Christian
There is at least one parameter you cannot estimate in this code, this is V2
(volume of distribution of the metabolite)
This parameter is not identifiable because the fraction of parent drug that is
metabolized is unknown and the volume of distribution simply relates the
concentration to the amount in the body.
A simple usual solution is to fix V2 to a constant value near the parent V1
value.
Also
- it is unlikely you will be able to estimate OMEGA values for all of the
parameters
- the initial SIGMA values are generally in the 0.3- 1 range for the
proportional component
Saik
----- Original Message -----
From: christian woloch
To: nmusers@globomaxnm.com
Sent: Wednesday, July 22, 2009 11:45 AM
Subject: [NMusers] Dynamic enzymatic model
Dear all,
I try to model simultaneously a parent drug and its metabolite using a 2
compartments dynamic enzymatic model
(using the general enzymatic reaction E+S-->(ES)-->E+P).
Here is the control stream :
$PROB TEST MODELISATION NLIN 2CPMT
$INPUT ID TIME DV AMT MDV CMT
$DATA DATA131.csv IGNORE=#
$SUBROUTINE ADVAN8 TOL4 ; Modèle Non linéaire 2 cpts
$MODEL
COMP= (CENTRAL,DEFDOSE) ; cpt 5FU
COMP= (CENTRAL,NODOSE) ; cpt 5FDHU
COMP= (CENTRAL,NODOSE) ; enzyme **(1)**
$PK
V1 =THETA(1)*EXP(ETA(1)) ; Vd 5FU
V2 =THETA(2)*EXP(ETA(2)) ; Vd 5FDHU
CL10 =THETA(3)*EXP(ETA(3)) ; Cl d'élimination 5FU
CL20 =THETA(4)*EXP(ETA(4)) ; Cl d'élimination 5FDHU
KE1 =THETA(5)*EXP(ETA(5)) ; Cstte de métabolisation
KE2 =THETA(6)*EXP(ETA(6)) ; Cstte de métabolisation
ET =THETA(7)*EXP(ETA(7)) ; total enzyme
K10=CL10/V1
K20=CL20/V2
S1=V1
S2=V2
$DES
B= ET-A(3)
DADT(1)= -K10 *A(1)-KE1*B*A(1)
DADT(2)= KE2 * A(3) - K20 * A(2)
DADT(3)= KE1 * B*A(1)-KE2*A(3)
$ERROR
IPRED=F
IF (CMT.EQ.1) THEN
Y=F*(1+EPS(1))+ EPS(2)
ELSE
Y=F*(1+EPS(3))
ENDIF
IRES=DV-IPRED
$THETA (0,10,20) (0,60,100) (0,1,10) (0,1,10) (0,1,10) (0,1,10) (0,1,10)
$OMEGA (1) (1) (1) (1) (1) (1) (1)
$SIGMA (0.01) (0.01) (0.01)
$EST METHOD=1 SIG=3 MAXEVAL=9999
**(1)** : this compartment was added because of the enzyme own equation
(DADT(3)).
With FO, I'm trying to estimate the parameters
But with FOCE, the run stop immediately!
ERROR IN NCONTR WITH INDIVIDUAL 1 ID=0.10000000E+01
NUMERICAL HESSIAN OF OBJ. FUNC. FOR COMPUTING CONDITIONAL ESTIMATE
IS NON POSITIVE DEFINITE
1)Are there any problems in the control stream?
2)The model is overparametrized? Should I have to fixed parameters?
(I have over one hundred couple of kinetics with 5-7 observations for each)
I would really appreciate aany advice, thanks.
Christian
