RE: $OMEGA blocks and log-likelihood profiling

From: Nick Holford n.holford@auckland.ac.nz Subject: RE: [NMusers] $OMEGA blocks and log-likelihood profiling Date: Tue, June 1, 2004 5:15 pm Mark, You will find some anecdotal evidence about the influence of including all runs or only successful runs for bootstrap CI here: http://www.cognigencorp.com/nonmem/nm/99jul152003.html Some remarks I made then were: "I bootstrapped the original data set using the preferred model and found 28% of bootstrap runs minimized successfully and 7.1% ran the $COV step. " "The mean of the parameters obtained from all bootstrap runs and the mean from those which ran the $COV step were all within 2%." "95% confidence intervals obtained from all the bootstrap runs were very similar to those obtained from minimization successful and $COV successful runs." "I computed the ratio of the mean standard error from the $COV successful runs to the bootstrap standard error obtained from all runs. For THETA:se estimates the $COV SE was on average 3% smaller but for OMEGA:se the $COV SE was 58% larger than the overall bootstrap SE." I would add that because $COV ran on some of the bootstrap samples the failure of NONMEM to run $COV does not mean the model is somehow badly formed (e.g. overparameterized). It must mean that NONMEM fails $COV because of the data because the only difference between runs that ran $COV and those that did not was a random sample of the data. BTW key parts of the results of the bootstrap exercise I described last year are on-line and will appear soon in print in BJCP: http://www.blackwell-synergy.com/links/doi/10.1111/j.1365-2125.2004.02114.x/full Nick -- Nick Holford, Dept Pharmacology & Clinical Pharmacology University of Auckland, 85 Park Rd, Private Bag 92019, Auckland, New Zealand email:n.holford@auckland.ac.nz tel:+64(9)373-7599x86730 fax:373-7556 http://www.health.auckland.ac.nz/pharmacology/staff/nholford/