Create a list of PKPDsim covariates for modeling

Data sets (like NHANES) typically come in rectangular formats while for MIPD trials we model each individual separately. PKPDsim expects covariates to be provided as a list of PKPDsim covariates. This function helps convert rectangular (wide) data to model-ready covariates.

Usage

create_cov_object(dat_i, mapping, implementation = NULL, time_column = NULL)

Arguments

dat_i

a rectangular data set, containing only rows for an individual

mapping

named vector indicating which columns to extract, and what the covariate names are expected in the model. Names should be model covariates and values should be data frame column names.

implementation

named vector indicating covariate implementation, one of c("interpolate", "locf"). If NULL, assumes "interpolate" for all covariates. See PKPDsim::new_covariate for details. Names should be model covariates and values should be implementation method.

time_column

if covariates are time-varying, supply a column for time. Otherwise, initial value only is taken for all covariates.

Value

Returns a named list of PKPDsim covariates.

Details

The data is assumed to already be numeric (e.g., categorical covariates like smoking status have already been mapped to 1 or 0).

The function returns NULL if required columns are missing or NA.

Examples

dat <- data.frame(ID = 1, weight = 70, CRCL = 4)
mapping <- c(WT = "weight", CRCL = "CRCL")
create_cov_object(dat, mapping)
#> $WT
#>   value times      unit implementation
#> 1    70     0 undefined    interpolate
#> 
#> $CRCL
#>   value times      unit implementation
#> 1     4     0 undefined    interpolate
#>