createNmParSamples: Simulate parameter sets from a NONMEM run object

In MSToolkit: The MSToolkit library for clinical trial design

Description

Creates a list containing a set of "THETA", "OMEGA" and "SIGMA" parameter values, based on either the initial or final estimates of a NONMEM run

Usage

createNmParSamples(N, 
  run, 
  seed = .deriveFromMasterSeed(), 
	method = c("Covariance", "Final", "Initial")) 

Arguments

N	Number of samples to take
run	Either a NONMEM run, NONMEM control file, or NONMEM output file, as imported using RNMImport
seed	Random number seed
method	Parameter method: Covariance, Final or Initial

Details

Firstly, the function extracts the information needed to simulate. * If method is "Covariance", it extracts the final estimates and covariance matrix from the run * If method is "Final", it extracts the final estimates from the run * If method is "Initial", it extracts the initial estimates from the run

If method is either "Final" or "Initial", then the parameters selected are just repeated "N" times and returned as a single list structure (ie. no difference between parameter sets)

If method is "Covariance" and the covariance matrix can be extracted, samples are taken from a multivariate normal distribution (using the mvrnorm function in the MASS library). These samples are formatted in a list and returned

Value

A list of length "N", where each element of the list contains: * An element called "THETA", containing a vector of fixed effects * An element called "OMEGA", containing a matrix of between subject effects * An element called "SIGMA", containing a matrix of within subject effects

Author(s)

Mike K Smith [e-mail]

MSToolkit documentation built on May 2, 2019, 6:30 p.m.