Dear NONMEM users,
I am trying to build a model to describe the drug toxicity on neuropathy.
Since I am having dose and toxicity data, the K-PD model was applied in our
model. The distribution of baseline is heavily left skewed. So I am also
trying included the Box-cox transformation to get a accurate estimate of
baseline. I followed Prof. Karlsson's paper. (Petersson KJ, Hanze E, Savic
RM, Karlsson MO. Semiparametric distributions with estimated shape
parameters. Pharm Res. 2009;26(9):2174-85.)
My problem is that NONMEM never gave an estimate of BXPAR (I got the
initial value of BXPAR in the output file).
The control stream and a sample of data follows.
Many thanks in advance.
Kehua
*
control stream*:
$SUBS ADVAN6 TOL=6
$MODEL
NCOMP=2
COMP=(DOSE)
COMP=(OBS)
$PK
KIN=THETA(1)*EXP(ETA(1))
BXPAR=THETA(2)
PHI=EXP(ETA(2))
ETATR=(PHI**BXPAR-1)/BXPAR
BASELINE=THETA(3)+(PHI**BXPAR-1)/BXPAR
KDE=THETA(4)*EXP(ETA(3))
EDK50=THETA(5)*EXP(ETA(4))
EMAX=THETA(6)*EXP(ETA(5))
KOUT=KIN/(BASELINE)
F2=BASELINE
S2=1
(I also tried to estimate KIN and KOUT. BASELINE=KIN/KOUT. the BOX-cox
transformation was added on KOUT. But did not get any estimate on BXPAR
either.)
$DES
DADT(1)=-KDE*A(1)
VIR=A(1)*KDE
IRG=VIR
COEF=1-(IRG*EMAX/(EDK50+IRG))
DADT(2)=KIN-KOUT*COEF*A(2)
$ERROR
IPRED=F
IRES=DV-IPRED
IF (F.EQ.0) FX=1
W=F+FX
IWRES=IRES/W
Y = F + ERR(1)+F*ERR(2)
*data*:
PATID DAY amt fgsum4 addl II CMT 1 0
3
2 1 1 150.3704
3 7 1 1 29 155.5556
3 7 1 1 54
6
2 1 57 155.5556
3 7 1 1 85 155.5556
3 7 1 1 108
13
2 1 113 155.5556
3 7 1 1 135
12
2 1 141 155.5556
3 7 1 1 162
9
2 1 169 150
3 7 1 1 190
14
2 1 197 155.5556
3 7 1 1 217
16
2 1 225 73.7234
5 7 1 1 267 139.8674
0 0 1 2 0
0
2 2 1 113.5556
3 7 1 2 27
3
2 2 29 116.6667
3 7 1 2 54
0
2 2 57 113.8148
3 7 1 2 81
2
2 2 85 108
3 7 1 2 109
2
2 2 113 112.9633
0 0 1 3 0
2
2 3 1 126
3 7 1 3 27
4
2 3 29 126
3 7 1 3 54
4
2 3 57 126
3 7 1 3 81
5
2 3 85 126
3 7 1
question in Box-Cox Transformations in K-PD model
From: Ribbing, Jakob
Sent: 12 August 2013 23:45
To: 'kehua wu'; nmusers
Cc: Ribbing, Jakob
Subject: RE: [NMusers] Fwd: question in Box-Cox Transformations in K-PD model
Hi Kehua,
When you say that you did not get the estimate of TH2 in the output file, but
you got the initial estimate. Did you mean that the model failed termination or
that it minimised, but that TH2 did not move from its final estimate? I think
we need more information from the control stream.
Also, the part of the control stream that you shared did not include initial
estimates. Did you start with a negative initial estimate for TH2? I would add
an upper boundary at zero as well.
For alternative statistical models with FOCE (or FOCEI, where appropriate) I
have seen a couple cases were likelihood profiling indicates that there is
information on the parameter, but where the estimate did not move from its
initial estimate (to describe shape of individual parameter distribution or
residual-error distribution) - These models were often complex or at least over
parameterized in some regards - In your case; do you have enough information to
estimate etas on KIN, KDE, EKD50 and EMAX, or are some of these omegas fixed?
In addition, estimating EKD50 (theta and omega) often is very helpful to avoid
correlation between the estimates (which is why this parameterisation was
suggested in the first place). However, there are also cases where this
parameterisation induces a correlation between the estimates and in that case
estimating EA50 may be more useful.
For the limited number of cases where I have tested different "semi-parametric"
distributions for individual parameters; Box-cox transformation I have found to
be one of the more stable alternatives.
Best regards
Jakob