NONMEM Users Network Archive

Hosted by Cognigen

RE: Significant Figures Continuation

From: Farrell, Colm <Colm.Farrell>
Date: Wed, 20 Jan 2010 17:15:12 -0000

Dear Indranil,

According to Bob Bauer's explanation of the search algorithm process in
the introduction to NM7, he proposes the following general rules.

Set SIGL, NSIG, and TOL such that:

SIGL<=TOL

NSIG<=SIGL/3

The SIGL is a new variable in NM7 that the user can now specify - if you
are still with NM VI, then the internal SIGL is set to 10.

Kind Regards,
Colm

Colm Farrell

PKPDM&S

ICON Development Solutions


________________________________

From: owner-nmusers
On Behalf Of Elassaiss - Schaap, J. (Jeroen)
Sent: 20 January 2010 16:36
To: Indranil Bhattacharya; nmusers
Subject: RE: [NMusers] Significant Figures Continuation


Dear Indranil,
 
Nick Holford has posted his observation on the use of SIGDIG in Nonmem
v6 to this list. You can find the results at his site:
http://www.fmhs.auckland.ac.nz/sms/pharmacology/holford/projects/sigdig.
aspx
 
If you use ADVAN 6 to 9, you may want to set TOL to a number really
higher (at least +3) than SIG to get consistent results, as mentioned
during the NM7 introduction by Bob bauer.
 
So you may interpret SIG as a convergence criterium and your results as
some instability. Try to play around with ADVAN 6-9, SIG, TOL and ML
approach (FOCE - LAPL) to search for stability in outcomes. Or better,
refine your model to stabilize it. A third, easier, solution would be to
do a bootstrap and check for stability / %CV outcomes that way.
 
Best regards,
Jeroen
 

Jeroen Elassaiss-Schaap
Modeling & Simulation Expert
Pharmacokinetics, Pharmacodynamics & Pharmacometrics (P3)
T: +31 (0)412 66 9320
F: +31 (0)412 66 2506
jeroen.elassaiss

 

________________________________

From: owner-nmusers
On Behalf Of Indranil Bhattacharya
Sent: Wednesday, 20 January, 2010 15:21
To: nmusers
Subject: [NMusers] Significant Figures Continuation


First, of all sorry about the last email, it somehow got sent before I
could complete it.

 
Hi, while running a KOKA model I came across an observation and would
greatly appreciate it if someone helps me understand it.
When using a SIG=3, the model converges, and provides lower and upper =
CI
bounds. All the CI bounds are positive except Q and the CV% of all
parameters are within 20 % (except Q).
This suggests (correct me if I am wrong) that the data might not be rich
enough for the model to calculate a precise value for Q so it gives a
wider range?
Now when I change the SIG=3 to SIG=4, the model converges but the =
CV%
are very high and more importantly I get negative lower bounds for all
the parameters. What does this suggest, that the model cannot get
precise estimates now at the level of significance (of numbers)
requested? Please advise.
 
Regards
 
Neil

--
Indranil Bhattacharya





--
Indranil Bhattacharya



________________________________

This message and any attachments are solely for the intended recipient.
If you are not the intended recipient, disclosure, copying, use or
distribution of the information included in this message is prohibited
--- Please immediately and permanently delete.
________________________________


ICON plc made the following annotations.
-------------------------------------------------------------------------=
-----
This e-mail transmission may contain confidential or legally privileged i=
nformation
that is intended only for the individual or entity named in the e-mail ad=
dress. If you
are not the intended recipient, you are hereby notified that any disclosu=
re, copying,
distribution, or reliance upon the contents of this e-mail is strictly pr=
ohibited. If
you have received this e-mail transmission in error, please reply to the =
sender, so that
ICON plc can arrange for proper delivery, and then please delete the mess=
age.
Thank You,
ICON plc
South County Business Park
Leopardstown
Dublin 18
Ireland
Registered number: 145835

Received on Wed Jan 20 2010 - 12:15:12 EST

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.