How to add IIV on residual unexplained variability when using SAEM

How-to
1

Hi!

I have encountered a problem when using Pumas to fit a PumasEMmodel, in emmodel, the random effect must be defined as a format of CL ~ 1 | LogNormal, but now I want to define a random effect of sigma, this is done easily in NONMEM by:

$ERROR
Y = IPRED + EPS(1) * EXP(ETA(1))

but in PumasEMmodel, the sigma is set in initial parameters tuple and I can’t add a random effect in @error block

@error begin
DV ~ Normal(IPRED)
end

if I set sdruv ~ 1 | LogNormal in random block and its initial value and omega vlaue and add it into @error block, like this

@error begin
DV ~ Normal(IPRED, sdruv)
end

it shows: ERROR: TaskFailedException

is there any way to solve this? Thanks in advance!

2

Hello Gbygbygby,

Welcome to the Pumas forum. It is currently not possible to model the conditional distribution of the data with a subject specific random effect on the variance when using @emmodel. Right now, it is only possible to model the conditional distribution like this when using @model. This is something that might change in the future.

Best
Andreas Noack