Hello, Is there a centering option in NLME, whereby the likelihood function is modified to result into a more centered fit by forcing the interindividual error terms to be distributed about zero? Also, in Nonmem, in cases when the residual variability (eps) is not independent of inter-individual variability (eta) (meaning that there is an interaction between eta and eps), one ca use FOCE with interaction (FOCE-I) to account for such interaction during the estimatin of the likelihoo function. Is there something similar we can use with NLME? Furthermore, hybrid models can be useful when lag-times are included as random effects in an extravascular administration model. How can I implement this approach in NLME? Thanks!
Hi Dora, yes there is - have a look on the Parameters tab, and under the Structual sub tab you will find several options; 
Hi Simon, I was actualy not asking about the centering of the covariates. I am asking about a less used option in Nonmem coded by the statement: $EST CENTERING. Typically, it is assumed that eta has the mean zero and that when the population model is a good fit, then the mean of eta will be close to zero. Such a fit is “centered”. However, none of he estimation algorithms guarantee a centered fit. With the centering option in Nonmem the likelihood function is modified to result into a more centered fit by forcing the interindividual error terms to be distributed about 0. I was wondering if I can somehow code this in Phoenix. Thanks!! Dora