mixed in R (EMA B ≠ FDA) [Design Issues]

posted by Helmut Homepage – Vienna, Austria, 2019-03-24 02:23 (2629 d 20:08 ago) – Posting: # 20077
Views: 36,417

Hi mittyri,

❝ If I remember correctly we were struggling with FDA model where sophisticated 'repeated' statement exists.

❝ We successfully crosschecked EMA method B (simple mixed effects model with a Subject as random effect).


Are you reminding Detlew and me obout our TODO-list? ;-)

❝ So linear mixed effects models are possible until FDA-style is required :cool:.


Yes, but that’s the point. However, recycled our code:

library(lmerTest) # (requires lme4, Matrix)
dta           <- read.table("exam45.dat", header=TRUE, na.strings="99999",
                            colClasses=c(rep("factor", 4), rep("numeric", 2)))
names(dta)[4] <- "treatment"
TR.only       <- dta[dta$treatment != "S", ]
ci            <- data.frame(rep(NA, 2), rep(NA, 4))
res           <- data.frame(method=c(rep("pooled", 2), rep("IBD", 2)),
                            PE=NA, CL.lo=NA, CL.hi=NA, CV=NA,
                            DFM=rep(c("Satterthwaite", "Kenward-Roger"), 2),
                            DF=NA, stringsAsFactors=FALSE)
for (j in 1:4) {
  if (j == 1) { # pooled (all at once)
    muddle <- lmer(log(AUC) ~ sequence + period + treatment + (1|subject),
                              data=dta)
  }
  if (j == 3) { # IBD (S excluded)
    muddle <- lmer(log(AUC) ~ sequence + period + treatment + (1|subject),
                              data=TR.only)
  }
  sum.muddle  <- summary(muddle, ddf=res$DFM[j])
  log.pe      <- sum.muddle$coefficients["treatmentT", "Estimate"]
  ci[j, 1:2]  <- round(100*exp(log.pe + c(-1, +1) *
                           qt(1-0.05, sum.muddle$coef["treatmentT", "df"]) *
                           sum.muddle$coef["treatmentT", "Std. Error"]), 2)
  res$PE[j]   <- round(100*exp(log.pe), 2)
  res$CL.lo[j] <- ci[j, 1]; res$CL.hi[j] <- ci[j, 2]
  res$CV[j]   <- round(100*sqrt(exp(sum.muddle$devcomp$cmp[["sigmaREML"]]^2)-1), 2)
  res$DF[j]   <- signif(sum.muddle$coefficients["treatmentT", "df"], 5)
}
print(res, row.names=FALSE)

Gives:

Method     PE  CL.lo  CL.hi    CV           DFM      DF
pooled 116.15 108.97 123.81 21.20 Satterthwaite 115.040
pooled 116.15 108.97 123.81 21.20 Kenward-Roger 114.630
   IBD 116.05 108.92 123.65 20.84 Satterthwaite  56.823
   IBD 116.05 108.91 123.65 20.84 Kenward-Roger  56.468


Similar ≠ same.

Though the DFs are slightly different, the CIs look only identical due to rounding.
—
Dif-tor heh smusma 🖖🏼 Довге життя Україна! [image]
Helmut Schütz
[image]

The quality of responses received is directly proportional to the quality of the question asked. 🚮
Science Quotes

Complete thread:

UA Flag
Activity
 Admin contact
23,653 posts in 4,991 threads, 1,570 registered users;
368 visitors (0 registered, 368 guests [including 38 identified bots]).
Forum time: 23:31 CEST (Europe/Vienna)

I’m all in favor of the democratic principle
that one idiot is as good as one genius, but I draw the line
when someone takes the next step and concludes
that two idiots are better than one genius.    Leo Szilard

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