From: "Xiao, Alan" alan_xiao@merck.com
Subject: [NMusers] erros messages related to data file and/or NONMEM I/O
Date: Wed, April 27, 2005 4:45 pm
Hello Friends,
Could anyone help me debug this message in the beginning of a NONMEM report file?
========================================================
.........
0ERROR SUBROUTINE CALLED WITH EVERY EVENT RECORD.
1
MONITORING OF SEARCH:
0PROGRAM TERMINATED BY OBJ
ERROR IN ELS WITH INDIVIDUAL 1127 ID=0.00000000E+00
VAR-COV OF DATA FROM INDIVIDUAL RECORD ESTIMATED TO BE SINGULAR
MESSAGE ISSUED FROM ESTIMATION STEP
AT INITIAL OBJ. FUNCTION EVALUATION
========================================================================
The individual 1127 is the last one in the data file and I have no idea how this ID=0000
came from. I checked my data file, which was created via SAS 8.2, but nothing useful was
found. I used the same SAS program to generate many different data files only 2 out
of 8 files had this problem.
Thanks,
Alan
erros messages related to data file and/or NONMEM I/O
From: "Xiao, Alan" alan_xiao@merck.com
Subject: [NMusers] RE: erros messages related to data file and/or NONMEM I/O
Date: Thu, April 28, 2005 10:51 am
Hello,
Thank all those friends who kindly responded to my last email. It looks like
the discussion posted in the following link is relevant - special thanks to
Leonid for forwarding this to me.
http://www.cognigencorp.com/nonmem/nm/99oct041996.html
Based on this discussion, I slightly modified my model - still use my original
proportional residual model but add a very small additive error model (0.000001)
for test, the problem is gone.
Now, my question for your discussion is that why this happened to a data file (with a
limited dose range) but not others (e.g. a data file with the full dose range) even
both data files start with a concentration record (predose, but different ID) followed
by a dosing record - I used calendar date and clock time. Or is there anything else I'm missing?
Alan
_______________________________________________________