Generate harmonized dataset(s) and associated metadata

Source: R/02-harmo_process_harmonization.R

Reads a DataSchema and Data Processing Elements to generate a harmonized dossier from input dataset(s) in a dossier and associated metadata. The function has one argument that can optionally be declared by the user (unique_col_dataset). It refers to the columns which contains name of each harmonized dataset. These two columns are added to ensure that there is always a unique entity identifier when datasets are pooled.

harmo_process(
  object = NULL,
  dataschema = attributes(dossier)$`Rmonize::DataSchema`,
  data_proc_elem = attributes(dossier)$`Rmonize::Data Processing Elements`,
  harmonized_col_dataset = attributes(dossier)$`Rmonize::harmonized_col_dataset`,
  harmonized_col_id = attributes(dossier)$`Rmonize::harmonized_col_id`,
  .debug = FALSE,
  dossier = object
)

Arguments

object

Data frame(s) or list of data frame(s) containing input dataset(s).

dataschema

A DataSchema object.

data_proc_elem

A Data Processing Elements object.

harmonized_col_dataset

A character string identifying the column to use for dataset names.

harmonized_col_id

A character string identifying the name of the column present in every dataset to use as a participant identifier.

.debug

Allow user to test the inputs before processing harmonization.

dossier

[Deprecated]

Value

A list of data frame(s) containing harmonized dataset(s). The DataSchema and Data Processing Elements are preserved as attributes of the harmonized dossier.

Details

A dossier is a named list containing one or more data frames, which are input datasets. The name of each data frame in the dossier will be used as the name of the associated harmonized dataset produced by harmo_process().

A DataSchema is the list of core variables to generate across datasets and related metadata. A DataSchema object is a list of data frames with elements named 'Variables' (required) and 'Categories' (if any). The 'Variables' element must contain at least the name column, and the 'Categories' element must contain at least the variable and name columns to be usable in any function. In 'Variables' the name column must also have unique entries, and in 'Categories' the combination of variable and name columns must also be unique.

The Data Processing Elements specifies the input elements and processing algorithms to generate harmonized variables in the DataSchema formats. It is also contains metadata used to generate documentation of the processing. A Data Processing Elements object is a data frame with specific columns used in data processing: dataschema_variable, input_dataset, input_variables, Mlstr_harmo::rule_category and Mlstr_harmo::algorithm. To initiate processing, the first entry must be the creation of a harmonized primary identifier variable (e.g., participant unique ID).

Examples

{
# Use Rmonize_examples to run examples.

library(dplyr)
library(stringr)
library(lubridate)

# Perform data processing
dossier <- Rmonize_examples[str_detect(names(Rmonize_examples),"input_dataset_study")]
names(dossier) <- str_remove(names(dossier), "input_")
dataschema <- Rmonize_examples$`DataSchema`
data_proc_elem <- Rmonize_examples$`Data_Processing_Element_no errors`

harmonized_dossier <- harmo_process(
  dossier,
  dataschema,
  data_proc_elem,
  harmonized_col_dataset = 'adm_study_id')
  
glimpse(harmonized_dossier$dataset_study1)
}
#> 
#> Attaching package: 'lubridate'
#> The following objects are masked from 'package:base':
#> 
#>     date, intersect, setdiff, union
#> - Data Processing Elements: ------------------------------------------------
#> 
#> --harmonization of : dataset_study1 -----------------------------------
#>     processing 1/9 : adm_unique_id                id created
#>     processing 2/9 : adm_study_id                 complete
#>     processing 3/9 : sdc_age_m                    complete
#>     processing 4/9 : sdc_marital_m                complete
#>     processing 5/9 : lsb_alc_binge_m_preg         complete
#>     processing 6/9 : lsb_alc_binge_m_first_tri    impossible
#>     processing 7/9 : preg_gestational_age_del     complete
#>     processing 8/9 : sdc_sex_c                    complete
#>     processing 9/9 : pm_birthweight               complete
#> 
#> --harmonization of : dataset_study2 -----------------------------------
#>     processing 1/9 : adm_unique_id                id created
#>     processing 2/9 : adm_study_id                 complete
#>     processing 3/9 : sdc_age_m                    complete
#>     processing 4/9 : sdc_marital_m                complete
#>     processing 5/9 : lsb_alc_binge_m_preg         impossible
#>     processing 6/9 : lsb_alc_binge_m_first_tri    impossible
#>     processing 7/9 : preg_gestational_age_del     complete
#>     processing 8/9 : sdc_sex_c                    complete
#>     processing 9/9 : pm_birthweight               complete
#> 
#> --harmonization of : dataset_study3 -----------------------------------
#>     processing 1/9 : adm_unique_id                id created
#>     processing 2/9 : adm_study_id                 complete
#>     processing 3/9 : sdc_age_m                    complete
#>     processing 4/9 : sdc_marital_m                impossible
#>     processing 5/9 : lsb_alc_binge_m_preg         complete
#>     processing 6/9 : lsb_alc_binge_m_first_tri    impossible
#>     processing 7/9 : preg_gestational_age_del     complete
#>     processing 8/9 : sdc_sex_c                    complete
#>     processing 9/9 : pm_birthweight               complete
#> 
#> --harmonization of : dataset_study4 -----------------------------------
#>     processing 1/9 : adm_unique_id                id created
#>     processing 2/9 : adm_study_id                 complete
#>     processing 3/9 : sdc_age_m                    complete
#>     processing 4/9 : sdc_marital_m                complete
#>     processing 5/9 : lsb_alc_binge_m_preg         complete
#>     processing 6/9 : lsb_alc_binge_m_first_tri    complete
#>     processing 7/9 : preg_gestational_age_del     complete
#>     processing 8/9 : sdc_sex_c                    complete
#>     processing 9/9 : pm_birthweight               complete
#> 
#> --harmonization of : dataset_study5 -----------------------------------
#>     processing 1/9 : adm_unique_id                id created
#>     processing 2/9 : adm_study_id                 complete
#>     processing 3/9 : sdc_age_m                    complete
#>     processing 4/9 : sdc_marital_m                complete
#>     processing 5/9 : lsb_alc_binge_m_preg         **ERROR**
#>     processing 6/9 : lsb_alc_binge_m_first_tri    complete
#>     processing 7/9 : preg_gestational_age_del     complete
#>     processing 8/9 : sdc_sex_c                    complete
#>     processing 9/9 : pm_birthweight               complete
#> 
#> 
#> - CREATION OF HARMONIZED DATA DICTIONARY : --------------------------------
#> 
#> dataset_study1 : done
#> dataset_study2 : done
#> dataset_study3 : done
#> dataset_study4 : done
#> dataset_study5 : done
#> 
#> 
#> ------------------------------------------------------------------------------
#> 
#> The Data Processing Elements contain 'error' statuses. Dataset with any 'error' 
#> status will appear as empty in the harmonized dossier. When the harmonization 
#> statuses are finalized, remember to rerun the process with the updated 
#> Data Processing Elements and generate the final harmonized datasets.
#> 
#> 
#> Useful tip: If you identified errors and want to correct them later, you can specify
#> 'undetermined' in the column 'Mlstr_harmo::algorithm' of your problematic 
#> Data Processing Elements(s). Such Data Processing Elements will be ignored.
#> 
#> - WARNING MESSAGES (if any): ----------------------------------------------
#> 
#> Rows: 4,000
#> Columns: 9
#> $ adm_unique_id             <chr> "PID7216", "PID5380", "PID5531", "PID7612", …
#> $ adm_study_id              <chr+lbl> "1", "1", "1", "1", "1", "1", "1", "1", …
#> $ sdc_age_m                 <int> 27, 38, 27, 26, 33, 31, 32, 29, 29, 26, 33, …
#> $ sdc_marital_m             <int+lbl> 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1…
#> $ lsb_alc_binge_m_preg      <int+lbl> 0, 0, 0, 0, NA, 0, 0, 0, 0, 0, 0, 0, 0, …
#> $ lsb_alc_binge_m_first_tri <int+lbl> NA, NA, NA, NA, NA, NA, NA, NA, NA, NA, …
#> $ preg_gestational_age_del  <int> 41, 38, 39, 41, 40, 38, 38, 39, 41, 40, 41, …
#> $ sdc_sex_c                 <int+lbl> 1, 2, 1, 1, 1, 2, 1, 1, 1, 1, 1, 2, 1, 1…
#> $ pm_birthweight            <dbl> 2603.164, 3295.346, 3426.888, 3112.095, 2394…