Interactive Patient Designer & Test Integration

library(PatientGenerator)
library(shiny)
library(testthat)

Introduction

The PatientGenerator package provides an interactive Shiny application, patientDesigner(), to visually construct and edit synthetic patient datasets for the OMOP Common Data Model (CDM). This vignette demonstrates how to utilize the application to generate test data and integrate these datasets into a testthat suite.

Operating the Patient Designer

To launch the interactive designer, execute the following command in the R console:

patientDesigner()

The application features:

• Person Table: Manage individual patient records (e.g., gender, birth year).
• CDM Tables: Add and edit clinical events such as Observation Period, Condition Occurrence, Drug Exposure, Measurement, and Procedure Occurrence.
• D3 Timeline: A visual, drag-and-drop timeline for adjusting event dates.
• Persistence: Export constructed datasets as JSON files.

Saving Test Sets for testthat Integration

By default, patientDesigner() searches for a tests/testthat/testCases directory within the current project. Clicking “Save Test Set” in the sidebar saves the JSON file directly to this location.

This facilitates a seamless development workflow: 1. Launch the application from the package root. 2. Visually design the test case. 3. Save the file (e.g., my_test_case.json). 4. The dataset is immediately available for unit testing.

Integrating with testthat

Saved test sets can be integrated into unit tests by loading them using the CDM constructor.

test_that("Custom test case behavior is correct", {
  # Locate the test case file relative to the tests/testthat directory
  test_file <- testthat::test_path("testCases", "my_test_case.json")
  
  # Instantiate the CDM constructor and load the JSON test set
  # Note: cdmConstructor is internal to PatientGenerator
  cdm <- PatientGenerator:::cdmConstructor$new()
  cdm$loadJsonTestSet(test_file)
  
  # Execute package logic using the synthetic data
  person_data <- cdm$person$data()
  
  expect_gt(nrow(person_data), 0)
  expect_equal(person_data$gender_concept_id[1], 8532) # Female
})