Jian,
It is probably not reasonable to expect that you can take the final run from
a potentially long analysis using NONMEM V and expect it to generate the
exact same results using NONMEM VI. There are many changes between V and VI
and the predominant thinking is that it has changed for the better (faster
and more stable). There are some cases where NONMEM V results may be
different (even better) than NONMEM VI. This is not unreasonable given that
it uses new algorithms. There are also some cases where you can replicate
NONMEM V results with NONMEM VI if you use the SLOW estimation option.
Your approach is similar to running an analysis without ever changing the
initial estimates. It may take some logical adjustments to your model to
get "good" estimates with NONMEM VI.
Also, it might help if you showed the output from NONMEM VI and not just the
control file.
Bill
_____
Quoted reply history
From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] On
Behalf Of Jian Xu
Sent: Wednesday, April 09, 2008 11:28 AM
To: [email protected]
Subject: [NMusers] NM VI and NMV gave different results
Dear all,
I did a Pop PK a year ago in NM V, and got a stable model and reasonable
estimation. Recently, I tried to rerun the same code in NM VI (with latest
patches from Metrum) . The results are huge different. Some p-values for Eta
from NM VI are even significant. I really want to know what the real
difference is between NM VI and NM V. I appreciate any help or reference.
Thanks in advance.
Jian
Below is the control stream for your consideration:
$PROB 2CM
$DATA DATA.CSV IGNORE=#
$INPUT ID AMT TIME CON=DV MDV FST
$SUBROUTINES ADVAN4 TRANS4
$PK
V2=THETA(1)*EXP(ETA(1)) ; VOLUME OF CENTRAL
V3=THETA(2)*EXP(ETA(2)) ; VOLUME OF PERIPHERAL
CL=THETA(3)*EXP(THETA(7)*ETA(1)) ; CLEARANCE
Q =THETA(4)*EXP(ETA(3)) ; DISTRIBUTION CLEARANCE
IF (FST.EQ.1) THEN
KA=THETA(5)*EXP(ETA(4)) ; ABSORPTION RATE CONSTANT FOR FASTED
ELSE
KA=THETA(6)*EXP(ETA(5)) ; ABSORPTION RATE CONSTANT FOR FED
END IF
S2=V2
$ERROR
Y=F*(1+EPS(1))
$THETA
(0,150) ;V2_F
(0,200) ;V3_F
(0,6) ;CL_F
(0,50) ;Q_F
(0,0.5) ;KA
(0,0.2) ;KA
(0,1) ;CL_V Correlation
$OMEGA
0.01 0.01 0.01 0.01 0.01
$SIGMA
0.04
$ESTIMATION SIG=3 MAX=5000 PRINT=10 METHOD=1 INTER NOABORT
$COVARIANCE
$TABLE ID TIME NOPRINT FILE=BASE
&&&&&&&&&&
&GO GATORS! &
&&&&&&&&&&
__________________________________________________
Do You Yahoo!?
Tired of spam? Yahoo! Mail has the best spam protection around
http://mail.yahoo.com
_____
No viruses found in this incoming message
Scanned by iolo AntiVirus 1.5.3.5
http://www.iolo.com http://www.iolo.com/iav/iavpop3
_______________________________________
No viruses found in this outgoing message
Scanned by iolo AntiVirus 1.5.3.5
http://www.iolo.com