**From biot0595@eurobretagne.fr Thu Jul 11 04:49:33 1996
**

I am currently studying a data file which contains 100 subjects, a unique dose per subject with covariates: AGE, WT, SEX, SECR...I have using the ADVAN4 and TRANS4 procedures towhich I radded a lag time. I radded a SEX covariate to the ALAG1 variable in this program. As well as the usual outputs, I have also noticed a "T MATRIX" output, which is not mentioned in the NONMEM users guide. Could you please give me some information on this?

Thank you for your help.

****

**From alison Thu Jul 11 08:26:31 1996
**

NONMEM's estimate of the precision of its parameter estimates is computed in the Covariance Step (COVARIANCE MATRIX OF ESTIMATE). This matrix is computed from the R and S matrices. Error messages from the covariance step indicating that these matrices are not positive definite or are singular indicate that the minimization routine did not find a true minimum. When the S matrix is singular, NONMEM attempts to compute and display a related matrix, the T matrix, which can be used to define a confidence region of the parameter space.

See Guide V, Chapter 13, in which errors in the Covariance step are discussed.

Possibly the covariate SEX has been added to the model for ALAG1 in a way that is incorrect. Perhaps some regression parameter does not influence the objective function, or perhaps two or more parameters are confounded and cannot be separately identified. Or perhaps the model has simply become too large for the amount of data.