I am running a simulation with subproblems = 1000. One of the
replicates generates the following error:
PROBLEM NO.: 1 SUBPROBLEM NO.: 160
0PRED EXIT CODE = 1
0INDIVIDUAL NO. 79 ID=0.13034000E+04 (WITHIN-INDIVIDUAL) DATA REC
NO. 2
THETA=
7.26E+00 1.06E+01 1.64E+00 3.99E+00 1.21E+01 5.60E+00
1.95E+00 -1.81E+00 4.31E-01 -1.39E+00
5.88E+00 1.93E+00 -4.94E-01 -4.73E-01
K32, OR K42 IS TOO CLOSE TO AN EIGENVALUE
MESSAGE ISSUED FROM SIMULATION STEP
When NONMEM encounters this the whole simulation stops. Is there anyway
to tell NONMEM to just keep going and ignore the replicates with this
error?
Thanks,
pete
Peter L. Bonate, PhD, FCP
Genzyme Corporation
Senior Director, Pharmacokinetics
4545 Horizon Hill Blvd
San Antonio, TX 78229 USA
[EMAIL PROTECTED]
phone: 210-949-8662
fax: 210-949-8219
blackberry cell: 210-315-2713
Pred Exit Error =1 with $SIM
Pete,
There are 2 things going on here. First the OMEGA you are using to simulate the
PK parameters is producing some unrealistic combinations for a true 2
compartment model and the analytical solution is having difficulty. The second
thing is that when NONMEM detects this problem it crashes and does not continue
as one would by hope by going to the next simulation subproblem.
This is a well known bug/feature of NONMEM V. It was on the list of things to
be fixed in NONMEM VI but I have not tried to see if it has gone away. Which
NONMEM version did you use?
One way to get around this problem is to specify the model as a set of
differential equations instead of relying on the closed form solution in
ADVAN4. NONMEM will blindly attempt to solve the DEs and won't crash even with
implausible parameters for the 2 cpt model. The simulation for that subproblem
may not be very realistic but at least NONMEM will continue to try the
remaining subproblems.
When the crash is caused by an error in the estimation step triggered by
strange values generated during simulation it can be worked around by doing the
simulation and estimation separately. I do this by using the nmgosim command
which is part of Wings for NONMEM ( http://wfn.sourceforge.net). This command
will use NONMEM to simulate say 1000 data sets then use NONMEM to use each data
set for estimation. If a simulated data set causes NONMEM to crash then nmgosim
logs this and continues with the next subproblem.
Nick
> "Bonate, Peter" wrote:
>
> I am running a simulation with subproblems = 1000. One of the replicates
> generates the following error:
>
> PROBLEM NO.: 1 SUBPROBLEM NO.: 160
>
> 0PRED EXIT CODE = 1
> 0INDIVIDUAL NO. 79 ID=0.13034000E+04 (WITHIN-INDIVIDUAL) DATA REC NO.
> 2
> THETA=
> 7.26E+00 1.06E+01 1.64E+00 3.99E+00 1.21E+01 5.60E+00 1.95E+00
> -1.81E+00 4.31E-01 -1.39E+00
> 5.88E+00 1.93E+00 -4.94E-01 -4.73E-01
> K32, OR K42 IS TOO CLOSE TO AN EIGENVALUE
> MESSAGE ISSUED FROM SIMULATION STEP
>
> When NONMEM encounters this the whole simulation stops. Is there anyway to
> tell NONMEM to just keep going and ignore the replicates with this error?
>
> Thanks,
>
> pete
>
> Peter L. Bonate, PhD, FCP
> Genzyme Corporation
> Senior Director, Pharmacokinetics
> 4545 Horizon Hill Blvd
> San Antonio, TX 78229 USA
> [EMAIL PROTECTED]
> phone: 210-949-8662
> fax: 210-949-8219
> blackberry cell: 210-315-2713
>
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New Zealand
email:[EMAIL PROTECTED] tel:+64(9)373-7599x86730 fax:373-7556
http://www.health.auckland.ac.nz/pharmacology/staff/nholford/