<- file stat 97roc.html ->
ROC Refs, Metz (1997)
<![CDATA[
"Radio Operating Characteristics" curves -
sensitivity vs specificty.
]]>
Error rates for Dx testing
<![CDATA[
=======================Charles Metz, 27 Feb 1997==========ssc
From: Charles Metz <c-metz@uchicago.edu>
Subject: Re: Q: Error Rates for Diagnostic Testing
Message-ID: <3315D224.251C@uchicago.edu>
Steve Lutes wrote:
> I'm working on a method of identifying people with
> vestibular disorders using discriminant analysis of test
> data. What I need to know is the following:
>
> (1) What is an acceptable error rate for a single
> test in a battery of tests?
>
> (2) Are there standards for test error rates?
>
> (3) Who/what sets these standards?
>
> Anything you can give (references, Web sites, books.
> etc.) will be helpful. If you reply by E-mail, please
> post your response to the newsgroup as well. I believe
> this is a matter of general interest.
From a cost/benefit or expected-utility viewpoint, the optimal
compromise between the false-negative and false-positive error rates of
a diagnostic test depends upon the prevalence of the disease in question
and upon the clinical consequences of the four kinds of correct
(true-positive and true-negative) and incorrect (false-negative and
false-positive) test results in a particular diagnostic setting. See,
for example:
McNeil BJ, Keeler E, Adelstein SJ. Primer on certain elements of
medical decision making. New England Journal of Medicine 1975; 293:
211-215.
Metz CE. Basic principles of ROC analysis. Seminars in Nuclear
Medicine 1978; 8: 283-298.
Phelps CE, Mushlin AI. Focusing technology assessment. Medical
Decision Making 1988; 8: 279-289.
Sainfort F. Evaluation of medical technologies: a generalized ROC
analysis. Medical Decision Making 1991; 11: 208-220.
Halpern EJ, Alpert M, Krieger AM, Metz CE, Maidment AD. Comparisons of
ROC curves on the basis of optimal operating points. Academic Radiology
1996; 3: 245-253.
* * * * * * * * * * * * * * * * * * * * * * * * * * * * *
]]> Document by Rich Ulrich. E-mail to wpilib+@pitt.edu
FAQ top.
Ulrich home page.
Ulrich FAQ.
http://www.pitt.edu/~wpilib/stats99.html