Re: [Fwd: CLIN PHAR STAT: Mixed Vs Fixed]

Date: Tue, 08 Aug 2000 07:34:04 +1200 From: Nick Holford <n.holford@auckland.ac.nz> Subject: Re: [Fwd: CLIN PHAR STAT: Mixed Vs Fixed] Leonid, Thanks for this suggestion but it doesn't seem to me that THETA(10) is identifiable. I think you need some additional covariate to distinguish the components of variance for CL when using this kind of approach. I was wondering if the undocumented Bayesian PRIOR feature of NONMEM would be appropriate here. By putting a prior on the variance of CL this is one way of putting "a random effect on the variances themselves". (Mats, Diane -- are you there? can you comment?). Nick "Gibiansky, Leonid" wrote: > > > > > Is it possible to put "a random effect on the variances themselves" in > NONMEM? > > > > One can try > > VAR=THETA(10)*EXP(ETA(10)) ;assuming that the other 9 THETAs and ETAs are > used elsewhere > > ET1=VAR*ETA(1) > CL=THETA(1)*EXP(ET1) > > $OMEGA > 1 FIXED ; variance of eta1 > ... > 0.5 ; variance of random effect for variance of eta1 > > I am not sure about convergence and whether you gain anything... > Leonid > -- Nick Holford, Dept Pharmacology & Clinical Pharmacology University of Auckland, Private Bag 92019, Auckland, New Zealand email:n.holford@auckland.ac.nz tel:+64(9)373-7599x6730 fax:373-7556 http://www.phm.auckland.ac.nz/Staff/NHolford/nholford.htm