st: RE: calculation of area under curve with Stata

["Seed, Paul" <paul.seed@kcl.ac.uk>]

To avoid further confusion, can I explain that the term "area under the curve" 
Has two completely different meanings in biostatistics.

The first is the area under a dose concentration curve;
 or any other measurement taken repeatedly over time.
It is mainly used in pharmacokinetics, and it is clearly that which 
Elmir Omerovic is interested in.  The -pk- suite of commands are 
designed to handle this:
    The pk commands are

        pkexamine        Calculate pharmacokinetic measures
        pksumm           Summarize pharmacokinetic data
        pkshape          Reshape (pharmacokinetic) Latin-square data
        pkcross          Analyze crossover experiments
        pkequiv          Perform bioequivalence tests
        pkcollapse       Generate pharmacokinetic measurement dataset


The second is the area under a ROC curve. 
It is used as a convenient non-parametric measure of the performance of a 
Diagnostic  or predictive test.  Appropriate commands include 
-lroc-, which Joe Canner referred to,  -roctab- roccomp- and -rocgold-.

To avoid confusion with the other AUC, it is sometimes referred to as AUROC; 
or as a' (a-prime), or as the c-statistic.  Although the last invites confusion 
with Harrell's C, which is another story .

ROC, by the way, stands for  receiver operating characteristic; and the 
term originates in signal detection theory. 

Best wishes,


Paul T Seed, Women's Health, KCL





> 
> Date: Mon, 14 Oct 2013 13:12:10 +0000
> From: Joe Canner <jcanner1@jhmi.edu>
> Subject: st: RE: calculation of area under curve with stata
> 
> Elmir,
> 
> One possibility, in addition to that already suggested is this:
> 
> . logit outcome predictors
> . lroc
> 
> You can also compare AUCs using:
> 
> . logit outcome predictor1
> . lroc
> . predict xb_predictor1 if e(sample), xb
> . logit outcome predictor2
> . lroc
> . predict xb_predictor2 if e(sample), xb
> . roccomp outcome xb_predictor1 xb_predictor2, graph summary
> 
> You can compare more than two AUCs as desired; just repeat the -logit-, -lroc-, -
> predict-  sequence.
> 
> Regards,
> Joe Canner
> Johns Hopkins University School of Medicine
> 
> - -----Original Message-----
> From: owner-statalist@hsphsun2.harvard.edu [mailto:owner-
> statalist@hsphsun2.harvard.edu] On Behalf Of elmir
> Sent: Monday, October 14, 2013 4:38 AM
> To: statalist@hsphsun2.harvard.edu
> Subject: st: calculation of area under curve with stata
> 
> Dear All,
> 
> I am analyzing the blood pressure (BP) levels from the three different groups of
> animals. BP was measured repeatedly in every animal at 10 different time points
> within 90 minutes from the start of the experiment.
> The animals were exposed to 3 different drugs (each group n=8).
> 
> I understand that one way to compare the effect of the drugs on blood pressure
> within the whole experimental period would be to calculate the area under the
> curve and to use this value in a linear mixed model.
> 
> I would be very grateful if someone could help me with step-by-step instructions
> how to calculate AUC in stata.
> 
> Kind regards
> 
> Elmir Omerovic MD, PhD
> Sahlgrenska University Hospital
> Gothenburg, Sweden


*
*   For searches and help try:
*   http://www.stata.com/help.cgi?search
*   http://www.stata.com/support/faqs/resources/statalist-faq/
*   http://www.ats.ucla.edu/stat/stata/