Back to the Top
The following message was posted to: PharmPK
We have just completed a paper using P-Pharm and I have one question
about the correct description of the methods we have used. The relevant
section of the text is as follows:
Compartmental models (1 or 2) were fitted to the plasma
concentration-time data and analyzed using a population pharmacokinetic
approach implemented in the program P-Pharm Version 1.5 (InnaPhase,
France). A heteroscedastic error variance (1/Y2) model was used to
describe residual error and a log normal distribution were used to
describe inter-patient variability in clearance (CL), volume of
distribution (V) and absorption rate constant (ka). Bioavailability was
assumed to be unity.
My question is:
With respect to heoeroscedastic error variance using theY squared option
- should it be written as (*Y2) or (1/Y2) or what ????
Back to the Top
The following message was posted to: PharmPK
Wheather it should be Y^2 or 1/Y^2 dependending on how the objective fuction
was expressed. For example,
if the objective function is expressed as
F = Sum of (Yobs(i)-Ypred(i))Wi +..., then Wi=1/Y^2; (1)
or F = Sum of (Yobs(i)-Ypred(i))/Wi + ..., then Wi=Y^2. (2)
In both Kinetica and P-Pharm, the obj function is defined as expression (2).
Scientific Support
InnaPhase Cor.
Philadelphia
USA
PharmPK Discussion List Archive Index page
Copyright 1995-2010 David W. A. Bourne (david@boomer.org)