```
From: "LSEN (Lars Erichsen)" lsen@novonordisk.com
Subject: [NMusers] \$DES and T
Date: Wed, February 9, 2005 7:29 am

Dear Nonmem users,

I have a question concerning the use of \$DES and the continous time
variable T:
I want to compute a concentration at any time point T and use it as
input to a PD model (see code below).

I get the warning (48): DES defined items are computed only when event
time increases...
Does this mean that CONC is only recomputed at observed time points (I
do not have any dosing events)? What happens between these time points?
When I print out to a table the first CONC value is 0 - even if TIME>0.
Results change if I add new data records to the data set with MDV=1 and
new TIME points.

Thanks in advance,

Lars.

\$SUBROUTINES ADVAN6 TOL 5

\$MODEL
COMP=(EFFECT)

\$PK

TVKIN=THETA(1)
KIN=EXP(TVKIN)
TVKOUT=EXP(THETA(2))
KOUT=TVKOUT*EXP(ETA(1))

TVTH=THETA(3)
TH=EXP(TVTH)

K=6.41
V=0.127
KA=1.35

\$ERROR

IPRE=A(1)

Y=1+IPRE+ERR(1);

\$DES

CONC=DOSE*KA/(K-KA)/V*(EXP(-KA*T)-EXP(-K*T))

KIN1=0
IF(CONC.GT.TH) KIN1=1

DADT(1)=KIN*KIN1-KOUT*A(1)
_______________________________________________________

From: "Sam Liao" sliao@pharmaxresearch.com
Subject:  Re: [NMusers] \$DES block and T
Date: Wed, February 9, 2005 6:38 pm

Hi Lars:

You do need to have the MDV record with time equal to zero in order to
solve the differential equation correctly.  You've done the right thing
paying attention to the warning msg from nmv.  It is trying to tell you
that the ODE solver start from a time point which is not zero.  That is
why the results are different.

Also on a different issue about your PKPD model, instead of using a
threshold concentration in your model, have you try a Emax model (Hill's
equation) with large gamma coefficient which can also give you a steep
function to mimic the threshold model you have.

Best regards,
Sam Liao
_______________________________________________________

```