$Error with log-transformation
Hi, NM users:
I am a new NONMEM user and wish to get help from NM experts.
I have a data from rats with concentration measured in both plasma and
cerebrospinal fluid . Compared to plasma concentration, the concentration
in brain is relatively small (up to 30 times difference). Due to the big
range of my data, I log-transformed my data. Below is the code I used. I
was able to get NM to run, but obvious bias existed in goodness-of-fit
plot. I worried that there are mistakes in my code. Could anyone take a
look of my code especially the $ERROR code to see what is wrong?
Also, I saw in an earlier discussion on how to get additive error with
log-transformed data, but that was for simple models. Can anybody give some
insights on how to do it with more complex data such as the data I have?
Thank you in advance for your time.
Siwei
$SUBROUTINES ADVAN6 TOL=3
$MODEL
NCOMP=3
COMP=(COMP1) ;Central compartment
COMP=(COMP2) ;Peripheral compartment
COMP=(COMP3) ;Brain compartment
$PK
TVCL=THETA(1)
CL=TVCL*EXP(ETA(1))
TVV1=THETA(2)
V1=TVV1*EXP(ETA(2))
TVQ1=THETA(3)
Q1=TVQ1*EXP(ETA(3))
TVV2=THETA(4)
V2=TVV2*EXP(ETA(4))
TVKEQ=THETA(5)
KEQ=TVKEQ*EXP(ETA(5)) ; Equilibration rate constant trough BBB
TVPC=THETA(6)
PC=TVPC*EXP(ETA(6)) ; Partition coefficient at BBB
K10=CL/V1
K12=Q1/V1
K21=Q1/V2
S1=MV1
$DES
DADT(1)=-K10*A(1)-K12*A(1)+K21*A(2)-KEQ*(A(1)*PC-A(3))
DADT(2)=K12*A(1)-K21*A(2)
DADT(3)=KEQ*(A(1)*PC-A(3))
$ERROR
IF(AMT.NE.0) THEN
IPRE=LOG(1)
ELSE
IPRE=LOG(F)
ENDIF
CM=0
IF (CMT.LE.2) CM=1
CF=0
IF (CMT.GE.3) CF=1
YM = IPRE+ERR(1) ; Plasma
YF = IPRE+ERR(2) ; Brain
Y=CM*YM+CF*YF
$EST METHOD=1 PRINT=1 MAX=9999 SIG=3
$THETA
$OMEGA
$SIGMA
Here is how the data look like:
ID TIME AMT DV_log MDV CMT
1 0 180 0 . 1 1
1 5 0 0 0.825 0 1
1 5 0 0 -0.127 0 3
1 15 0 0 0.954 0 1
1 15 0 0 -0.011 0 3
1 25 0 0 0.937 0 1
1 25 0 0 0.137 0 3
1 60 0 0 1.015 0 1
1 60 0 0 0.188 0 3
1 100 0 0 0.567 0 1
1 100 0 0 -0.311 0 3
1 150 0 0 0.378 0 1
1 150 0 0 -0.493 0 3
1 180 0 0 -0.159 0 1
1 180 0 0 -0.74 0 3