Dear Francois,
Unless you create a new data set where events are allowed to occur at any possible time (usually a dense time grid), then the result that you have got is rather the expected (i.e. too good to be true). Best regards, Mats Mats Karlsson, PhD Professor of Pharmacometrics Dept of Pharmaceutical Biosciences Faculty of Pharmacy Uppsala University Box 591 75124 Uppsala Phone: +46 18 4714105 Fax + 46 18 4714003 From: owner-nmus...@globomaxnm.com [mailto:owner-nmus...@globomaxnm.com] On Behalf Of Francois Gaudreault Sent: 19 September 2012 22:49 To: nmusers@globomaxnm.com Subject: [NMusers] time-to-event simulation based diagnostic Dear NONMEM users I am currently performing a time-to-event analysis on binary data using a weibull distribution (time-varying hazard function). Using simulation based diagnositcs, I compared `simulated kaplan-Meier` to `observed kaplan-Meier`. However, fit looks to good to be true and I was actually suspecting a mispecification during the simulation process with PsN Here is the code used for vpc: vpc run2.mod -tte=RTTE -flip_comments -samples=100 I used (ICALL.EQ.4) and the flip-comment option in the model file. However, I am not sure if I have to generate a new simulation data set (each individual having a row for each possible observation time) or use the original one ? Any advises or comments are welcome Regards, -- François Gaudreault, Ph.D. Candidate Pharmacométrie / Pharmacometrics Chargé de cours / Lecturer Faculté de pharmacie / Faculty of Pharmacy Université de Montréal
