Perform a stepwise covariate modeling (SCM)

Perform a stepwise covariate modeling (SCM). Note currently only the forward inclusion is implemented (i.e. no backward deletion)

scm_IQRnlmeProject(
  projectPath,
  out_dir,
  relations,
  p_lev = 0.01,
  pp_fun = NULL,
  verbose = TRUE,
  clean = TRUE,
  gof = FALSE,
  Nparallel = 1,
  nlme = TRUE,
  ...
)

Arguments

projectPath

The directory of the IQR model to be used for the reference

out_dir

The directory to be used for the SCM outputs

relations

A tibble with the following mandatory columns: cov for the covariate name, prm for the parameter name, cont a TRUE/FALSE flag to indicate wether the covariate is continuous.An optional column value can be provided to set the intial estimate for the covariate effect, if not provided 0.01 will be used instead. OFV.

p_lev

The significance level to use for the covariate selection. Should be a number between 0 and 1.

pp_fun

A post processing function for the model repository. This function should only have a single argument. The input to this function will be the the path to a model directory.

verbose

A logical flag indicating wether informative messages should be printed to the console

clean

A logical flag indicating wether the scm process cleanup the provided dir upon start.

gof

Logical flag indicating whether GOF plots should be produced or not (default is FALSE).

Nparallel

The number of models run in parallel (default is 1).

nlme

TRUE for NONMEM/MONOLIX, or FALSE for SYSFIT.

...

Options to be passed to the IQRnlmeProject function (e.g. K1, K2). Currently only the tool, the toolversion, and the estimation method are automatically applied from the original run. If specific settings were used in the original run it is best to re-redefine them here.

See also

Other IQRnlmeProject: IQRnlmeEst(), IQRnlmeProject(), addCovariateToModelSpec_IQRest(), addPar_modelSpec(), as_IQRnlmeProject(), as_IQRnlmeProjectMulti(), bootstrap_IQRnlmeProject(), compareModels_IQRnlmeProjectMulti(), convertETAINDIVPRED_IQRnlmeProject(), covariateEffect_IQRnlmeProject(), data_IQRest(), dosing_IQRest(), duplicate_IQRnlmeProject(), eigenvalues_IQRnlmeProject(), exportVirtualSubjects_IQRnlmeProject(), getData_IQRnlmeProject(), getETAs_IQRnlmeProject(), getEst_IQRnlmeProject(), getHeader_IQRnlmeProject(), getIndivParameters_IQRnlmeProject(), getIndivPredictions_IQRnlmeProject(), getModel_IQRnlmeProject(), getObsPred_IQRnlmeProject(), getPopParameters_IQRnlmeProject(), getResults_IQRnlmeProject(), getResults_IQRnlmeProjectMulti(), hasrun_IQRnlmeProject(), hasrun_IQRnlmeProjectMulti(), informationContent_IQRnlmeProject(), is_IQRnlmeEst(), is_IQRnlmeProject(), is_IQRnlmeProjectMulti(), is_MONOLIX_IQRnlmeProject(), is_NLMIXR_IQRnlmeProject(), is_NONMEM_IQRnlmeProject(), modelSpec_IQRest(), outlier_IQRnlmeProject(), plot.IQRnlmeProject(), plot.IQRnlmeProjectMulti(), plotBLQVPC_IQRdataVPC(), plotConvergence_IQRnlmeProject(), plotETACOV_IQRnlmeProject(), plotETA_IQRnlmeProject(), plotGOF_IQRnlmeProject(), plotINDIVSIM_IQRnlmeProject(), plotINDIV_IQRnlmeProject(), plotMEANSIM_IQRnlmeProject(), plotVPC_IQRdataVPC(), pred_IQRnlmeProject(), print.IQRnlmeEst(), print.IQRnlmeProject(), print.IQRnlmeProjectMulti(), print_modelSpec(), run_IQRnlmeProject(), run_IQRnlmeProjectMulti(), sample_IQRnlmeProject(), summary.IQRnlmeProject(), summary.IQRnlmeProjectMulti(), summaryComments_IQRnlmeProjectMulti(), summaryCorrelations_IQRnlmeProjectMulti(), summaryCovariates_IQRnlmeProjectMulti(), summaryParameters_IQRnlmeProjectMulti(), vpc_IQRnlmeProject()

Author

Benjamin Guiastrennec, IntiQuan GmbH

Examples

if (FALSE) { # \dontrun{
# Creating the relation object
relations <- tidyr::crossing(
  cov = c("AGE", "WTKG", "RACE", "SEXF"),
  prm = c("CL", "VC")
) %>%
  arrange(prm, cov) %>%
  mutate(cont = ifelse(cov %in% c("AGE", "WTKG"), TRUE, FALSE))

# Running the SCM
scm_IQRnlmeProject(
  projectPath = "../Models/01_base_model/",
  out_dir = "../Models/02_covariate_selection",
  relations = relations,
  p_lev = 0.05,
  nlme = "NONMEM"
)
} # }