Generate variability terms — generate

To enable comparison of multiple treatment conditions in a reproducible manner, it is recommended that interindividual variability terms and residual variability terms be generated prior to all analyses. This design also allows for resuming a simulation part-way through, when the random seed position may not be known.

Generate IIV for one or more individuals and one or more iterations per individual according to the supplied omega matrix.

Generate unexplained variability for one or more individuals and one or more iterations per individual according to the supplied proportional and additive error.

Usage

generate_iiv(
  sim_model,
  omega,
  parameters,
  ids = 1,
  n_iter = 1,
  seed = NULL,
  ...
)

generate_ruv(tdm_sample_time, prop, add, ids = 1, n_iter = 1, seed = NULL)

Arguments

sim_model: model used for simulated patient response ("truth").
omega: omega matrix, with covariance terms. See PKPDsim::sim for details.
parameters: simulation model parameters (population estimates), a named list.
ids: vector of ids, can be numeric or character.
n_iter: number of sets of individual parameters to generate per id
seed: set random seed
...: arguments passed on to PKPDsim::sim
tdm_sample_time: time of tdm, since start of treatment course (or other vector of identifiers to use for each tdm). For example, for three days of daily dosing and peak-trough sample collection, c(1, 24, 25, 48, 49, 73).
prop: proportional error
add: additive error

Value

generate_iiv a data frame with columns id (corresponding to ids), iter ( numbers 1 to n_iter) and columns for each individual parameter value.

generate_ruv returns a data frame with identifier columns of tdm_number, iteration, id, plus columns for proportional (prop) and additive (add) error.

Details

This family of functions generates variability terms to allow for reproducible analyses. Using multiple iterations per individual ID allows for PK variability within one set of covariates.

By default, generate_iivassumes a log-normal (exponential) distribution. See PKPDsim::sim documentation for the omega_type argument to provide finer grain control.

generate_ruv Assumes a normal distribution for proportional and additional error.