Re: [NMusers] DEFDOSE/DEFOBS Required for Many Compartments?

From: Alison Boeckmann <alisonboeckmann_at_fastmail.fm>
Date: Mon, 01 May 2017 12:28:16 -0700

Bill,
Here's how it looks to me right now.

I put together a little test case:
$INPUT      ID DOSE=AMT TIME CP=DV WT CMT EVID 
$MODEL 
COMP=(One,INITIALOFF,NODOSE)
COMP=(PKCENT)

I ran this with a data set in which CMT=2 for both dose and observation
records.
PREDPP gives me the same error message that you reported:
DATA REC    1: COMPARTMENT ASSOCIATED WITH THE PREDICTION IS OFF

NMTRAN makes the assignment of "Default for observations"
inappropriately:

COMPT. NO.   FUNCTION   INITIAL    ON/OFF    =
  DOSE      DEFAULT  
DEFAULT
                    =
   STATUS     ALLOWED    ALLOWED   =
 FOR DOSE   FOR OBS.
    1         ONE        =
  OFF        YES        NO =
      NO         YES
    2         PKCENT       ON =
        YES        YES   =
   YES        NO
    3         OUTPUT       OFF=
        YES        NO    =
     NO         NO

The PCMT data item is needed with dose records to identify the
prediction compartment associated with the dose record, and, when PCMT
is not
defined, this is defaulting to "Default for observations," which is
compartment 1,
which is initially off.

The DEFAULT FOR OBS compartment should have the attribute "INITIAL ON".

Your fix is a good one (specify both DEFDOSE and DEFOBS).
Another fix is to supply the PCMT data item in the data set, and (in my
case) set PCMT=2

This is something that can be fixed with a future version of NONMEM (its
too late for 7.4).

-- Alison

On Mon, May 1, 2017, at 09:33 AM, Bill Denney wrote:
> Hi,


> I have a model with many compartments (42 different compartments
      are defined).  It is currently setup this way to future-proof the
      dataset in case we want to combine analyses of many drugs (about
      20 different drugs with central/peripheral compartments) and
      several endpoints.


> In the current model, I'm only fitting the PK of one drug, so
      only two of the compartments are enabled:  compartments 39 and
      42.  After the IGNORE statement, the data is relatively small wi=
th
      400 records between 16 subjects, and all the data are in
      compartment 39 (the central compartment for the current drug).


> CMT, EVID, and MDV (along with everything else) is correctly
      specified.  When I run the model with CMT 39 defined by:


> COMP=(PKCENT)  ; 39 central PK compartment (ug/mL)


> the model fails with the error "COMPARTMENT ASSOCIATED WITH THE
      PREDICTION IS OFF" noted for all the dosing rows, but when I
      specify that compartment 39 is the default observation and dose
      compartment (due to IV dosing), the model works.


> COMP=(PKCENT,DEFOBS,DEFDOSE)  ; 39 central PK compartment (ug/mL)


> I thought that when CMT and EVID are specified, DEFOBS and
      DEFDOSE are not important.


> Does anyone know why adding DEFOBS and DEFDOSE fixed my model?


> Thanks,


> Bill


> Human
          Predictions Logo[http://www.humanpredictions.com/]*William S.
          Denney, PhD*
> Chief Scientist, Human Predictions LLC
> +1-617-899-8123
> wdenney_at_humanpredictions.com
> This e-mail communication is
          confidential and is intended only for the individual(s) or
          entity named above and others who have been specifically
          authorized to receive it. If you are not the intended
          recipient, please do not read, copy, use or disclose the
          contents of this communication to others. Please notify the
          sender that you have received this e-mail in error by replying
          to the e-mail or by calling +1-617-899-8123. Please then
          delete the e-mail and any copies of it. Thank you.



--
  Alison Boeckmann
  alisonboeckmann_at_fastmail.fm


Received on Mon May 01 2017 - 15:28:16 EDT

This archive was generated by hypermail 2.3.0 : Fri Sep 27 2019 - 16:54:52 EDT