Declare a public structural model

The structural model and its typical parameter values are public inputs: they must be established without inspecting the confidential dataset. For a first-in-human compound they normally come from preclinical allometric scaling, which is also what selected the starting dose.

Usage

pmx_structural_model(
  pk,
  typical,
  pd = "none",
  source,
  rx = NULL,
  iiv = c(cl = 0.3, v = 0.2),
  residual_cv = 0.15
)

Arguments

pk

One of "1cmt_iv", "1cmt_oral", "1cmt_infusion", "2cmt_iv", "2cmt_oral".

typical

Named numeric vector of typical parameter values, interpreted as the median of a lognormal population. Requires cl and v (the central volume), plus ka for oral models and q and v2 for two-compartment models. Optional f defaults to 1. PD shapes additionally require baseline, plus slope for "linear" and plateau and rate for "exponential".

pd

The PD time course, with no exposure dependence. One of "none" (PK only), "constant", "linear" (needs slope), or "exponential" (needs plateau and rate, covering both decay and growth). A simple shape is adequate for exercising longitudinal code and calibrates through a well-conditioned level correction.

source

Required provenance string recording where the model and its typical values came from. Recorded in the release ledger.

rx

Reserved for an rxode2 model. Not yet implemented: the value is stored on the returned object but the generator always evaluates the built-in analytic solution, so supplying it warns.

iiv

Named vector of between-subject variability, as CV on the log scale. A public assumption; it consumes no privacy budget.

residual_cv

Proportional residual error, as a CV.

Value

A pmx_structural_model.