In metrumresearchgroup/lastdose: Calculate Time and Amount of Last Dose

lastdose

Overview

Calculate the time since and amount of the last dose. Additional (ADDL) dosing records are expanded and included in the calculation.

knitr::opts_chunk$set(comment = '.', message=FALSE, warning = FALSE, 
                      fig.path="man/figures/readme-")

Installation

remotes::install_github("metrumresearchgroup/lastdose")

library(lastdose)
library(tidyverse)
theme_set(theme_bw())

A PK profile

We'll use this PK profile as an example

file <- system.file("csv/data1.csv", package = "lastdose")

df <- read.csv(file) 

head(df)

The dosing runs over 12 weeks and there are 3 epochs, with 3 different doses, most of which are scheduled into the future via ADDL.

df %>% filter(EVID==1) %>% count(TIME,AMT,ADDL)

ggplot(df, aes(TIME,DV)) + geom_line() + theme_bw()

Calculate TAD, TAFD, and LDOS

Use the lastdose() function

df <- lastdose(df)

head(df)

Now we have TAD, TAFD, and LDOS in our data set.

Plot last dose versus time

ggplot(df, aes(TIME,LDOS)) + geom_line()

Plot time after dose versus time

ggplot(df, aes(TIME,TAD)) + geom_line()

Observations before doses at the same time by default

ggplot(df, aes(TIME,TAD)) + geom_line() + 
  scale_x_continuous(breaks = seq(0,72,4), limits=c(0,72)) + 
  scale_y_continuous(breaks = seq(0,24,4), limits=c(0,24)) 

You can also make doses "happen" first

dd <- lastdose(df, addl_ties = "dose_first")

ggplot(dd, aes(TIME,TAD)) + geom_line() + 
  scale_x_continuous(breaks = seq(0,72,4), limits=c(0,72)) + 
  scale_y_continuous(breaks = seq(0,24,4), limits=c(0,24)) 

All doses explicit in the data set

df2 <- mrgsolve::realize_addl(df) %>% lastdose()

ggplot(df2, aes(TIME,TAD)) + geom_line() + 
  scale_x_continuous(breaks = seq(0,72,4), limits = c(0,72)) + 
  scale_y_continuous(breaks = seq(0,24,4))

How does it perform on bigger data?

Same setup as the previous profile, but more individuals.

We have 500K rows and 1000 individuals

file <- system.file("csv/data_big.RDS", package = "lastdose")

big <- readRDS(file)

dim(big)

length(unique(big$ID))

Timing result

system.time(x2 <- lastdose(big))

Compare against the single profile

system.time(x1 <- lastdose(df))

x3 <- filter(x2, big[["ID"]]==1) %>% as.data.frame()

all.equal(x1,x3)

Observations prior to the first dose

When non-dose records happen prior to the first dose, lastdose calculates the time before the first dose (a negative value) for these records.

file <- system.file("csv/data2.csv", package = "lastdose")

df <- read_csv(file)

lastdose(df) %>% head()

The user can alternatively control what happens for these records

lastdose(df, fill = NA_real_, back_calc=FALSE) %>% head()

---

More info

See inst/doc/about.md for more details.

metrumresearchgroup/lastdose documentation built on April 12, 2025, 3:51 p.m.