From: "Milly de Jonge" <Apmil@SLZ.NL>
Subject: [NMusers] mixture model
Date: Wed, 14 Nov 2001 11:44:16 +0100

Dear NONMEM users,

At the moment I am working on a pk model for both parent and metabolite. Kinetic data is rich with more than 8 samples per patient. With this data FOCE with interaction is the best method to use. However, due to long run times I am using log-transformed data with the first-order method. Inclusion of a mixture model for the residual error of the parent compound produced a large decrease in objective function. For the metabolite a similar reduction was observed. In my next model I would like to implement (independent) mixture models for both the parent and metabolite. Does someone know whether (and how) this is possible?

Thanks in advance.

Milly de Jonge