mattfidler
commented
3 months ago But I'm unable to access my_model as a function any longer:
That is because that it isn't a function. If you get back the functional form you can do the following:
library(rxode2)
#> rxode2 2.1.3.9000 using 8 threads (see ?getRxThreads)
#> no cache: create with `rxCreateCache()`
library(babelmixr2)
#> Loading required package: nlmixr2
#> Loading required package: nlmixr2data
my_model <- function() {
ini({
# Typical Value of System Parameters
TV_Vc = c(.Machine$double.eps, 1.0, Inf)
TV_Vt = c(.Machine$double.eps, 1.0, Inf)
TV_CL = c(.Machine$double.eps, 1.0, Inf)
TV_Q = c(.Machine$double.eps, 1.0, Inf)
TV_ka = c(.Machine$double.eps, 1.0, Inf)
TV_fb = c(.Machine$double.eps, 1.0, Inf)
# Error model parameters
prop_err = c(.Machine$double.eps, 0.1, .Machine$double.xmax)
add_err = c(.Machine$double.eps, 0.1, .Machine$double.xmax)
})
model({
# System Parameters
Vc = TV_Vc
Vt = TV_Vt
CL = TV_CL
Q = TV_Q
ka = TV_ka
fb = TV_fb
# Placeholders for infusion rates.
# These should be defined in the dataset.
# time scale: 1 units: (hours)
# mass scale: 1 units: (mg/hour)
Dinf = 0.0
# Defining ODEs
d/dt(At) = (-ka*At)
d/dt(Cc) = (ka*At*fb/Vc - CL/Vc*Cc -Q*(Cc-Cp)/Vc + Dinf/Vc)
d/dt(Cp) = (+Q*(Cc-Cp)/Vt)
# Outputs and error models
Cc_mg_ml = Cc
Cc_mg_ml ~ prop(prop_err) + add(add_err)
})
}
my_model = my_model |>
model({Vc = exp(TV_Vc + eta_Vc)})
#> ℹ add between subject variability `eta_Vc` and set estimate to 1
print(is.function(my_model))
#> [1] FALSE
my_model = my_model |>
as.function()
print(is.function(my_model))
#> [1] TRUECreated on 2024-08-03 with reprex v2.1.1
-
Evaluating a function creates a non-function object with class
rxUi. This is what is needed and used for piping, so any piping will return this object too. -
You can always convert back to a function with
as.function()though you would have to convert it back to accessrxUiproperties and functions like$nonmemModel
john-harrold
This is related to this issue:
https://github.com/nlmixr2/babelmixr2/issues/107
Consider this example here.
One option I've tried is the following:
The monolix model works fine but the NONMEM model fails with:
I think this is because I have no IIV in the model, so I add a IIV to one of the parameters:
But I'm unable to access my_model as a function any longer:
Now I try to access these as list elements:
Which seems to work, however the monolixModel element has a length of 3.
Can you tell me what I'm doing wrong?