error at initial obj. function evaluation

From: "Gibiansky, Ekaterina" <Katyag@mocr.oapi.com> Subject: error at initial obj. function evaluation Date: 2 Feb 1998 21:09:51 -0500 Dear NONMEM users, Modeling infusion data, I have the following error message: PROGRAM TERMINATED BY OBJ, ERROR IN CELS WITH INDIVIDUAL 1 (IN INDIVIDUAL RECORD ORDERING) INTRAINDIVIDUAL VARIANCE OF DATA FROM OBS RECORD 1 ESTIMATED TO BE 0 MESSAGE ISSUED FROM ESTIMATION STEP AT INITIAL OBJ. FUNCTION EVALUATION I do use a proportional intraindividual model, but there are no zero concentrations in the observation records. What problems in the data or the code could cause this error? Below is the data for one person: ---------------------------------------- ID MDV RATE AMT TIME CONC ---------------------------------------- 210101 1 37.5 75.00 0.00 . 210101 0 0 0.00 1.00 58.24 210101 0 0 0.00 1.50 69.32 210101 0 0 0.00 2.00 45.81 210101 1 75 168.75 2.00 . 210101 0 0 0.00 3.25 94.49 210101 0 0 0.00 3.75 107.50 210101 0 0 0.00 4.25 163.58 210101 1 150 325.00 4.25 . 210101 0 0 0.00 4.58 157.60 210101 0 0 0.00 4.75 137.52 210101 0 0 0.00 24.00 11.19 ---------------------------------------- Thank you Katya Gibiansky ___________________________________ Otsuka America Pharmaceutical, Inc. 2440 Research Blvd, Rockville, MD 20850 E-mail: katyag@mocr.oapi.com Phone: (301)-527-4911 Fax: (301)-721-7511

From: alison@c255.ucsf.EDU (ABoeckmann) Subject: Re: error at initial obj. function evaluation Date: 2 Feb 1998 16:15:03 -0500 My advice is: Change the proportional model for epsilon to additive. Delete the $ESTIMATION record so that you can see what is happening with the initial estimates Include the record $TABLE ID TIME Rerun. Now you should be able to see predictions PRED for all records. Look to see if they are 0 for any observation record, in particular for the first one. Alison Boeckmann