RE: Significant Figures Continuation

________________________________ From: [email protected] [mailto:[email protected]] On Behalf Of Elassaiss - Schaap, J. (Jeroen) Sent: 20 January 2010 16:36 To: Indranil Bhattacharya; [email protected] Subject: RE: [NMusers] Significant Figures Continuation Dear Indranil, Nick Holford has posted his observation on the use of SIGDIG in Nonmem v6 to this list. You can find the results at his site: http://www.fmhs.auckland.ac.nz/sms/pharmacology/holford/projects/sigdig. aspx If you use ADVAN 6 to 9, you may want to set TOL to a number really higher (at least +3) than SIG to get consistent results, as mentioned during the NM7 introduction by Bob bauer. So you may interpret SIG as a convergence criterium and your results as some instability. Try to play around with ADVAN 6-9, SIG, TOL and ML approach (FOCE - LAPL) to search for stability in outcomes. Or better, refine your model to stabilize it. A third, easier, solution would be to do a bootstrap and check for stability / %CV outcomes that way. Best regards, Jeroen Jeroen Elassaiss-Schaap Modeling & Simulation Expert Pharmacokinetics, Pharmacodynamics & Pharmacometrics (P3) T: +31 (0)412 66 9320 F: +31 (0)412 66 2506 [email protected] <mailto:[email protected]> ________________________________ From: [email protected] [mailto:[email protected]] On Behalf Of Indranil Bhattacharya Sent: Wednesday, 20 January, 2010 15:21 To: [email protected] Subject: [NMusers] Significant Figures Continuation First, of all sorry about the last email, it somehow got sent before I could complete it. Hi, while running a KOKA model I came across an observation and would greatly appreciate it if someone helps me understand it. When using a SIG=3, the model converges, and provides lower and upper CI bounds. All the CI bounds are positive except Q and the CV% of all parameters are within 20 % (except Q). This suggests (correct me if I am wrong) that the data might not be rich enough for the model to calculate a precise value for Q so it gives a wider range? Now when I change the SIG=3 to SIG=4, the model converges but the CV% are very high and more importantly I get negative lower bounds for all the parameters. What does this suggest, that the model cannot get precise estimates now at the level of significance (of numbers) requested? Please advise. Regards Neil -- Indranil Bhattacharya -- Indranil Bhattacharya ________________________________ This message and any attachments are solely for the intended recipient. If you are not the intended recipient, disclosure, copying, use or distribution of the information included in this message is prohibited --- Please immediately and permanently delete. ________________________________