Stepwise Parameter Modeling for QSP - random effects

Supports identification of random effects to consider in the estimation. The algorithm is based on reduction of OBJ by inclusion of IIV on a parameter. In each iteration single IIV elements are considered and the IIV is retained that results in the lowest objective function value. Then the next iteration is done. Currently limited to positive parameters only ("L" IIV).

spmIIV_IQRsysProject(path, model, dosing, data, POPvalues0,
  SPM_IIVvalues0 = NULL, IIVvalues0 = NULL, errorModel = NULL,
  ncores = 1, Nparallel = 1, FLAGclean = TRUE,
  SIMOPT.nauxtimes = 0, SIMOPT.hmax = NULL, SIMOPT.atol = 1e-06,
  SIMOPT.rtol = 1e-06)

Arguments

path

Path to where to store all the models in all the iterations

model

IQRmodel object to use

dosing

See dosing_IQRest

data

See data_IQRest

POPvalues0

Named vector with fixed effect parameters to use in the model

SPM_IIVvalues0

Named vector with random effect parameters to test in the model. If not provided the default is 50 percent CV for all parameters in POPvalues0

IIVvalues0

Named vector with random effect parameters to always have in the model

errorModel

See modelSpec_IQRsysEst

ncores

Number of cores for a single model run

Nparallel

Number of models to run in parallel

FLAGclean

If TRUE then all models are removed, keeping only the table result files - saves a lot of diskspace!

SIMOPT.nauxtimes

Number of additional auxiliary times for integrator

SIMOPT.hmax

Maximum stepsize of the integrator. Passed to IQRsysProject. Set if experiencing numerical issues with TIME offsets in the dataset.

SIMOPT.atol

Absolute error tolerance for integrator

SIMOPT.rtol

Relative error tolerance for integrator

Value

Results are displayed in the console and a list is returned with a table and a vector with parameters and values identified. In addition in the path and in each iteration folder a table with information is exported.

See also