Re: Simulating different populations

From: Liping Zhang ZHANG_LIPING@lilly.com Subject: Re: [NMusers] Simulating different populations Date: Wed, 10 Aug 2005 11:09:16 -0500 Hi, Max, what I typically do, is to sample (let's say 1000 replicates) from the distribution of "fixed effects" parameters in Splus, given estimated population parameter value and their variance-covariance matrix generated by NONMEM. So now you have a 1000 different "population", each has different "fixed effects", but overall they reflect the population parameter estimates and their uncertainty. Then through some simple Unix C-shell script I simulate the outcome from each "population" (with fixed population parameters and BSV and WSV) and do statistical summary according to modeling needs. As Ken Kowalski pointed out not long ago, this procedure is computer-labor intensive, you need to simulate large number of population, and within each population, a sufficient number trials. I would not go for this route unless I know for my modeling need it requires this kind of "truthful representation". best regards, Liping Liping Zhang, PhD Global PK/PD Modeling and Trial Simulation ELi Lilly & Company Work: 317-277-8687 Fax: 317-433-6661 DC 0734 Email: zhang_liping@lilly.com