Re: OMEGA HAS A NONZERO BLOCK

From:Nick Holford [mailto:n.holford@auckland.ac.nz] Subject:Re: [NMusers] OMEGA HAS A NONZERO BLOCK Date:Wednesday, October 02, 2002 4:06 PM Ken, While I cannot disagree with the technical details of your proposal I would wonder why one would deliberately impose the assumption that the correlation between two parameters is 1. This seems very unlikely -- even more unlikely than the other extreme that the correlation is 0. If the data and estimation method do not allow one to estimate the correlation then I think it would be preferable to fix the covariance so that a reasonable correlation is assumed. I would assume that any stability benefits would also apply with this more realistic implementation. An analogy would be the well accepted procedure of fixing KA to a reasonable value when the data do not allow one to define the absorption process well. This topic was discussed on nmusers in May 2001 ( http://www.cognigencorp.com/nonmem/nm/97may092001.html) Mats Karlsson proposed this example of how to fix the covariance: CL= THETA(1)*EXP(THETA(3) * (ETA(1) + THETA(5)*ETA(3)) V = THETA(2) *EXP(THETA(4) * (ETA(2) + SQRT(THETA(5)*THETA(5))*ETA(3)) Where omega for ETA(1), ETA(2) and ETA(3) are fixed to 1. IF THETA(5) is negative, correlation is negative, whereas if it is positive correlation is positive. Nick PS Users of Wings for NONMEM ( http://wfn.sourceforge.net/) will of course readily diagnose the case when correlations approach 1 because these correlations are automatically calculated and presented to the user instead of just offering the less informative off-diagonal elements of OMEGA. Nick Holford, Divn Pharmacology & Clinical Pharmacology University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New Zealand email:n.holford@auckland.ac.nz tel:+64(9)373-7599x6730 fax:373-7556 http://www.health.auckland.ac.nz/pharmacology/staff/nholford/