On 5 Oct 1999 at 23:17:27, "INGRAM Wendy, Pharmacy" (wendy.ingram.-a-.phnt.swest.nhs.uk) sent the message
Can someone point me in the right direction?!!
I'm writing a model in WinNonlin - I need to model my data to a
three-comp, first order input, first-order output model which
incorporates a lag time and has primary parameters expressed as
macro-constants.
I've done this so far:-
MODEL
COMM
NPARM 8
PNAMES 'A', 'Alpha', 'B','Beta','C','Gam', 'k01', 'tlag'
remark NSEC 14
remark SNAMES 'k10', 'k12', 'k21', 'k13', 'k31', 'k10_HL', 'k01_HL',
'Alpha_HL', 'Beta_HL', 'Gam_HL', 'Cl', 'Cmax', 'auc', 'v'
NCON 1
END
TEMP
D=CON(1)
A=P(1)
Alpha=P(2)
B=P(3)
Beta=P(4)
C=P(5)
Gam=P(6)
k01=P(7)
tlag=P(8)
e=-(A+B+C)
T=X - tlag
END
FUNC1
F = A*DEXP(-Alpha*T) + B*DEXP(-Beta*T) + C*DEXP(-Gam*T) + e*DEXP(-k01*T)
END
EOM
I'm unsure about the time lag? Have I written this properly?
Any comments will be much appreciated.
Thanks.
Wendy Ingram
Research Student
email: wendy.ingram.-at-.phnt.swest.nhs.uk
Back to the Top
On 6 Oct 1999 at 22:39:32, David_Bourne (david.at.boomer.org) sent the message
[Two replies - db]
X-ExtMailInfo: thompson.ga.at.notes.ntssmtpgw.na.pg.com
bdc-notes041.na.pg.com [155.125.116.41]
X-Lotus-FromDomain: PGI
From: "Gary Thompson-GA"
To: pharmpk.-at-.boomer.org
Date: Wed, 6 Oct 1999 08:33:35 -0400
Subject: Re: PharmPK WinNonlin model writing????
Wendy
try using the following regarding lagtime - the first two statements
are what is
needed with the Dmax argument
T1=T-TL
TA=DMAX1(0,T1)
C1=A1*(DEXP(-G1*(TA)))
C2=A2*(DEXP(-G2*(TA)))
C3=A3*(DEXP(-G3*(TA)))
C4=-(A1+A2+A3)*(DEXP(-KA*(TA)))
F=C1+C2+C3+C4
it will set the time to zero prior to the lagtime and use the difference after.
---
From: Jeff Wald
X-Accept-Language: en
To: PharmPK.at.boomer.org
Subject: Re: PharmPK WinNonlin model writing????
Just a reminder to the WinNonlin (and other Pharsight software users) on the
PharmPK list. Technical support for PK, PD, and statistical questions is
available to Pharsight users by sending an e-mail to support.aaa.pharsight.com.
By using this e-mail, it will provide you with the most timely response and
it will allow us to track the types of question that users have so we can be
more responsive in the future.
For the interest of the group, I have copied a working version of the model
in question below. It is also worthwhile to note that the parameters in
such a 4 exponent model are not mathematically identifiable when fitting to
data on only one compartment.
Jeff Wald, Ph.D.
Scientific Support Services
Pharsight Corporation
MODEL 1
COMM
NPARM 8
PNAMES 'A', 'Alpha', 'B','Beta','C','Gam', 'k01', 'tlag'
END
TEMP
A=P(1)
Alpha=P(2)
B=P(3)
Beta=P(4)
C=P(5)
Gam=P(6)
k01=P(7)
tlag=P(8)
e=-(A+B+C)
T=x-tlag
T=max(0,T)
END
FUNC 1
F = A*DEXP(-Alpha*T) + B*DEXP(-Beta*T) + C*DEXP(-Gam*T) + e*DEXP(-k01*T)
END
EOM
[my 2 cents - I think the eight parameters might be identifiable if
there is enough data spread over a wide time span and the values of
the rate constant are sufficiently separated. I'm not so sure about
the lag time but that should also be identifiable - could this be
another thread? - db]
Back to the Top