Construction of Receiver Operating Characteristic Curves when Disease Verification Is Subject to Selection Bias
- 1 June 1984
- journal article
- Published by SAGE Publications in Medical Decision Making
- Vol. 4 (2) , 151-164
- https://doi.org/10.1177/0272989x8400400204
Abstract
The estimates of ROC curves, which are frequently used in the assessment of diagnostic tests, may be biased if the sample is restricted to subjects whose disease status has been definitively verified. A method to provide an unbiased estimate of the ROC curve under these sample conditions is proposed. The new method requires information on the probability distribution of test results in the population from which the verified sample is drawn. It is illustrated using data from a study of computed tomography for fever of uncertain origin.Keywords
This publication has 9 references indexed in Scilit:
- Assessment of Diagnostic Tests When Disease Verification is Subject to Selection BiasPublished by JSTOR ,1983
- The meaning and use of the area under a receiver operating characteristic (ROC) curve.Radiology, 1982
- A prospective evaluation of computed tomography and ultrasound of the pancreas.Radiology, 1982
- A prospective study of computed tomography, ultrasound, and gallium imaging in patients with fever.Radiology, 1981
- ROC Analysis Applied to the Evaluation of Medical Imaging TechniquesInvestigative Radiology, 1979
- Problems of Spectrum and Bias in Evaluating the Efficacy of Diagnostic TestsNew England Journal of Medicine, 1978
- Basic principles of ROC analysisSeminars in Nuclear Medicine, 1978
- Maximum-likelihood estimation of parameters of signal-detection theory and determination of confidence intervals—Rating-method dataJournal of Mathematical Psychology, 1969