CMT is important for dosing: it directs the dose to the appropriate
compartment (equation number).
It is a common misconception that CMT has to be related to observations.
It is relevant for observations only if F notation is used (and
therefore it is better to avoid using this notation).
For observations, one can set any compartment number (within the range
from 1 up to the number of equations in the model). Then, a separate
variable TYPE can be used, for example:
Y=A(2) + error ; for parent (TYPE=1)
IF(TYPE.EQ.2) Y = A(6)+A(7)+A(8) + error ; for metabolite (TYPE=2)
In this case, compartment number found on observation records can be
ignored. It will direct NONMEM to set F to the A(CMT) value, but if F is
not used, then it has no effect on the model or results.
Leonid
On 4/24/2017 11:26 AM, Fanny Gallais wrote:
Dear NMusers,
I'm having trouble writing the code for my model. I have IV and IP
administrations data, that I would like to model simultaneously. We
found out that IP was best modeled by a first order absorption, using an
input compartment (DADT(1)=-KA*A(1)). At first, we used a CMT column in
the dataset to indicate for each dosing event if it is IV or IP. But
then, as we made the model more complicated, we realized that we
couldn't use a CMT column. We study the parent compound, as well as its
metabolite at the same time but the problem is we cannot put a
compartment number for the metabolite's observations. This is because,
given the model structure, the prediction for the metabolite
concentrations is a sum of 3 concentrations in 3 different compartments
(e.g. IPRED=A(6)+A(7)+A(8)). So we think that putting a compartment
number for the observation would be confusing for NONMEM. What do you
think? Do you see how we could model IV and IP simultaneously, without
using the CMT column ? or any other solution?
Thank you for your help
Fanny Gallais
