README

Does my cohort picked the correct number patients? Am I calculating an intersection in the right way? Is that the expected value for treatment duration? It just takes one incorrect parameter to get incoherent results in a pharmacoepidemiological study, and it is very challenging to test calculations on huge and complex databases.

That is why TestGenerator is useful to push a small sample of patients to unit test a study on the OMOP-CDM. It includes tools to create a blank CDM with a complete vocabulary and check if the code is doing what we expect in very specific cases.

Installation

# CRAN version

install.packages("TestGenerator")

Example

The user can provide an Excel file (link to sample) or a set of CSV files that represent tables of the OMOP-CDM, with a micro population of just 8 patients for testing purposes.

readPatients() will read either Excel or CSVs, and then saves the data in a JSON file. This is useful if the user wants to create more than one Unit Test Definitions. If the parameter outputPath is NULL The files are saved in the testthat/testCases folder of the package. Alterna

TestGenerator::readPatients(filePath = "~/pathto/testPatients.xlsx",
                            testName = "test",
                            outputPath = NULL,
                            cdmVersion = "5.3")

Alternatively, the user can use the functions readPatients.xl or readPatients.csv directly.

TestGenerator::readPatients.xl(filePath = "~/pathto/testPatients.xlsx",
                               testName = "test",
                               outputPath = NULL,
                               cdmVersion = "5.3")

TestGenerator::readPatients.csv(filePath = "~/pathto/csv/files",
                                testName = "test",
                                outputPath = NULL,
                                cdmVersion = "5.3",
                                reduceLargeIds = FALSE)

patientCDM() pushes one of those Unit Test Definitions into a blank CDM reference with a complete version of the vocabulary. If the pathJSON parameter is NULL, TestGenerator will look for the JSON test files in the testthat/testCases folder.

cdm <- TestGenerator::patientsCDM(pathJson = NULL, 
                                  testName = "test",
                                  cdmVersion = "5.3")

Now the user has a CDM reference with a complete vocabulary and just 8 patients.

filePath <- system.file("extdata/icu_sample_population.xlsx", 
                        package = "TestGenerator")

outputPath <- file.path(tempdir(), "test")

dir.create(outputPath)

TestGenerator::readPatients(filePath = filePath, 
                            testName = "test", 
                            outputPath = outputPath, 
                            cdmVersion = "5.3")
#> ✔ Unit Test Definition Created Successfully: 'test'

cdm <- TestGenerator::patientsCDM(pathJson = outputPath, 
                                  testName = "test",
                                  cdmVersion = "5.3")
#> ! cdm name not specified and could not be inferred from the cdm source table
#> ✔ Patients pushed to blank CDM successfully


cdm[["person"]] %>% glimpse()
#> Rows: ??
#> Columns: 18
#> Database: DuckDB v0.9.1 [cbarboza@Windows 10 x64:R 4.3.1/C:\Users\cbarboza\AppData\Local\Temp\Rtmp44K6CT\file6a812027396.duckdb]
#> $ person_id                   <int> 1, 2, 3, 4, 5, 6, 7, 8
#> $ gender_concept_id           <int> 8532, 8507, 8532, 8507, 8532, 8507, 8532, …
#> $ year_of_birth               <int> 1980, 1990, 2000, 1980, 1990, 2000, 1980, …
#> $ month_of_birth              <int> NA, NA, NA, NA, NA, NA, NA, NA
#> $ day_of_birth                <int> NA, NA, NA, NA, NA, NA, NA, NA
#> $ birth_datetime              <dttm> NA, NA, NA, NA, NA, NA, NA, NA
#> $ race_concept_id             <int> 0, 0, 0, 0, 0, 0, 0, 0
#> $ ethnicity_concept_id        <int> 0, 0, 0, 0, 0, 0, 0, 0
#> $ location_id                 <int> 0, 0, 0, 0, 0, 0, 0, 0
#> $ provider_id                 <int> 0, 0, 0, 0, 0, 0, 0, 0
#> $ care_site_id                <int> 0, 0, 0, 0, 0, 0, 0, 0
#> $ person_source_value         <chr> "0", "0", "0", "0", "0", "0", "0", "0"
#> $ gender_source_value         <chr> "M", "F", "M", "F", "M", "F", "M", "F"
#> $ gender_source_concept_id    <int> NA, NA, NA, NA, NA, NA, NA, NA
#> $ race_source_value           <chr> NA, NA, NA, NA, NA, NA, NA, NA
#> $ race_source_concept_id      <int> NA, NA, NA, NA, NA, NA, NA, NA
#> $ ethnicity_source_value      <chr> NA, NA, NA, NA, NA, NA, NA, NA
#> $ ethnicity_source_concept_id <int> NA, NA, NA, NA, NA, NA, NA, NA

test_cohorts <- system.file("extdata",
                            "test_cohorts",
                            package = "TestGenerator")

cohort_set <- CDMConnector::readCohortSet(test_cohorts)
  
cdm <- CDMConnector::generate_cohort_set(cdm,
                                         cohort_set,
                                         name = "test_cohorts")
#> ℹ Generating 3 cohorts
#> ℹ Generating cohort (1/3) - diazepam✔ Generating cohort (1/3) - diazepam [345ms]
#> ℹ Generating cohort (2/3) - hospitalisation✔ Generating cohort (2/3) - hospitalisation [309ms]
#> ℹ Generating cohort (3/3) - icu_visit✔ Generating cohort (3/3) - icu_visit [191ms]


cohortAttrition <- CDMConnector::attrition(cdm[["test_cohorts"]])
  
excluded_records <- cohortAttrition %>% 
    pull(excluded_records) %>% 
    sum()
  
expect_equal(excluded_records, 0)

With graphCohort() it is possible to visualise the timeline for particular patient.