Parameter estimates don't converge with a multiplicative model

[david.reith_1]

I have an unusual problem where I have a pretty simple model (one compartment, first order extravascular absorption and first order elimination). The model converges fine with an additive, log-additive or exponential residual error, but with a multiplicative or combined additive + multiplicative residual error model the random effects converge but the fixed effects (Ka, CL.F and V/F) stay static - and don’t converge.

Has anybody else encountered this situation?

(PS I am using Phoenix 8.1 NLME)

[Samer_Mouksassi]

do you have any DV= zeros that you are trying to fit ? unless you have mdv = 1 and you map it or you have empty dv instead of zero the engine will try to fit the zero

[david.reith_1]

@smouksassi1

do you have any DV= zeros that you are trying to fit ? unless you have mdv = 1 and you map it or you have empty dv instead of zero the engine will try to fit the zero

Thanks, good advice. I had already checked for observations of 0 and there were none. I just checked again and there were missing observations where MDV was 0. I fixed this, but it still won’t iterate the PK parameters. So I think I need to keep looking at this dataset.