RE: OMEGA HAS A NONZERO BLOCK

From:"Serge Guzy" Subject:RE: [NMusers] OMEGA HAS A NONZERO BLOCK Date:Tue, 8 Oct 2002 09:29:53 -0700 Ken, you stated correctly what I said. The problem is that I create an artifact by simulating an hypothetical situation and try to retrieve the characteristics of the population I simulated. The artifact is that I am using only Mathematics and Statistics and just a little bit of physiology (PK parameters must be positive). The result of my simulations/fitting show that high simulated correlation occur for all the bootstrap samples I generated. If I would not know there is correlation I would stand by what I got using the scientific tool I used and that is it. Here come the experts in the biologic field (I am not an expert) and tell me "Serge, this high correlation is not realistic". I do not have an answer to that but I am a little concerned about what to do. What should I do indeed? Should fitting results match always with what we know from Physiology? To what extent Physiology is relevant when in fact we use a mathematical/statistical method to find a maximum of a specific objective function? Should it be enough just to make sure the fitting parameter are positive? If selecting a correlation to a specific value which correspond to an objective function that is not as good as the one I got, then why not changing the value of all the parameters I believe are not realistic? Serge