Dear Matt, We published a multiple-dose implementation of the transit model a little while back which may be of help (although it uses the mathematical solution from Savic et al 2007, and assumes the entire bioavailable dose has left the absorption compartment before the next dose is given). I'm not sure that hard-coding the number of compartments as you suggest is an easier solution, though, given the time it might take to empirically (?) narrow down the number of transit compartments you need, and the computational overhead of using so many differential equations. Why did you choose 25 as a starting point, incidentally?
We had similar experiences (numerical instability under some circumstances) in our own implementation of the Stirling approximation to the gamma function in NONMEM VI. Wilkins JJ, Savic RM, Karlsson MO, Langdon G, McIlleron H, Pillai G, Smith PJ, Simonsson US. Antimicrob Agents Chemother. 2008 Jun;52(6):2138-48. Savic RM, Jonker DM, Kerbusch T, Karlsson MO. J Pharmacokinet Pharmacodyn. 2007 Oct;34(5):711-26 Best Justin On 16 July 2010 18:00, Zierhut, Matt <matt.zier...@amylin.com> wrote: > All, > Has anyone ever used a transit compartment model in NONMEM for PK with > multiple doses. We are looking to use ~25 transit compartments (and thus > ~25 DiffEqs in the $DES block). This seems easier for multiple doses, > compared to the mathematical solution detailed in Rada Savic's 2004 PAGE > presentation. We are also trying to implement the gamma scaling factor in > the transit compartment equations (Sun YN, Jusko WJ. J Pharm Sci. 1998 > Jun;87(6):732-7), but it seems to be fairly unstable in NONMEM. Can > anyone comment/help? > > Thanks in advance, > Matt > -- Justin Wilkins ----------------------------- Colmarerstrasse 31 4055 Basel Switzerland ----------------------------- Email: justin...@gmail.com Skype: kestrel_za2002 Mobile: +41 76 561 0949