Pharmacometrics CHLD0085: Assignment
Candidate number
Examplokacin is a new aminoglycoside antibiotic. It has been given at a dose of 1000 mg by intravenous bolus
to 50 volunteers, followed by intensive PK sampling. The data are provided in the file examplokacin.csv.
Using the Rmarkdown template provided, answer the questions below to create a Word or PDF document
containing your answers along with the commented R code within each R chunk. You must label your
assignment with your candidate number rather than your name to facilitate anonymous marking.
1 Write a sentence about each of the 4 PK processes (ADME)
from your reading about aminoglycoside PK (4 marks)
2 Read in the data and make a plot of the data, with lines con-
necting data from each individual, appropriate axis labels and
using a log scale for the y-axis (4 marks)
# CODE HINT: In lattice a log scale can be specified with:
# xyplot( ..., scales=list(y = list(log = TRUE)))
3 Undertake a non-compartmental analysis of the data, reporting
the median, 2.5th, and 97.5th percentiles of Cmax, Elimination
half-life, AUC(0−t), AUC(0−∞) (4 marks)
# CODE HINT: an amended version of the function in Hands-on 1:
trap.rule <- function(x,y) sum(diff(x) * (y[-1] + y[-length(y)])) / 2
# Specify function to do NCA for each individual
calc.nca <- function(ID){
#####
##### Assumes the dataset is called df
#####
out <- df[df$ID==ID,]
# Calculate AUC(0-t) using trap.rule function
auct <- trap.rule(out$TIME, out$DV)
# Select last 3 rows for ke calc
temp1 <- tail(out, n = 3)
# linear regression for slope of log of DVs
abc <- lm(log(temp1$DV) ~ temp1$TIME)
slope <- summary(abc)[[4]][[2]]
1
ke <- -slope
# Cmax
cmax <- max(out$DV)
# Tmax
tmax <- out$TIME[out$DV==cmax]
# AUC(0-inf)
# first find AUC(t-inf)
C_last <- tail(temp1$DV, 1)
auct_inf <- C_last / ke
auc0_inf <- auct + auct_inf
data.frame(ID, cmax, tmax, ke, auct, auc0_inf)
}
4 Using nlmixr() fit 1 and 2 compartment models and test addi-
tive or proportional residual errors. Use the Objective Func-
tion Value, plots of NPDE versus TIME and |IWRES| ver-
sus IPRED to choose your preferred structural and statistical
model. Make a VPC of your final model, and create a table of
parameter estimates (8 marks)
# CODE HINT 1: Remember you can save your run output with e.g.:
# saveRDS(run1.lst, "run1.rds")
# And read it back in with e.g.:
# run1.lst <- readRDS("run1.rds")
# This can save you running the model every time you compile your answer document
# CODE HINT 2: Some useful functions for the plots you need:
npde.gof <- function(run.fit){
# Set some options for colour, point type line styles/width
# Have a go at changing these to see how the plots change
cex.n <- 1 ; cex.pan <- 1 ; cex.pt <- 1
type.n <- "b"; pch.n <- 16 ; lty.n <- 2
lwd.n <- 0.5 ; col.n <- grey(0.6)
# run.fit <- run1.lst
goftab <- data.frame(run.fit)
# NPDE vs TIME
npde.time <- xyplot(NPDE ~ TIME, data = goftab, groups = ID, type = type.n,
pch = pch.n, lty = lty.n, lwd = lwd.n,
col = col.n, cex = cex.pt,
ylab=list(label = "NPDE", cex = cex.n),
xlab=list(label = "Time after first dose (h)", cex = cex.n),
xlab.top=list(label = "NPDE vs time", cex = cex.n),
panel=function(...){
panel.xyplot(...)
# Add zero and +/-2 lines
panel.abline(0, 0, lwd = 2) ; panel.abline(h = 2,
2
lwd = 2, lty = 2)
panel.abline(h = -2, lwd = 2, lty = 2)
panel.loess(goftab$TIME, goftab$NPDE,
lwd = 2, col = "red", span = 1)})
print(npde.time)
}
iwres.gof <- function(run.fit){
# Set some options for colour, point type line styles/width
# Have a go at changing these to see how the plots change
cex.n <- 1 ; cex.pan <- 1 ; cex.pt <- 1
type.n <- "b"; pch.n <- 16 ; lty.n <- 2
lwd.n <- 0.5 ; col.n <- grey(0.6)
# run.fit <- run1.lst
goftab <- data.frame(run.fit)
iwres.ipred <- xyplot(abs(IWRES) ~ IPRED, data = goftab, groups = ID, type = type.n,
pch = pch.n, lty = lty.n, lwd = lwd.n,
col = col.n, cex = cex.pt,
ylab=list(label = "|IWRES|", cex = cex.n),
xlab=list(label = "Individual prediction (mg/L)", cex = cex.n),
xlab.top=list(label = "|IWRES| vs IPRED", cex = cex.n),
panel=function(...){
panel.xyplot(...)
panel.loess(goftab$IPRED, abs(goftab$IWRES),
lwd = 2, col = "red", span = 1)})
print(iwres.ipred)
}
3








































































































学霸联盟