Skip to contents

Create a minimalist model

Source: vignettes/v01_model_creation.Rmd
v01_model_creation.Rmd

Prerequisite

For this exercise, we’ll need the campsismod package. This package can be loaded as follows:

Create a minimalist model in the Notepad++ editor

Assume a very simple 1-compartment PK model with first-order eliminate rate K. Say this parameter has a typical value of log(2)/12≈0.06 (where 12 is the elimination half life) and has 15% CV. Let’s also initiate the central compartment to 1000.

This can be translated into the following Campsis model (

download Notepad++ plugin for Campsis

):

Minimalist model edited in Notepad++
Minimalist model edited in Notepad++

Let’s now create our theta.csv with our single parameter K as follows:

Edition of THETA’s in Excel
Edition of THETA’s in Excel

And finally, let’s also create our omega.csv to include inter-individual variability on K:

Edition of OMEGA’s in Excel
Edition of OMEGA’s in Excel

This model can now be loaded by campsismod…

model <- read.campsis("resources/minimalist_model/")
## Warning in read.allparameters(folder = folder): No file 'sigma.csv' could be
## found.

Let’s simulated this model in Campsis:

library(campsis)
dataset <- Dataset(25) %>% add(Observations(seq(0,24,by=0.5)))
results <- model %>% simulate(dataset=dataset, seed=1)
spaghettiPlot(results, "A_CENTRAL")

Create the same model programmatically

The same model can be created programmatically. First, let’s create an empty Campsis model.

model <- CampsisModel()

Then, let’s define the equation of our model parameter K.

model <- model %>% add(Equation("K", "THETA_K*exp(ETA_K)"))

We can add an ordinary differential equation as follows:

model <- model %>% add(Ode("A_CENTRAL", "-K*A_CENTRAL"))

We can init the central compartment as well on the fly:

model <- model %>% add(InitialCondition(compartment=1, "1000"))

Finally, let’s define our THETA_K and ETA_K:

model <- model %>% add(Theta("K", value=0.06))
model <- model %>% add(Omega("K", value=15, type="cv%"))

This model can simulated by Campsis as well. Powerful, isn’t it?

library(campsis)
dataset <- Dataset(25) %>% add(Observations(seq(0,24,by=0.5)))
results <- model %>% simulate(dataset=dataset, seed=2)
spaghettiPlot(results, "A_CENTRAL")