OMOP - Generating Characterization Files • ohdsilab

This vignette summarizes the code that can be found here: https://ohdsi.github.io/Characterization/ and illustrates how this code can be used within the OHDSI Lab against OHDSI Lab data.

Install one new package (Characterization) and load it along with the other typical packages.

renv::install("OHDSI/Characterization")

library(ohdsilab)
library(DatabaseConnector)
library(keyring)
library(ROhdsiWebApi)
library(CohortGenerator)
library(FeatureExtraction)
library(Characterization)

Set the connection details. I connect to the database here so I can set the option(con.default.value) setting in the next line.

Sys.setenv("DATABASECONNECTOR_JAR_FOLDER" = "insert path to JDBC driver")

connectionDetails <- DatabaseConnector::createConnectionDetails(
    dbms = "redshift",
    server = "ohdsi-lab-redshift-cluster-prod.clsyktjhufn7.us-east-1.redshift.amazonaws.com/ohdsi_lab",
    port = 5439,
    user = keyring::key_get("db_username"),
    password = keyring::key_get("db_password"))

atlas_url = "https://atlas.roux-ohdsi-prod.aws.northeastern.edu/WebAPI"
synpuf_schema = "omop_cdm_synpuf_110k_531"
write_schema = paste0("work_", keyring::key_get("db_username"))

con =  DatabaseConnector::connect(connectionDetails)

options(con.default.value = con)
options(schema.default.value = synpuf_schema)
options(write_schema.default.value = write_schema)

Connect to ATLAS

ROhdsiWebApi::authorizeWebApi(
    atlas_url,
    authMethod = "db",
    webApiUsername = keyring::key_get("atlas_username"),
    webApiPassword = keyring::key_get("atlas_password"))

Instead of setting one cohort Id, this time you’re setting a target cohort Id and an outcome id as the purpose of the characterization package is to compare a target cohort with an outcome cohort.

targetId <- 4675
outcomeId <- 4681

Pull the cohorts definitions from ATLAS, and use them to populate tables in your write_schema (this code comes from the “Generating a cohort” vignette)

cohortDefinitionSet <- ROhdsiWebApi::exportCohortDefinitionSet(
    baseUrl = atlas_url,
    cohortIds = c(targetId, outcomeId))


cohortTableNames <- CohortGenerator::getCohortTableNames(
    cohortTable = "cohort")


createCohortTables(
    connectionDetails = connectionDetails,
    cohortTableNames = cohortTableNames,
    cohortDatabaseSchema = write_schema)


cohortsGenerated <- CohortGenerator::generateCohortSet(
    connectionDetails = connectionDetails,
    cdmDatabaseSchema = synpuf_schema,
    cohortDatabaseSchema = write_schema,
    cohortTableNames = cohortTableNames,
    cohortDefinitionSet = cohortDefinitionSet)

Now, set the covariate settings. The createCovariateSettings() function can be used to customize the output of characterization. Run ?createCovariateSettings to view all possible arguments. For now, we’ll just use createDefaultCovariateSettings() as we don’t need anything specific. The aggreegateCovariate Settings will be used as an argument when we set the characterization settings. Other argument options for characterization are timeToEventSettings and dechallengeRechallengeSettings.

aggregateCovariateSettings <- Characterization::createAggregateCovariateSettings(
  targetIds = targetId,
  outcomeIds = outcomeId,
  riskWindowStart = 1,
  startAnchor = 'cohort start',
  riskWindowEnd = 365,
  endAnchor = 'cohort start',
  covariateSettings = FeatureExtraction::createDefaultCovariateSettings())

Set the characterization settings

characterizationSettings <- Characterization::createCharacterizationSettings(
  aggregateCovariateSettings = aggregateCovariateSettings)

Run the characterization. This will generate a series of csv files that can later be analyzed using traditional or programmatic methods. You may want to create a folder before running the following code and set the path of outputDirectory and executionPath to that folder.

Characterization::runCharacterizationAnalyses(
  connectionDetails = connectionDetails,
  cdmDatabaseSchema = synpuf_schema,
  targetDatabaseSchema = write_schema,
  targetTable = 'cohort',
  outcomeDatabaseSchema = write_schema,
  outcomeTable = 'cohort',
  characterizationSettings = characterizationSettings,   
  outputDirectory = "insert path to output folder",
  executionPath = "insert path to output folder",
  csvFilePrefix = 'insert custom prefix')

And that’s it! The characterization files may take a while to generate depending on the size of your cohorts and the amount of variables you include in your characterization settings.