Bioequivalence and Bioavailability Forum

❝ ❝ I used the optimized values of 0.0303 and 0.0282. No significant inflation of the TIE (maxima 0.05015 and 0.05010).

❝

❝ Clever!

❝ ❝ ...and running the simulations took me close to two months.

❝ I heard that all the "for..." constructions are not so effective in R and better to use vectors for all "for ...to" and calculate them.

runs  <- 1e5
run   <- 1:runs
ptm   <- proc.time()
v1    <- NULL
for(j in seq_along(run)) v1 <- c(v1, rnorm(1, mean=0, sd=1))
t1    <- proc.time()-ptm
ptm   <- proc.time()
v2    <- vector("numeric", length=runs)
for(j in seq_along(run)) v2[j] <- rnorm(1, mean=0, sd=1)
t2    <- proc.time()-ptm
ptm   <- proc.time()
v3    <- rnorm(number, mean=0, sd=1)
t3    <- proc.time()-ptm
cat("\n",
"increasing vector:", t1[3], "seconds\n",
"indexed vector   :", t2[3], "seconds\n",
"internal function:", t3[3], "seconds\n")


 increasing vector: 11.74 seconds
 indexed vector   : 0.29 seconds
 internal function: 0.02 seconds

❝ Have you any experience on this?

library(Power2Stage)
power.2stage(method="C", alpha=rep(0.0282, 2), CV=0.1, n1=16,
             theta0=1.25, nsims=1e6)

Total time consumed (secs):
   user  system elapsed
    3.9     0.1     4.0

Method C: alpha0 = 0.05, alpha (s1/s2) = 0.0282 0.0282
Target power in power monitoring and sample size est. = 0.8
BE margins = 0.8 ... 1.25
CV = 0.1; n(stage 1)= 16; GMR = 0.95
GMR = 0.95 and mse of stage 1 in sample size est. used
Futility criterion Nmax = Inf

1e+06 sims at theta0 = 1.25 (p(BE)='alpha').
p(BE)    = 0.0501
p(BE) s1 = 0.0501
Studies in stage 2 = 0%

Distribution of n(total)
- mean (range) = 16 (16 ... 22)
- percentiles
 5% 50% 95%
 16  16  16

❝ (Read the R developers are thinking about an API for distributed computing. so we can share all our machines for large simulations )