OMOP - Generating a Synthetic Dataset • ohdsilab

This tutorial will illustrate how to use the Eunomia package created by the OHDSI community and described here: https://ohdsi.github.io/PatientLevelPrediction/. Researchers can use this package to create a synthetic OMOP CDM (Common Data Model) dataset for demonstration and testing purposes. This package may have limited usefulness within the OHDSI Lab due to thealready-included SynPUF synthetic dataset, but the Eunomia dataset can also be loaded outside the OHDSI Lab on a local machine or other workspace. This can be useful if you prefer to develop and test analytical code on your personal computer.

We’ll need to install one new package: Eunomia

renv::install("OHDSI/Eunomia")

library(Eunomia)
library(DatabaseConnector)

Now we create our connection details. This step is far simpler than creating connection details for the OHDSI Lab databases (PharMetrics and SynPUF).

connectionDetails <- Eunomia::getEunomiaConnectionDetails()

We connect to the Eunomia database is the same way we connected to the OHDSI Lab databases.

con <- DatabaseConnector::connect(connectionDetails)
options(con.default.value = con)

Now we can use the Eunomia database in much the same ways as we have the OHDSI Lab databases. Let’s generate a cohort of patients with Otitis Media to prove it. The only notable difference is the cohortDatabaseSchema and cdmDataBaseSchema will be set to “main”.

# Load the necessary packages
library(ROhdsiWebApi)
library(CohortGenerator)

# Set ATLAS url
atlas_url = "https://atlas.roux-ohdsi-prod.aws.northeastern.edu/WebAPI"

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

# Identify the cohort definition
cohortId <- 4734

# Export the cohort definition from ATLAS
cohortDefinitionSet <- ROhdsiWebApi::exportCohortDefinitionSet(
    baseUrl = atlas_url,
    cohortIds = cohortId)

# Set the naming convention for the cohort tables
cohortTableNames <- getCohortTableNames(cohortTable = "cohort")

# Create the cohort tables
createCohortTables(
    connectionDetails = connectionDetails,
    cohortTableNames = cohortTableNames,
    cohortDatabaseSchema = "main")

# Populate the cohort tables with persons in the Eunomia dataset who meet the
# cohort definition
cohortsGenerated <- generateCohortSet(
    connectionDetails = connectionDetails,
    cdmDatabaseSchema = "main",
    cohortDatabaseSchema = "main",
    cohortTableNames = cohortTableNames,
    cohortDefinitionSet = cohortDefinitionSet)

From here, we can do anything we normally do with cohorts (see vignettes 3-9). The Eunomia package can be a quick way to test or demonstrate code intended to be run against an OMOP CDM dataset.