lmer: Method B is ready for scaling [🇷 for BE/BA]

posted by mittyri – Russia, 2016-11-05 21:01 (2700 d 09:24 ago) – Posting: # 16779
Views: 19,180

❝ mittyri!!!

❝ May I hug you?!

:flower::party:

❝ Why didn't you provide the result of the code (that's the way of secret initiation?)


I was so surprized by the results that was in hurry to share my code :-D

❝ Please explain how to get Std.Error and Value with more digits?

Oh, you are right, please substitute confint string to
  FormulationT<-summary(muddle.lmer)$coefficients["FormulationT",]   cat(round(exp(FormulationT[1]+FormulationT[2]*qt(0.05, FormulationT[3]))*100, 2), "-",
      round(exp(FormulationT[1]-FormulationT[2]*qt(0.05, FormulationT[3]))*100, 2))

confint is not the right thing here. Here we go:

lmer treats Sequence, Period and Formulation as factors, Random is /~1|Subject/ (close to Method B), then
[1] 0.1460882
PE is 115.73
SAS gives 115.73
what about LSMeans?
PHX gives for Method B 7.67001367911898 and 7.81610190985527
and lmer gives
               Formulation Estimate
Formulation  R           R   7.6700
Formulation  T           T   7.8161

what about CI? SAS gives 107.17-124.97
107.17 - 124.97


❝ Why does the model have (1|Subject) as random? I also tried (1|Subject/Sequence) and the result (PE and CI) remains the same. Can we use it instead?


Yes, we can. Subject in Sequence is SAS legacy. By the way as ElMaestro wrote, that's meaningless, because the reference grid (G-matrix) doesn't change. It will be useful only if we are enumerating the subjects in each sequence from 1 (Seq ABAB: 1, 2, 3...; Seq BABA: 1, 2, 3...)

❝ The other question is what from the output we can use to estimate R-variance in order to apply scABE?


No, we cannot. Sorry, I must confess I don't have time now to implement this, but I would do it as follows:
- exclude all T data first and then
- build a linear model with lm() using only reference data.

If nobody does it during the next week I'll try to code it by myself.

❝ We are in front of the final step - from now we can make R tutorial to hold scABEL as did Helmut in WinNonLin, am I right? If it is so, Es ist fantastisch!:ok:


Ja, new feature to implement in BEAR, Yung-jin will be happy :cool:

Kind regards,
Mittyri

Complete thread:

UA Flag
Activity
 Admin contact
22,957 posts in 4,819 threads, 1,636 registered users;
110 visitors (0 registered, 110 guests [including 6 identified bots]).
Forum time: 06:25 CET (Europe/Vienna)

With four parameters I can fit an elephant,
and with five I can make him wiggle his trunk.    John von Neumann

The Bioequivalence and Bioavailability Forum is hosted by
BEBAC Ing. Helmut Schütz
HTML5