Bringing in data dictionary labels to many dataframes

labelling_data_from_dictionary.R

# Read In Data Tables ----------------------------------------------------

raw_transfusion_data <- list(
  dti_untx = readr::read_csv(here(data_dir, "DTI_UnTx_V2.1.csv")),
  dti_tx = readr::read_csv(here(data_dir, "Transfused_DTI_50_V2.csv")),
  o2_vitals_untx = readr::read_csv(here(data_dir, "Vitals_O2_UnTx_V2.1.csv")),
  o2_vitals_tx = readr::read_csv(here(data_dir, "Transfused_02_Vitals_V2.csv")),
  blood_type_untx = readr::read_csv(here(data_dir, "ABO_UnTx_V2.csv")),
  blood_type_tx = readr::read_csv(here(data_dir, "Transfused_ABO_V2.1.csv")),
  encounter_untx = readr::read_csv(here(data_dir, "Encounter_UnTx_V2.csv")),
  encounter_tx = readr::read_csv(here(data_dir, "Transfused_Encounter_V2.csv"))
)


# Read in Data Dictionary -------------------------------------------------

data_dictionary <- list(
  dti = readxl::read_excel(here(data_dir, "Data Dictionary.xlsx"), sheet = 'DTI Table'),
  o2_vitals = readxl::read_excel(here(data_dir, "Data Dictionary.xlsx"), sheet = 'Vitals_O2 Table'),
  encounter = readxl::read_excel(here(data_dir, "Data Dictionary.xlsx"), sheet = 'Encounter Table'),
  blood_type = readxl::read_excel(here(data_dir, "Data Dictionary.xlsx"), sheet = 'ABO_Table'),
  icd = readxl::read_excel(here(data_dir, "Data Dictionary.xlsx"), sheet = 'ICD Table'),
  labs = readxl::read_excel(here(data_dir, "Data Dictionary.xlsx"), sheet = 'Labs Table'),
  patient = readxl::read_excel(here(data_dir, "Data Dictionary.xlsx"), sheet = 'Patient Table')
)



# apply data dictionary to columns ----------------------------------------

# a helper function to make a named vector out of the data dictionary
# descriptions -- an entry for each column
make_datadict_label_mapping <- function(dict_df) {
  dict_df |> select(`Column Name`, `Description`) |> 
    transmute(
      name = as.character(`Column Name`), 
      label = as.character(Description)) |> 
    deframe()
}

# a function to apply the column labels using labelled::var_label
apply_labels <- function(df, label_map) {
  common_colnames <- intersect(names(df), names(label_map))
  labelled::var_label(df[common_colnames]) <- label_map[common_colnames]
  return(df)
}

# create label maps
datadict_label_maps <- purrr::map(data_dictionary, make_datadict_label_mapping)

# apply the label maps using modify_at from purrr
raw_transfusion_data <- raw_transfusion_data |>
  modify_at(
    c("dti_untx", "dti_tx"),
    ~ apply_labels(.x, datadict_label_maps$dti)
  ) |>
  modify_at(
    c("o2_vitals_untx", "o2_vitals_tx"),
    ~ apply_labels(.x, datadict_label_maps$o2_vitals)
  ) |>
  modify_at(
    c("encounter_untx", "encounter_tx"),
    ~ apply_labels(.x, datadict_label_maps$encounter)
  ) |>
  modify_at(
    c("blood_type_untx", "blood_type_tx"),
    ~ apply_labels(.x, datadict_label_maps$blood_type)
  )

The end result is that all of the underlying tables have nice labels that appear when you View(raw_transfusion_data$dti_tx) in RStudio for example: