time_to_event

From: lewis@c255.ucsf.EDU (LSheiner) Subject: time_to_event Date: 22 Jan 1998 15:44:23 -0500 Responding to the emails reproduced below, I have modified the old LOGIST.DIR in NONMEM.DIR at http://pkpd.icon.palo-alto.med.va.gov (IP address 204.161.113.34) as follows. I have 1. Changed the name of LOGIST.DIR to NON_CONTINUOUS.DIR 2. Modified README.009 2. Added a control stream fora Weibull-type time-to-event model The time-to-event code has the limitation that it does not deal with continuously time varying hazard. This is something that, as I said in my original note, I am still working on, and I hope to add a control stream for this in the near future. =============================================================== From penend1@pfizer.com Thu Jan 15 07:19:10 1998 Dear NONMEM Users, It seems to me that the example Lew gave is dealing with a class of survival models called multiplicative intensity models and do not necessarily extend to mixed effect modeling. To my knowledge no marketed software is available for this type of modeling. It would be rather exciting if some one could develop these models in NONMEM, though I still maintain that the NONMEM software is not necessarily the most efficient or practical place for these models. The best reference for these models is Counting Processes and Survival Analysis by Thomas R. Fleming and David P. Harrington published by Wiley Interscience. Regards Darryl Penenberg Assistant Director Biometrics, Clinical Operations Pfizer Pharmaceuticals