Skip to contents

Create model

Source: R/create_model.R
create_model.Rd

This is essentially a wrapper around the model-creation and -modification functionality in pharmr/Pharmpy.

Usage

create_model(
  route = c("auto", "oral", "iv"),
  lag_time = FALSE,
  n_transit_compartments = 0,
  bioavailability = FALSE,
  n_cmt = 1,
  elimination = c("linear", "michaelis-menten"),
  iiv = "all",
  iiv_type = "exp",
  ruv = c("additive", "proportional", "combined", "ltbs"),
  covariates = NULL,
  data = NULL,
  name = NULL,
  estimation_method = c("foce", "saem"),
  estimation_options = list(),
  uncertainty_method = c("sandwich", "smat", "rmat", "efim", "none"),
  tool = c("nonmem", "nlmixr", "nlmixr2"),
  tables = c("fit"),
  auto_init = TRUE,
  auto_stack_encounters = TRUE,
  mu_reference = FALSE,
  settings = list(),
  verbose = FALSE
)

Arguments

route

route of administration, either oral or iv

lag_time

add a lag time, default is FALSE

n_transit_compartments

number of transit-compartments for absorption model. Default is 0.

bioavailability

Add a bioavailability parameter? Default is FALSE. Will add using a logit function.

n_cmt

number of elimination and distribution compartments. Default is 1, i.e. no peripheral distributions.

elimination

elimination type, either linear or michaelis-menten.

iiv

either character or a list object. If character, should be either "basic" (only CL and V parameters) or "all" (IIV on all parameters). If specified as a list object, it should contain the IIV magnitude (on SD scale) for parameters, e.g. list(CL = 0.2, V = 0.3).

ruv

one of proportional, additive, or combined.

data

data.frame as input to NONMEM / nlmixr.

name

name of model

estimation_method

estimation method.

estimation_options

options for estimation method, specified as list, e.g. NITER or ISAMPLE.

uncertainty_method

Compute uncertainty for parameter estimations. One of sandwich (default), smat, fmat, efim.

tool

output model type, either nonmem or nlmixr

tables

which pre-specified tables to add, defaults to parameters and fit tables.

auto_init

automatically update initial estimates to reasonable values based on a crude assessment of the PK data. Default is TRUE.

auto_stack_encounters

detects if TIME within an individual is decreasing from one record to another, which NONMEM cannot handle. If this happens, it will add a reset event (EVID=3) at that time, and increase the TIME for subsequent events so that NONMEM does not throw an error. It will increase the time for the next encounter to the maximum encounter length across all subjects in the dataset (rounded up to 100). If no decreasing TIME is detected, nothing will be done (most common case). This feature is useful e.g. for crossover trials when data on the same individual ispresent but is included in the dataset as time-after-dose and not actual time since first overall dose.

mu_reference

MU-reference the model, useful for SAEM estimation method.

settings

additional settings for model creation and model estimation. TBD

verbose

verbose output?

iiv_effect

either character or list. If character, one of c("exp", "add", "prop", "log", "re_log"). If list, should specify for each parameter the effect type, e.g. list(CL = "add", V = "exp"). Default is "exp" for all.