search

nlmixr2 / nlmixr2est

nlmixr2 estimation routines
https://nlmixr2.github.io/nlmixr2est/
GNU General Public License v2.0
7 stars 3 forks source link

`est = "posthoc"` provides the same values for PRED and IPRED #436

Closed billdenney closed 6 months ago

billdenney commented 6 months ago

The same model I've been working on for the past few days has an issue where when I use est = "posthoc", I get the same values for the PRED and IPRED columns. Is there any reason that this should be happening?

mattfidler commented 6 months ago

If the EBE for all subjects is zero

On Sat, Apr 6, 2024, 7:02 PM Bill Denney @.***> wrote:

The same model I've been working on for the past few days has an issue where when I use est = "posthoc", I get the same values for the PRED and IPRED columns. Is there any reason that this should be happening?

— Reply to this email directly, view it on GitHub https://github.com/nlmixr2/nlmixr2est/issues/436, or unsubscribe https://github.com/notifications/unsubscribe-auth/AAD5VWXIBFR7K7JD425JFZLY4CEKJAVCNFSM6AAAAABF22BR66VHI2DSMVQWIX3LMV43ASLTON2WKOZSGIZDSNBWHA4DSMI . You are receiving this because you are subscribed to this thread.Message ID: @.***>

billdenney commented 6 months ago

The EBE aren't all zero.

For the moment, my work-around is to remove the iiv from the ini() block of the model, left_join(data, ranef(model)), and then use est = "rxSolve".

mattfidler commented 6 months ago

That's an OK workaround, though it shouldn't be the goto solution.

mattfidler commented 6 months ago

Without a reproducible example, I cannot fix this :face_with_head_bandage: I will close this for now. Once you get a reproducible example I will look at this a bit more.