RxODE
Facilities for Simulating from ODE-Based Models

Package index
Search the RxODE package
Vignettes
Functions
641
Source code
185
Man pages
155
Home
/
CRAN
/
RxODE
/
RxODE events"

RxODE events"
In RxODE: Facilities for Simulating from ODE-Based Models 

library(learnr)
library(RxODE)
library(ggplot2)
## Adapted from gradethis::grade_result for now
## Wait for gradethis to make it to CRAN
## Model from RxODE tutorial
m1 <-RxODE({
    KA=2.94E-01;
    CL=1.86E+01;
    V2=4.02E+01;
    Q=1.05E+01;
    V3=2.97E+02;
    Kin=1;
    Kout=1;
    EC50=200;
    ## Added modeled bioavaiblity, duration and rate
    fdepot = 1;
    durDepot = 8;
    rateDepot = 1250;
    C2 = centr/V2;
    C3 = peri/V3;
    d/dt(depot) =-KA*depot;
    f(depot) = fdepot
    dur(depot) = durDepot
    rate(depot) = rateDepot
    d/dt(centr) = KA*depot - CL*C2 - Q*C2 + Q*C3;
    d/dt(peri)  =                    Q*C2 - Q*C3;
    d/dt(eff)  = Kin - Kout*(1-C2/(EC50+C2))*eff;
    eff(0) = 1
});

From Rstudio 1.3+ tutorial pane you can use the expand button () to navigate tutorial sections.

Creating events

Model for Event discussion

For these models, we can illustrate by using the model shared in the RxODE tutorial; Compile it so we can continue:

## Model from RxODE tutorial
m1 <-RxODE({
    KA=2.94E-01;
    CL=1.86E+01;
    V2=4.02E+01;
    Q=1.05E+01;
    V3=2.97E+02;
    Kin=1;
    Kout=1;
    EC50=200;
    ## Added modeled bioavaiblity, duration and rate
    fdepot = 1;
    durDepot = 8;
    rateDepot = 1250;
    C2 = centr/V2;
    C3 = peri/V3;
    d/dt(depot) =-KA*depot;
    f(depot) = fdepot
    dur(depot) = durDepot
    rate(depot) = rateDepot
    d/dt(centr) = KA*depot - CL*C2 - Q*C2 + Q*C3;
    d/dt(peri)  =                    Q*C2 - Q*C3;
    d/dt(eff)  = Kin - Kout*(1-C2/(EC50+C2))*eff;
    eff(0) = 1
});

print(summary(m1))

What is an event table

An event table in RxODE is a specialized data frame that acts as a container for all of RxODE's events and observation times. While not required to solve, the event table allows easy generation of a data table in the format that RxODE expects.

To create an RxODE event table you may use the code eventTable(), et(), or even create your own data frame with the right event information contained in it. This is closely related to the types of events that RxODE supports.

Once the event table has been created you can add sampling/observations or doses by piping or direct access.

This is a short table of arguments the two main functions (et() and add.dosing():

| add.dosing() | et() | Description | |-----------------|------|-------------------------------------| | dose | amt | Dose/Rate/Duration amount | | nbr.doses | addl | Additional doses or number of doses | | dosing.interval | ii | Dosing Interval | | dosing.to | cmt | Dosing Compartment | | rate | rate | Infusion rate | | start.time | time | Dosing start time | | | dur | Infusion Duration |

Sampling times can be added with add.sampling( sampling times ) or et( sampling times ). Dosing intervals and sampling windows are also supported.

Exercise: Create an empty table and add a 100 mg dose

e <- et()## you can add the units by `amount.units` and `time.units`

print(e)
You may add a dose by `$add.dosing(dose=...)` or `e %>% et(amt=...)`

Solving the system

With the model built from RxODE, and the event table built above, we can solve and plot the scenario; Run the code and see if you notice anything:

ev <- et() %>% 
    et(dose=5000, dosing.interval=12, until=48)

rxSolve(m1, ev) %>% plot(C2)

Quiz 1

Lets see if you can recall everything so far:

quiz(
  question("What function(s) can create event table for solving in RxODE?",
    answer("`data.frame()`", correct=TRUE),
    answer("`eventTable()`", correct=TRUE),
    answer("`et()`", correct=TRUE),
    answer("ev()", correct = FALSE),
    answer("add.sampling()", correct=FALSE),
    answer("add.dosing()", correct=FALSE)
  ),
  question("What did you notice about the event table plot?",
           answer("There was a missing dose at the end of the plot"), 
           answer("The samples times were chose by RxODE", correct=TRUE))
)

Adding sampling

In the last example you may have noticed that the sample times were chosen by RxODE; This is because the event table that was supplied to the solving routine only included dosing information. If you add observations to the event table you can control the observations of each event.

Recall RxODE you can use add.sampling( samples ) or et( samples ) to add observations to an RxODE event table.

Exercise, add observations for the first 24 hours

As an exercise, add an observation for the first 24 hours every 0.5 hours:

ev <- et() %>% 
    et(dose=5000, ii=12, until=48)

rxSolve(m1, ev) %>% plot(C2)

ev <- et() %>% 
    et(dose=5000, ii=12, until=48) %>%
    et(seq(0, 24, by=0.5))

rxSolve(m1, ev) %>% plot(C2)

et convenience notation

If you are like me, you may have used seq() as your solution to adding observations to the event table. For the function et() you can simply drop the seq() and enclose everything in the et() function. For et(seq(0,24,by=0.5)) this can be written as et(0,24,by=0.5).

As an exercise change the event table below to reflect this convenience:

ev <- et() %>% 
    et(dose=5000, ii=12, until=48) %>%
    et(seq(0, 24, by=0.5))

rxSolve(m1, ev) %>% plot(C2)

ev <- et() %>% 
    et(dose=5000, ii=12, until=48) %>%
    et(seq(0, 24, by=0.5))

rxSolve(m1, ev) %>% plot(C2)

Event classification

Here are the legal entries to a RxODE event table (which can be a tibble, data.frame, data.table or RxODE event table):

| Column | Meaning | Notes | |-----------|-----------------------|------------------------------------------------------------------------------| | id | Individual identifier | Will convert to a factor/integer (sorted) | time | Individual time | For each ID must be ascending and non-negative | | amt | dose amount | Positive for doses zero/NA for observations | | rate | infusion rate | When specified the infusion duration will be dur=amt/rate | | | | rate = -1, rate modeled; rate = -2, duration modeled | | dur | infusion duration | When specified the infusion rate will be rate = amt/dur | | evid | event ID | 0=Observation; 1=Dose; 2=Other; 3=Reset; 4=Reset+Dose; 5=Replace; 6=Multiply | | cmt | Compartment | Represents compartment #/name for dose/observation | | ss | Steady State Flag | 0 = non-steady-state; 1=steady state; 2=steady state +prior states | | ii | Inter-dose Interval | Time between doses. | | addl | # of additional doses | Number of doses like the current dose. |

Quiz -- Legal event names

What are some of the legal event names:

quiz(
  question("What are legal names for event columns:",
           answer("steadyState", correct=FALSE),
           answer("ii", correct=TRUE), 
           answer("dosingInterval", correct=FALSE),
           answer("rate", correct=TRUE),
           answer("dur", correct = TRUE),
           answer("ss", correct=TRUE)))

Event differences

If you are familiar with NONMEM or Monolix, you may notice that RxODE event tables are very similar to the NONMEM/Monolix dataset convention with these exceptions:

Event ID (evid quiz)

Now that we covered all the event types, quick quiz on what they all mean:

quiz(
  question("What does evid=1 mean?",
           answer("An observation event", correct=FALSE),
           answer("A reset event", correct=FALSE),
           answer("A reset then dose", correct=FALSE), 
           answer("A multiplication event", correct=FALSE),
           answer("An addivite event (adding say one dose)", correct=TRUE),
           answer("Other type of event", correct=FALSE),
           answer("Replacement event", correct=FALSE),
           answer("Nothing; illegal evid", correct=FALSE)),
  question("What does evid=0 mean?",
           answer("An observation event", correct=TRUE),
           answer("A reset event", correct=FALSE),
           answer("A reset then dose", correct=FALSE), 
           answer("A multiplication event", correct=FALSE),
           answer("An addivite event (adding say one dose)", correct=FALSE),
           answer("Other type of event", correct=FALSE),
           answer("Replacement event", correct=FALSE),
           answer("Nothing; illegal evid", correct=FALSE)),
  question("What does evid=2 mean?",
           answer("An observation event", correct=FALSE),
           answer("A reset event", correct=FALSE),
           answer("A reset then dose", correct=FALSE), 
           answer("A multiplication event", correct=FALSE),
           answer("An addivite event (adding say one dose)", correct=FALSE),
           answer("Other type of event", correct=TRUE),
           answer("Replacement event", correct=FALSE),
           answer("Nothing; illegal evid", correct=FALSE)), 
  question("What does evid=3 mean?",
           answer("An observation event", correct=FALSE),
           answer("A reset event", correct=TRUE),
           answer("A reset then dose", correct=FALSE), 
           answer("A multiplication event", correct=FALSE),
           answer("An addivite event (adding say one dose)", correct=FALSE),
           answer("Other type of event", correct=FALSE),
           answer("Replacement event", correct=FALSE), 
           answer("Nothing; illegal evid", correct=FALSE)),
  question("What does evid=4 mean?",
           answer("An observation event", correct=FALSE),
           answer("A reset event", correct=FALSE),
           answer("A reset then dose", correct=TRUE), 
           answer("A multiplication event", correct=FALSE),
           answer("An addivite event (adding say one dose)", correct=FALSE),
           answer("Other type of event", correct=FALSE),
           answer("Replacement event", correct=FALSE), 
           answer("Nothing; illegal evid", correct=FALSE)),
  question("What does evid=5 mean?",
           answer("An observation event", correct=FALSE),
           answer("A reset event", correct=FALSE),
           answer("A reset then dose", correct=FALSE), 
           answer("A multiplication event", correct=FALSE),
           answer("An addivite event (adding say one dose)", correct=FALSE),
           answer("Other type of event", correct=FALSE),
           answer("Replacement event", correct=TRUE), 
           answer("Nothing; illegal evid", correct=FALSE)), 
  question("What does evid=6 mean?",
           answer("An observation event", correct=FALSE),
           answer("A reset event", correct=FALSE),
           answer("A reset then dose", correct=FALSE), 
           answer("A multiplication event", correct=TRUE),
           answer("An addivite event (adding say one dose)", correct=FALSE),
           answer("Other type of event", correct=FALSE),
           answer("Replacement event", correct=FALSE),
           answer("Nothing; illegal evid", correct=FALSE)),
  question("What does evid=7 mean?",
           answer("An observation event", correct=FALSE),
           answer("A reset event", correct=FALSE),
           answer("A reset then dose", correct=FALSE), 
           answer("A multiplication event", correct=FALSE),
           answer("An addivite event (adding say one dose)", correct=FALSE),
           answer("Other type of event", correct=FALSE),
           answer("Replacement event", correct=FALSE),
           answer("Nothing; illegal evid", correct=TRUE))
  )

Remembering event types

One of the challenges of flexible event types, is remembering the event identifiers; If you ever want to figure out what an event type is, it is print out in an RxODE event table.

In this exercise, print out event types 1 to 6 by creating an event table:

e <- et() %>%
    et(time=1, amt=1, evid=1) %>%
    et(time=2, amt=2, evid=2) %>%
    et(time=3, amt=3, evid=3) %>%
    et(time=4, amt=4, evid=4)

## add evid=5 through 6 to the above event table

print(e)

e <- et() %>%
    et(time=1, amt=1, evid=1) %>%
    et(time=2, amt=2, evid=2) %>%
    et(time=3, amt=3, evid=3) %>%
    et(time=4, amt=4, evid=4) %>% 
    et(time=5, amt=5, evid=5) %>%
    et(time=6, amt=6, evid=6)
## add evid=5 through 6 to the above event table

print(e)

Convert to data.frame

Notice that the evid shows #:Description, this is only for display; When you convert the event table to a data.frame, it becomes a number.

Use as.data.frame to convert the event table e to an data.frame and print-out the results:

e <- et() %>%
    et(time=1, amt=1, evid=1) %>%
    et(time=2, amt=2, evid=2) %>%
    et(time=3, amt=3, evid=3) %>%
    et(time=4, amt=4, evid=4) %>%
    et(time=6, amt=6, evid=6)

print(e)

e <- et() %>%
    et(time=1, amt=1, evid=1) %>%
    et(time=2, amt=2, evid=2) %>%
    et(time=3, amt=3, evid=3) %>%
    et(time=4, amt=4, evid=4) %>%
    et(time=6, amt=6, evid=6) %>%
    as.data.frame

print(e)

Bolus/Additive Doses

A bolus dose is the default type of dose in RxODE and only requires the amt/dose. Note that this uses the convenience function et() described in the RxODE event tables

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12,until=24) %>%
    et(seq(0, 24, length.out=100))

print(ev)

rxSolve(m1, ev) %>% plot(C2) +
    xlab("Time")

Infusions

There are a few different type of infusions that RxODE supports:

Constant Infusion (in terms of duration and rate)

The next type of event is an infusion; There are two ways to specify an infusion; The first is the dur keyword.

An example of this is:

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12,until=24, dur=8) %>%
    et(seq(0, 24, length.out=100))

print(ev)

rxSolve(m1, ev) %>% plot(depot, C2) +
    xlab("Time")

It can be also specified by the rate component:

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12,until=24, rate=10000/8) %>%
    et(seq(0, 24, length.out=100))

print(ev)

rxSolve(m1, ev) %>% plot(depot, C2) +
    xlab("Time")

Did you spot a difference in the event tables?

If you have a keen eye you may have noticed the two event tables. The first had a rate column and the second has a dur column. Keep this in mind as we continue our infusion discussion.

What does bioavailability do to the system?

Now that we have infusion, a natural question is how does bioavailability affect the RxODE system?

Let check; Change the bioavailability to fdepot=0.25 when the event table is parameterized in terms of rate to see what happens (you may have to run the code then change it to see the difference):

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12,until=24, rate=10000/8) %>%
    et(seq(0, 24, length.out=100))

## Change the fdepot to 0.25
rxSolve(m1, ev, c(fdepot=1)) %>% plot(depot, C2) +
    xlab("Time")

Bioavailability change for durations

In the case of modeling rate, a bioavailability decrease, decreases the infusion duration, as in NONMEM.

Conversely, a bioavailability increase should therefore increase the infusion duration; A simple check could be to change the fdepot to 2:

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12,until=24, rate=10000/8) %>%
    et(seq(0, 24, length.out=100))

## Change the fdepot to 2
## If you cannot clearly see the effect, continue to the next section
rxSolve(m1, ev, c(fdepot=1)) %>% plot(depot, C2) +
    xlab("Time")

Rationale for infusion changes when rate is specified

The rationale for this behavior is that the rate and amt are specified by the event table, so the only thing that can change with a bioavailability increase is the duration of the infusion.

However a bioavailability change may not affect the infusion duration of an IV infusion, so this thinking about IV infusion bioavilability has led to confusion in the past.

This is the reason why parameterizing using the duration can be useful.

Exercise -- parameterize using duration

If you specify the amt and dur components in the event table, bioavailability changes affect the rate of infusion.

Note that if there is no bioavialability change the solution is the same; Change the bioavilability to see the effect:

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12,until=24, dur=8) %>%
    et(0, 24, length.out=100)

## If you cannot see a clear difference, go to the next section
rxSolve(m1, ev, c(fdepot=1)) %>% plot(depot, C2) +
    xlab("Time")

Duration effect

Did you figure out the effect? It can be hard to notice; the y-axis changed; To more clearly see the effect you can look at the side-by-side comparison; Run the following code to see the difference:

library(ggplot2)
library(gridExtra)

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12,until=24, dur=8) %>%
    et(0, 24, length.out=100)

p1 <- rxSolve(m1, ev, c(fdepot=1.25)) %>% plot(depot) +
    xlab("Time") + ylim(0,5000)

p2 <- rxSolve(m1, ev, c(fdepot=0.25)) %>% plot(depot) +
    xlab("Time")+ ylim(0,5000)

grid.arrange(p1,p2, nrow=1)

Now you can see the clear changes in infusion rate when dur and amt is constant. This is more likely what would happen in clinical practice; The duration of infusion is the same, but the rate is changed.

Modeled rate

You may also model rate. This is equivalent to NONMEM's rate=-1 and is how RxODE's event table specifies the data item as well. You can also use rate=model as a mnemonic. Note that you chan change the rateDepot to change the rate of the infusion. Also fdepot in this case changes the duration of infusion.

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12,until=24, rate=model) %>%
    et(seq(0, 24, length.out=100))

print(ev)

rxSolve(m1, ev, c(rateDepot=10000/3, fdepot=1)) %>% plot(depot, C2) +
    xlab("Time")

Note that the event table gives you clues about the type of event rate=-1 is by the text. As before, this is simply a displayed output and the actual value is -1

Modeled duration

You can model the duration, which is equivalent to NONMEM's rate=-2. As a mnemonic you can use the dur=model instead of rate=-2

As an exercise you can change the duration of infusion (durDepot) and the fdepot. The durDepot will change the duration of the infusion; The fdepot will change the infusion rate:

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12,until=24, dur=model) %>%
    et(seq(0, 24, length.out=100))

print(ev)

rxSolve(m1, ev, c(durDepot=7, fdepot =1)) %>% plot(depot, C2) +
    xlab("Time")

Steady State dosing

These doses are solved until a steady state is reached with a constant inter-dose interval. There are three types of steady state doses that RxODE supports:

The RxODE event table needs to be setup specially for each type of steady state record.

Simple Steady State dosing

Simple steady-state dosing can be added by using the ss=1 flag coupled with the inter-dose interval flag ii. This is interpreted as the same bolus dose or infusion being applied separated by a dosing interval ii until the steady state is reached.

This particular example has a steady-state bolus dose separated by 12 hours; You may run it and see how it behaves.

As an exercise you can change it to a 30 minute duration infusion with dur=0.5; You can also explore the effect of the inter-dose interval ii on the steady state concentrations

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12, ss=1) %>%
    et(seq(0, 24, length.out=100))

print(ev)

rxSolve(m1, ev) %>% plot(depot, C2)

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12, ss=1, dur=0.5) %>%
    et(seq(0, 24, length.out=100))

rxSolve(m1, ev) %>% plot(depot, C2)

Simple steady state dosing observations

You probably observed the steady state solution did not start at zero, but was at a higher concentration, as expected. This simple way of specifying steady state works for both bolus doses and infusion doses; It also assumes a constant interval between doses;

But, what if the interval or dose was not constant; For example if you were dosing 100 mg in the morning and 150 mg in the evening. How would you get the steady state solution?

The answer is adding a bit of complexity to the steady state solutions

Complex steady state

By using the ss=2 flag, you can use the super-positioning principle in linear kinetics to get steady state nonstandard dosing (i.e. morning 100 mg vs evening 150 mg). This is done by:

This takes a full dose cycle to reach a true steady state solution;

As an exercise, run the example below and then change the evening dose to 200 mg:

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=24, ss=1) %>%
    et(time=12, amt=15000, ii=24, ss=2) %>%
    et(time=24, amt=10000, ii=24, addl=3) %>%
    et(time=36, amt=15000, ii=24, addl=3) %>%
    et(seq(0, 64, length.out=500))

library(ggplot2)

rxSolve(m1, ev,maxsteps=10000) %>% plot(C2) +
    annotate("rect", xmin=0, xmax=24, ymin=-Inf, ymax=Inf, alpha=0.2) +
    annotate("text", x=12.5, y=7, label="Initial Steady State Period") +
    annotate("text", x=44,   y=7, label="Steady State AM/PM dosing")

Infusion steady states

The last type of steady state that RxODE supports is steady-state constant infusion rate. This can be specified the same way as NONMEM, that is:

Note that rate=-2 where we model the duration of infusion doesn't make much sense since we are solving the infusion until steady state. The duration is specified by the steady state solution.

Also note that bioavailability changes on this steady state infusion also do not make sense because they neither change the rate or the duration of the steady state infusion. Hence modeled bioavailability on this type of dosing event is ignored.

As an exercise you can run this example and then change the rate to see how it affects the steady state solution; The steady state solution (on the top) should be the same as the long term infusion (on the bottom)

library(gridExtra)
ev <- et(timeUnits="hr") %>%
    ## The infusion amount is unknown, rather we are using as much as
    ## needed to get to steady state so `ss=0`
    et(amt=0, ss=1,rate=10000/8)

p1 <- rxSolve(m1, ev) %>% plot(C2, eff)


ev <- et(timeUnits="hr") %>%
    ## Use a very large dose so that the infusion reaches steady state
    et(amt=200000, rate=10000/8) %>%
    et(0, 250, length.out=1000)

p2 <- rxSolve(m1, ev) %>% plot(C2, eff)


grid.arrange(p1,p2, ncol=1)

Resetting or turning off Compartments

RxODE supports two different types of events that may be useful:

Resetting events

Reset events are implemented by evid=3 or evid=reset, for reset and evid=4 for reset and dose. The exercise below resets the dose 6 hours post dose.

Run the following example to see what happens 6 hours post-dose:

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12, addl=3) %>%
    et(time=6, evid=reset) %>% ## this is a mnemonic for evid=3
    et(seq(0, 24, length.out=100))

rxSolve(m1, ev) %>% plot(depot,C2, eff)

Instead of reset, what about reset and dose?

You can see all the compartments are reset to their initial values at time=6. The next dose start the dosing cycle over.

What if you used the reset and dose event evid=4 of amt=1000?

Adjust the last example to see what happens:

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12, addl=3) %>%
    et(time=6, evid=reset) %>% ## Change this to evid=4
    et(seq(0, 24, length.out=100))

rxSolve(m1, ev) %>% plot(depot,C2, eff)

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12, addl=3) %>%
    et(time=6, evid=4) %>% ## Change this to evid=4
    et(seq(0, 24, length.out=100))

rxSolve(m1, ev) %>% plot(depot,C2, eff)

Turning off compartments

In this case, the whole system is reset and the dose is given to the depot compartment. This related but not the same as turning off a compartment.

What is the difference between "resetting" the system and turning off compartments?

To find out try turning off the depot compartment; This is done by changing the reset event to evid=2 and turning off the compartment. This can be done by dosing either to the negative compartment, ie -# or dosing to the -cmtName compartment. In this case it can be -1 or -depot

As an exercise change the reset dose to turn off the depot compartment:

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12, addl=3) %>%
    et(time=6, evid=4) %>% ## turn off the compartment with evid=2 and cmt=?
    et(seq(0, 24, length.out=100))

rxSolve(m1, ev) %>% plot(depot,C2, eff)

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12, addl=3) %>%
    et(time=6, cmt="-depot", evid=2) %>%
    et(seq(0, 24, length.out=100))

rxSolve(m1, ev) %>% plot(depot,C2, eff)

Turning the compartment back on?

In this case, the depot is turned off; The depot compartment concentrations are set to the initial values but the other compartment concentrations/levels are not reset.

So what do you need to do to turn the compartment back on? Add another dose to the depot is administered the depot compartment is turned back on. This dose can be even a bolus of 0 just to turn the compartment back on.

Note: a dose to a compartment only turns back on the compartment that was dosed. Hence if you turn off the effect compartment, it continues to be off after another dose to the depot.

As an exercise, run the model without turning on the effect compartment; Then as an exercise turn the effect compartment back on at time=12:

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12, addl=3) %>%
    et(time=6, cmt="-eff", evid=2) %>%
    et(seq(0, 24, length.out=100))

rxSolve(m1, ev) %>% plot(depot,C2, eff)

ev <- et(timeUnits="hr") %>%
    et(amt=10000, ii=12, addl=3) %>%
    et(time=6, cmt="-eff", evid=2) %>%
    et(time=12,cmt="eff",evid=2) %>%
    et(seq(0, 24, length.out=100))

rxSolve(m1, ev) %>% plot(depot,C2, eff)

Result of turning on the effect compartment again

Once the effect compartment was turned back on, you can see a sharp increase in effect since the amount in the C2 starts to equilabrate with the effect compartment

Combining/Expanding event tables

In this section of the events tutorial we will cover:

Expanding the event table into a multi-subject event table

You may want to expand the event table into a multiple subject event table; et() will expand the event table by duplicating an event table for each individual specified in id=...

As an exercise, expand the solve to be 4 IDs

ev <- et(timeUnits="hr") %>%
    et(amt=10000, until = set_units(3, days), ii=12) %>% # loading doses
    et(seq(0,48,length.out=200)) %>%
    et(id=1:4)

print(ev)

set.seed(42)
rxSolve(m1, ev,
      params=data.frame(KA=0.294*exp(rnorm(4)), 18.6*exp(rnorm(4)))) %>%
    plot(C2)
You can add `et(id=1:4)` to the end of the event table pipline to solve for 4 individuals

Combining event tables

You may have noticed that the rxSolve will work if you add et(id=1:4) or if you do not; Where the expanding of individual IDs really shines is by combining arms in a study with different sampling times or sampling intervals. There the IDs will not have the exact same structure.

To combine event tables, you can use c, seq, rep, and rbind.

Sequencing event tables for complex dosing

One way to combine event table is to sequence them by c(), seq() or etSeq(). This takes the two dosing groups and adds at least one inter-dose interval between them:

etSeq() can combine two event tables or time intervals; As an exercise, change the event table to have qd dosing followed by bid dosing:

## bid for 5 days
bid <- et(timeUnits="hr") %>%
       et(amt=10000,ii=12,until=set_units(5, "days"))

## qd for 5 days
qd <- et(timeUnits="hr") %>%
      et(amt=20000,ii=24,until=set_units(5, "days"))

## bid for 5 days followed by qd for 5 days
## Change to qd for 5 days and bid for 5 days
et <- seq(bid,qd) %>% et(seq(0,11*24,length.out=100));

rxSolve(m1, et) %>% plot(C2)

Using seq to add a drug holiday

You can also add drug holidays using the etSeq or seq function. The way to do this is to simply add a time-period between the two dosing tables; For example

seq(eventTable1,separationTime,eventTable2)

Would combine eventTable1 add a separationTime and finally add eventTable2

Below is an example of a 3 day drug holiday; Change it to a week of drug holiday

## bid for 5 days
bid <- et(timeUnits="hr") %>%
       et(amt=10000,ii=12,until=set_units(5, "days"))

## qd for 5 days
qd <- et(timeUnits="hr") %>%
      et(amt=20000,ii=24,until=set_units(5, "days"))

## bid for 5 days followed by qd for 5 days
et <- seq(bid,set_units(3, day), qd) %>%
    et(seq(0,18*24,length.out=100));

rxSolve(m1, et) %>% plot(C2)
You can change `set_units(3, day)` to `set_units(1, week)`

Notes on adding the one-week drug holiday

Note that in this worked-out example, the time between the bid and the qd event tables is exactly one week, not 1 week plus 24 hours because of the inter-dose interval. If you want that behavior, you can sequence it using the wait="+ii".

Also note, that RxODE assumes that the dosing is what you want to space the event tables by, and clears out any sampling records when you combine the event tables. If that is not true, you can also use the option samples="use"

Repeating event tables

You can have an event table that you can repeat with etRep or rep.

For example 4 rounds of 2 weeks on QD therapy and 1 week off of therapy can be simply specified by the default exercise below.

As an exercise, change the drug holiday to 2 weeks

qd <-et(timeUnits = "hr") %>% et(amt=10000, ii=24, until=set_units(2, "weeks"), cmt="depot")

et <- rep(qd, times=4, wait=set_units(1,"weeks")) %>%
      add.sampling(set_units(seq(0, 12.5,by=0.005),weeks))

rxSolve(m1, et)  %>% plot(C2)

Combining event tables with rbind

You may combine event tables with rbind. This does not consider the event times when combining the event tables, but keeps them the same times. If you space the event tables by a waiting period, it also does not consider the inter-dose interval.

Using the previous seq you can clearly see the difference.

Please run the bid/qd example to have it fresh in you memory. After you have run it, change seq to rbind to see what changesr

## bid for 5 days
bid <- et(timeUnits="hr") %>%
       et(amt=10000,ii=12,until=set_units(5, "days"))

## qd for 5 days
qd <- et(timeUnits="hr") %>%
      et(amt=20000,ii=24,until=set_units(5, "days"))

et <- seq(bid,qd) %>%
    et(seq(0,18*24,length.out=500));

rxSolve(m1, et) %>% plot(C2)

rbind vs seq

As you noticed in the example rbind combined the event tables as they are, creating a double dose at the qd time and a single dose at the bid time. What about if you use a waiting period?

Using rbind with a waiting period

Still the waiting period applies (but does not consider the inter-dose interval).

As an exercise, change the waiting period to 3 days.

## bid for 5 days
bid <- et(timeUnits="hr") %>%
       et(amt=10000,ii=12,until=set_units(5, "days"))

## qd for 5 days
qd <- et(timeUnits="hr") %>%
      et(amt=20000,ii=24,until=set_units(5, "days"))


et <- rbind(bid,wait=set_units(10,days),qd) %>%
    et(seq(0,18*24,length.out=500));

rxSolve(m1, et) %>% plot(C2)

rbind with a waiting interval of 3 days

Notice that when waiting for 3 days, the days 3-5 duplicate the QD doses

Combining multi-subject event tables with rbind

You can also bind the tables together and make each ID in the event table unique; This can be good to combine cohorts with different expected dosing and sampling times. This requires the id="unique" option; Using the first example shows how this is different in this case:

## bid for 5 days
bid <- et(timeUnits="hr") %>%
       et(amt=10000,ii=12,until=set_units(5, "days"))

## qd for 5 days
qd <- et(timeUnits="hr") %>%
      et(amt=20000,ii=24,until=set_units(5, "days"))


## bid for 5 days
et <- etRbind(bid,qd, id="unique") %>%
    et(seq(0,150,length.out=500));

library(ggplot2)
rxSolve(m1, et) %>% plot(C2) + facet_wrap( ~ id)

Add samples/doses in a window

Dosing windows

In addition to adding fixed doses and fixed sampling times, you can have windows where you sample and draw doses from. For dosing windows you specify the time as an ordered numerical vector with the lowest dosing time and the highest dosing time inside a list.

In the initial example, the dosing is within a 6 hour dosing window; This can be specified by list(c(0,6));

For an exercise change the dosing windows to c(0,2):

set.seed(42)
ev <- et(timeUnits="hr") %>%
    et(time=list(c(0,6)), amt=10000, until = set_units(2, days), ii=12) %>% # loading doses
    et(id=1:4)

print(ev)
ev <- ev %>% et(seq(0,48,length.out=200))

solve(m1, ev, params=data.frame(KA=0.294*exp(rnorm(4)), 18.6*exp(rnorm(4)))) %>%
    plot(C2)

set.seed(42)
ev <- et(timeUnits="hr") %>%
    et(time=list(c(0,2)), amt=10000, until = set_units(2, days), ii=12) %>% # loading doses
    et(id=1:4) %>%
    et(seq(0,48,length.out=200))

solve(m1, ev, params=data.frame(KA=0.294*exp(rnorm(4)), 18.6*exp(rnorm(4)))) %>% plot(C2)

Sampling windows

The same sort of thing can be specified with sampling times. To specify the sampling times in terms of a sampling window, you can create a list of the sampling times. Each sampling time will be a two element ordered numeric vector.

set.seed(42)
ev <- et(timeUnits="hr") %>%
    et(time=list(c(0,2)), amt=10000, until = set_units(2, days), ii=12) %>% # loading doses
    et(id=1:4)

## Create 20 samples in the first 24 hours and 20 samples in the second 24 hours
samples <- c(lapply(1:20, function(...){c(0,24)}),
             lapply(1:20, function(...){c(20,48)}))

## Add the random collection to the event table
ev <- ev %>% et(samples)

library(ggplot2)
solve(m1, ev, params=data.frame(KA=0.294*exp(rnorm(4)), 18.6*exp(rnorm(4)))) %>% plot(C2) + geom_point()

This shows the flexibility in dosing and sampling that the RxODE event tables allow.

Units in event tables

As commented in the last exercise, you may specify units in the table event table. When specified, the units use the units package to keep track of the units and convert them if needed. Additionally, ggforce uses them to label the ggplot axes. The set_units and drop_units are useful to set and drop the RxODE event table units.

This example creates a data-set with doses in mg and time in hr.

Adding units in data-set and converting to different overall units

In this exercise use set_units to change the dose to g and the time to days:

ev <- et(amount.units="mg", time.units="hr") %>% 
    et(dose=5000, dosing.interval=12, until=48) %>%
    etExpand() ## Expand dosing so there is no ii/addl record

## Use set_units to change the  dose to g and time to days
print(ev)
Use `ev %>% set_units(g) %>% set_units(day)`

Dropping units

You may have used the units package in the past; In the case of RxODE event tables, you do not have to apply to the column(s) in the dataset, but RxODE applies the unit changes where it needs to apply them.

This is also true when dropping units in an event table. In this exercise, drop the units using drop_units from the event table dataset:

ev <- et(amount.units="mg", time.units="hr") %>% 
    et(dose=5000, dosing.interval=12, until=48) %>%
    etExpand() %>% ## Expand dosing so there is no ii/addl record
    set_units(day) %>% # change the units to day/grams
    set_units(g)

## Use drop_units to remove unit information
print(ev)
Use `drop_units()` on the `ev` dataset

Dropping units keeps the original transformation

If you simply dropped the units at the end of the prior example, you may have noticed that the event table ev did not revert to the units originally specified in the event table. It will keep the conversions completed before unit information is dropped.

You could also add units to the event table by simply using set_units.

In this example, try setting the units to mg and hr by the set_units function:

ev <- et() %>% 
    et(dose=5000, dosing.interval=12, until=48) %>%
    etExpand() ## Expand dosing so there is no ii/addl record

print(ev)
Use the `set_units(mg) %>% set_units(hr)` to change the units

Using deSolve data frames

Sometimes you may have a deSolve model you may wish to convert to RxODE to take advantage of its threading capabilities or pre-compiled models.

Using the deSolve's example for the event data frame and adapting it for RxODE you simply wrap the deSolve data frame in the et() function;

For example, using the model and event table the March 2020 deSolve's vignette, we can reproduce the results

rx <- RxODE({
    d/dt(v1) = -0.1 * v1
    d/dt(v2) = -0.1 * v2
    v1(0) = 1
    v2(0) = 2
})

eventdat <- data.frame(var = c("v1", "v2", "v2", "v1"), time = c(1, 1, 5, 9),
                       value = c(1, 2, 3, 4),
                       method = c("add", "mult", "rep", "add"))

print(eventdat)


## When we have a
et <- et(eventdat) %>% et(0, 10, by=0.1)

print(et)

## In this case we have to add the dosing records
rxSolve(rx, et, times=seq(0, 10, by=0.1)) %>% plot()


Try the RxODE package in your browser

Any scripts or data that you put into this service are public.

RxODE documentation built on March 23, 2022, 9:06 a.m.

R Package Documentation

rdrr.io home R language documentation Run R code online

Browse R Packages

CRAN packages Bioconductor packages R-Forge packages GitHub packages

We want your feedback!

Note that we can't provide technical support on individual packages. You should contact the package authors for that.
GitHub issue tracker
ian@mutexlabs.com
Personal blog

 
Embedding an R snippet on your website

Add the following code to your website.

For more information on customizing the embed code, read Embedding Snippets.

Close