Back to the Top
I have encountered an issue that when I ran bootstrap using WFN, I get
a larger eta than the FOCE estimates.
I would appreciate if someone can comment on the code below. I have
attached the code here:
$PROB 1;
$DATA data.csv IGNORE=C
$INPUT C STUD PHAS DUR AMT RATE EVID II SS ID TIME DV SEX AGE WT HT CLCR
$ESTIM PRINT=0
$WARN WARN=NONE
$SUBROUTINE ADVAN3 TRANS4
$PK
TVCL=THETA(1)+THETA(5)*CLCR+THETA(8)*WT
CL=TVCL*EXP(ETA(1))
TV1=THETA(2)+THETA(6)*CLCR+THETA(7)*WT
V1=TV1*EXP(ETA(2))
TVQ=THETA(3)
Q=TVQ*EXP(ETA(3))
TVV2=THETA(4)+THETA(9)*WT+THETA(10)*CLCR
IF (AGE.LT.75) THEN
TVV3=TVV2
ELSE
TVV3=TVV2*THETA(11)
ENDIF
V2=TVV3*EXP(ETA(4))
S1=V1
S2=V2
$THETA
(0,0.554) ;CL
(0,1.94) ;V1
(0,3.5) ;Q
(0,5) ;V2
(0.005) ;SlCLCRCL
(0.000005) ;slCLCRV2
(0, 0.05) ;slWTV1
(0, 0.002) ;slWTCL
(0, 0.052) ;slWTV2
(0.00005) ;slCLCRV2
(0, 1) ;V3
$OMEGA
0.3 ;[P]
0.5 ;[P]
$ERROR
IPRED = F
IRES = DV-IPRED
Y=F*(1+ERR(1))+ERR(2)
$SIGMA 0.66
0.5
Back to the Top
Jarrett,
Please provide some details of what you are talking about. What do you
mean by "I get a larger eta than the FOCE estimates"? Details,
details, details,...
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
n.holford.-at-.auckland.ac.nz
www.health.auckland.ac.nz/pharmacology/staff/nholford
Back to the Top
Hello Nick,
As you can see in the table below, the etas and sigmas from the FOCE
algorithm are outside the 95% prediction interval from the 200
bootstrap replicates. I tried to use FO estimations, omega block, but
still got the same problem.
Final estimates for pharmacokinetic parameters based on FOCE
estimation versus 200 bootstrap replicates
************************************************************
FOCE 200 Bootstrap replicates
Parameter Estimates Mean 95% Pred Interval
************************************************************
Theta(1) 0.524 0.493 0.285 - 0.700
Theta(2) 4.28 3.28 1.34 - 5.21
Theta(3) 5.16 5.12 4.20 - 6.04
Theta(4) 1.35 1.43 -0.38 - 3.23
Theta(5) 0.00238 0.00238 0.000963 - 0.00379
Theta(6) -0.031 -0.0251 -0.0372 - -0.0131
Theta(7) 0.0569 0.0573 0.0374 - 0.0771
Theta(8) 0.00317 0.00336 0.000253 - 0.00647
Theta(9) 0.0752 0.0835 0.0583 - 0.109
Theta(10) -0.0129 -0.0150 -0.0265 - -0.00342
Theta(11) 1.41 1.77 1.35 - 2.18
Eta(1) 0.095 0.317 0.266 - 0.369
Eta(2) 0.218 0.399 0.326 - 0.471
Eta(3) 0.186 0.577 0.387 - 0.767
Eta(4) 0.0986 0.473 0.319 - 0.627
Sigma(1) 0.0233 0.132 0.114 - 0.150
Sigma(2) 0.24 0.875 0.258 - 1.49
************************************************************
Thank you very much for your help. I really appreciate it.
Regards,
Jarrett
Back to the Top
The following message was posted to: PharmPK
Is it that you may be extracting CVs or SD in bootstraps and comparing
to variances from FOCE estimates?
varun goel
Back to the Top
Jarrett,
I think that others (thank you Yaning and Varun) have already pointed
out that WFN tries to help you by reporting the square root of OMEGA
and SIGMA in all its output (including bootstrap summaries).
If you would prefer to see only variances then set the environment
variable nmraw as follows before using WFN:
set nmraw=y
Best wishes,
Nick
--
Nick Holford, Dept Pharmacology & Clinical Pharmacology
University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New
Zealand
n.holford.aaa.auckland.ac.nz
www.health.auckland.ac.nz/pharmacology/staff/nholford
Back to the Top
The following message was posted to: PharmPK
Dear Jarrett,
Could you give some more details, in particular: how did you do the
bootstrap analysis? In your first message you wrote:
> I have encountered an issue that when I ran bootstrap using WFN
I don't know what is WFN.
> As you can see in the table below, the etas and sigmas from the FOCE
> algorithm are outside the 95% prediction interval from the 200
bootstrap
> replicates.
I assume you mean omegas, i.e. the variance of the etas. Are you sure
that
all values for omega (eta) and sigma refer to variances, not to standard
deviations? This could make the difference.
best regards,
Hans Proost
Johannes H. Proost
Dept. of Pharmacokinetics and Drug Delivery
University Centre for Pharmacy
Antonius Deusinglaan 1
9713 AV Groningen, The Netherlands
Email: j.h.proost.-a-.rug.nl
Want to post a follow-up message on this topic?
If this link does not work with your browser send a follow-up message to PharmPK@boomer.org with "Wings for Nonmem" as the subject | Support PharmPK by using the |
Copyright 1995-2011 David W. A. Bourne (david@boomer.org)