From: "Bonate, Peter" - pbonate@ilexonc.com
Subject: [NMusers] Modeling average data
Date: 2/13/2004 9:19 AM

Hi, everyone.

I am having trouble how to model a particular problem.  It seems quite
simple conceptually and is a very simple matter to program in Proc Mixed in
SAS, but I can't figure out how to do this in NONMEM.  I have repeated measures
data taken essentially at steady-state at trough.  Essentially every subject is
a flat line that varies around some population average.  There is no absorption
data and no elimination data.  What I want to model is the following:

DV  = Theta(1) + Eta(1) + EPS(1)

So that I get an estimate of the population average, the between-subject
variability and residual variability.

Can someone help me out here and provide me with what the control stream should look like.

Thanks,

Pete

Peter L. Bonate, PhD, FCP
Director, Pharmacokinetics
ILEX Oncology
4545 Horizon Hill Blvd
San Antonio, TX  78229
phone: 210-949-8662
fax: 210-949-8219
email: pbonate@ilexonc.com

_______________________________________________________

From: LSEN (Lars Erichsen) 
Subject: RE: [NMusers] Modeling average data 
Date: 2/13/2004 10:18 AM

You can fit this model by using $PRED instead of $PK and $ERROR:

$PRED

Y=THETA(1)+ETA(1)+EPS(1)
 
 
 
BR
 
Lars
_______________________________________________________

From: phil.lowe@pharma.novartis.com
Subject: RE: [NMusers] Modeling average data
Date: 2/13/2004 11:55 AM


Dear Pete.

Similar to a previous reply, something I used a couple of years back to monitor
an approximately flat set of placebo cell counts (before making a PK/PD connection).
You can have great fun getting into covariate space even with this simple a model
of a marker.

$PRED
   CELL=THETA(1)*EXP(ETA(1))
   F=CELL
   Y=F+ERR(1)

Best regards, Phil. 
_______________________________________________________