last 3 time points [🇷 for BE/BA]

posted by martin  – Austria, 2008-10-12 02:04 (6449 d 13:12 ago) – Posting: # 2519
Views: 26,349

dear Yung-jin Lee !

good to hear that your code will be much more sophisticated than WinNonLin! I appreciate your efforts. you may find of interest that in the case of manual selection of the last 3 time points, the middle time point is “useless” to calculate lambda_z by using the OLS fitting criteria in the case of equidistant measurement time points. in this case, the middle data tuple only effects the intercept but not the slope (=lambda_z). you may think about implementing a warning message in this case.

the theoretical background for this unexpected behavior rather complicated but I can provide you some information upon request.

some examples:
t <- c(10, 12, 14) # equidistant time points
c1 <- c(5, 4, 1)
c2 <- rnorm(3)
c3 <- runif(3)

lm(c1~t)
lm(c1[-2]~t[-2])

lm(c2~t)
lm(c2[-2]~t[-2])

lm(c3~t)
lm(c3[-2]~t[-2])

Complete thread:

UA Flag
Activity
 Admin contact
23,653 posts in 4,991 threads, 1,570 registered users;
373 visitors (0 registered, 373 guests [including 29 identified bots]).
Forum time: 15:16 CEST (Europe/Vienna)

The epistemological value of probability theory is based on the fact
that chance phenomena, considered collectively and on a grand scale,
create non-random regularity.    Andrey Kolmogorov

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