NONMEM Users Network Archive

Hosted by Cognigen

RE: Model building

From: Jurgen Bulitta <jbulitta>
Date: Thu, 3 Dec 2009 21:38:39 -0500

Dear Neil,

Two questions:
1) Could you please give us some information on the number of thetas and om=
number of cellular sub-populations with different drug susceptibility, and =
rates of growth and killing (e.g. n Log10 killing per hour or day) in your =

2) Could you please let us know what your observed variables are? (i.e. are=
 you just
measuring total cell concentration or are you specifically measuring resist=
ant cells)

Especially for the FOCE method, too many omegas on initial conditions of re=
subpopulation(s) that only show up later after the susceptible cells have b=
een killed
may cause the FOCE method to not finish successfully. I am not overly worri=
ed about
this and tend to go for the better curve fits. Two solutions would be to
1) use MC-PEM instead of FOCE or 2) to reduce the number of OMEGAs.
At least for in vitro studies, the run to run variability may be small.

Best wishes

From: owner-nmusers
 Behalf Of Indranil Bhattacharya
Sent: Thursday, December 03, 2009 10:41 AM
To: nmusers
Subject: [NMusers] Model building

Hi, I am in the process of developing a PK/PD model and have a naive questi=
on regarding model building.
I currently do not have all the PD data and more data would be available in=
 the future. For the current data set, I have tried models say A to D.
Now model A (cell kill) and B (cell kill +transduction) converges using FO=
CE (no CV% but I am willing to live with that) but from the RES and WRES pl=
ots we can clearly see that there is some bias. The fit is OK but not great=
. The ofv values are around 400.
Now models B (cell cycle specific kill), C (cell + precursor cell kill +tra=
nsduction) and D (cell cycle specific + indirect response model) do not con=
verge using FO or FOCE methods but when I look at the fits from the termina=
ted runs, the fits are much better than those obtained with Model A, and th=
ere seems very little bias. Also the ofv values are between 160-250.

So my question is whether the fits, RES, WRES plots and the ofv values have=
 meaning even when the minimization terminates.



Indranil Bhattacharya

Received on Thu Dec 03 2009 - 21:38:39 EST

The NONMEM Users Network is maintained by ICON plc. Requests to subscribe to the network should be sent to:

Once subscribed, you may contribute to the discussion by emailing: