NONMEM Users Network Archive

Hosted by Cognigen

Re: Problem with changing parameter estimates

From: Nitin Mehrotra <nitin_utmem>
Date: Wed, 14 Mar 2007 10:48:20 -0700 (PDT)

Hi Paul,
   
  Its hard to tell much about the problem with the information you have provided. However I could think of one likely explanation (if it fits into your case). It is known that NONMEM is sensitive to initial parameter estimates and if you have overparameterized model such that there is insufficient data to describe the relevant parameters, things become worse.
   
  In nutshell, lot of it depends upon what kind of data you have and which parameter you want to estimate. For instance, if your data is not informative enough (with regard to density and/or time points of samples) such that it is difficult to estimate a particular parameter, even small change in initial parameter estimates may lead to convergence problem or infinite looping, or physiologically unreasonable parameter estimate.
   
  Now the fact that you had 13 successful runs and you are using NONMEM VI (which is likely to be less sensitive to initial parameter estimates), it would be nice if you provide a bit more details of your problem (for eg., which parameter estimate was changed and by what magnitude, what kind of data and structural model you are working with etc.) to better understand the issue.
   
  Would like to hear views from other members of the group on this.
   
  Hope it Helps
   
  Regards
   
  Nitin

Paul Westwood <pwestwood02
  Hello all,

I am running nonmem 6, using wings for nonmem. I have ran into a slight
problem. Whilst gradually building up a model, an initial value for a
parameter was changed. This caused many executable files to be created in
the control file folder, and the run to go on infinitely as if in an
infinite loop. Anyone any ideas on what happened? The change in the initial
parameter value was the only difference in this particular control file to
the others, and it was the 14th run at that point.

As always, much obliged to anyone with any replies.
Best Regards,
Paul Westwood.



Nitin Mehrotra, Ph.D
Post Doctoral Research Fellow
Department of Pharmaceutical Sciences
University of Tennessee Health Science Center
Memphis, Tennessee, USA-38105
nmehrotr

 
---------------------------------
8:00? 8:25? 8:40? Find a flick in no time
 with theYahoo! Search movie showtime shortcut.
Received on Wed Mar 14 2007 - 13:48:20 EDT

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to: nmusers-request@iconplc.com.

Once subscribed, you may contribute to the discussion by emailing: nmusers@globomaxnm.com.