''gold standard.'' A marker-b ased test is considered positive when at least one of a subject's mark er values exceeds a designated cutoff. If subjects with positive and n egative underlying responses differ in the number of marker measuremen ts per subject, the ROC curve for a noninformative marker is bowed abo ve or below the diagonal line representing the ''null'' curve for a ma rker that is measured just once per subject. If subjects with negative responses tend to have more measurements than those with positive res ponses, the ROC curve for even an informative marker may lie beneath t he curve for the same marker measured once per subject. The form of th e ROC curve for a marker used in this way is strongly influenced by th e strength of the correlation of measurements within subjects.