nlmixr2autoinit (version 1.0.0)

getPPKinits: Automated pipeline for generating initial estimates in population PK models

Description

Provides a unified and fully automated workflow to generate initial pharmacokinetic and residual variability parameters for population PK models using concentration–time data from bolus, infusion, or oral administration.

Usage

getPPKinits(dat, control = initsControl(), verbose = TRUE)

Value

An object of class getPPKinits containing recommended initial parameter estimates, intermediate results, and computation diagnostics.

Arguments

dat

A data frame containing pharmacokinetic records in standard nlmixr2 format, including ID, TIME, EVID, and DV.

control

A list created by initsControl() specifying configuration for pooling, non-compartmental analysis, steady-state detection, fallback rules, statistical model components, and parameter selection metrics.

verbose

Logical (default = TRUE); when TRUE, displays key progress messages and stepwise updates during the initialization process. When FALSE, the function runs quietly without printing intermediate information.

Author

Zhonghui Huang

Details

The pipeline integrates four model-informed analytical components applied to raw or pooled concentration–time profiles:

1. Adaptive single-point methods

2. Naive pooled graphic methods

3. Naive pooled non-compartmental analysis (NCA) with optional Wagner–Nelson Ka calculation for oral dosing

4. Parameter sweeping across one-, two-, three-compartment and Michaelis–Menten models

In addition to structural PK parameters, the framework also initializes statistical model components:

• Inter-individual variability (IIV): pragmatic fixed \(\omega^2\) values are assigned to random effects.

• Residual unexplained variability (RUV): estimated either using a data-driven method based on trimmed residual summaries or a fixed-fraction approach consistent with NONMEM User Guide recommendations.

• Model applicability: the automated and model-informed strategy generates robust initial values suitable for both linear and nonlinear mixed-effects pharmacokinetic models.

See Also

initsControl, run_single_point, run_graphcal, run_pooled_nca, sim_sens_1cmpt_mm, sim_sens_2cmpt, sim_sens_3cmpt, metrics.

Examples

# \donttest{
## Bolus example
getPPKinits(Bolus_1CPT,verbose = TRUE)
## Oral example (run quietly)
getPPKinits(Oral_1CPT,verbose = FALSE)
# }