From:  "Liesbeth Rook" 
Subject: [NMusers] Coding for Bioavailability
Date:Tue, 22 Apr 2003 09:54:09 +0200

Dear NM users

At the moment I am modelling data of a compound administered to patients intravenously
and by inhalation. The bioavailability after inhalation is known to be in the order of
0.5. 
Because of the expected large inter-individual variation in inhalation technique, we did
not want to fix the F but estimate it. Bioavailability however should be restricted to a
value varying between 0-1. Therefore, we have defined F in the control stream as follows 

F1=EXP((LOG(TVF1*(1-TVF1))+ETA(5))/(1+EXP(LOG(TVF1*(1-TVF1))+ETA(5))))

However, according to this equation, when F is estimated as 0.5 and ETA=0, the calculated
outcome for F1 is 0.33.
The PRED are indeed much lower than the DV when I apply this equation. 

If I use F without any restriction in the control stream, then NM has problems in finding
the solution. 

Has anybody tackled this problem before or have I made a mistake in the equation (which
was cited from Karlson et al., Assumption testing in PPK models, J. of Pharmacokinetics
and Biopharmaceutics, 1998, page 211) ?

Kind regards
Liesbeth Rook
_______________________________________________________

From:VPIOTROV@PRDBE.jnj.com
Subject: RE: [NMusers] Coding for Bioavailability
Date:Tue, 22 Apr 2003 10:27:51 +0200

Liesbeth, 

The right coding is as follows: 

F1=EXP(LOG(TVF1/(1-TVF1))+ETA(5))/(1+EXP(LOG(TVF1/(1-TVF1))+ETA(5))) 

Best regards, 
Vladimir 

_______________________________________________________

From:  "Liesbeth Rook" 
Subject: RE: [NMusers] Coding for Bioavailability
Date: Tue, 22 Apr 2003 14:16:12 +0200

Vladimir and Siv
Thank you for the correct equation. NM is running smoothly now..
Liesbeth Rook
_______________________________________________________