FHIR Resource Generator

This program allows you to generate a bulk of FHIR resources in JSON format with randomly generated data. At its core, the program takes a JSON representation of a FHIR resource and applies a map of FHIRPath expressions to generator functions to obtain the value for the value path and sets the value in the resource whose initial value is obtained from the JSON representation.

Prerequisites

• Java 17
• Maven
• Access to a FHIR Terminology Server that supports the $expand Operation for ValueSets and has all required ValueSets

Usage

Load a FHIR Resource using Hapis IParser. Load a TestDataBluePrint using the FhirResourceFactory.loadFhirPathToValueFunctionFromJson function. Use the FhirResourceFactory.modifyResource function to modify the loaded resource based on the Blueprint FHIR Defintions. GeneratorFunctions provides the defintion of available functions to create the value defined by the fhir path key in the BluePrint. If you have many different profiles i.e. because you have a separate profile for each LaboratoryValue with its own specific ucum code. You want to generate the TestDataBluePrint based on the Profile Information or other sources. While the values might change for each profile the required FHIR Resource you want to modify want need to be changed for each case as long as all but also only the fhir path you want to have in your final resource are present. A good starting point for creating a resource you want to modify is ask Chatgpt to generate a FHIR Resource of specific type with a value for all fhir paths in your blueprint.

Example

{
  "resourceType": "Condition",
  "code": {
    "coding": [
      {
        "system": "http://snomed.info/sct",
        "code": "195967001",
        "display": "Hypertension"
      }
    ]
  },
  "subject": {
    "reference": "Patient/1234"
  },
  "recordedDate": "2023-02-15T08:30:00Z"
}

Defines a Condition that can be modified using the TestDataBlueprint information:

{
  "Condition.code.coding": "randomCoding(http://fhir.de/ValueSet/bfarm/icd-10-gm)",
  "Condition.subject": "randomPatientReference()",
  "Condition.recordedDate": "randomDateTime()"
}

A modified Condition might look something like this:

{
  "resourceType": "Condition",
  "code": {
    "coding": [
      {
        "system": "http://fhir.de/CodeSystem/bfarm/icd-10-gm",
        "code": "I11.0",
        "display": "Hypertensive Herzkrankheit mit (kongestiver) Herzinsuffizienz"
      }
    ]
  },
  "subject": {
    "reference": "Patient/c91a530a-72c0-46f3-b337-fcd097166e69",
    "display": "Berat Kustermann"
  },
  "recordedDate": "2006-07-21T09:26:24+02:00"
}

The modified FHIR resource can now be used for testing or other purposes. All test data generated have no real-world medical relation at all.

The program utalizes:

• https://github.com/DiUS/java-faker for fake names, adresses etc.
• https://github.com/hapifhir/hapi-fhir for FHIR Resource parsing and evaluating FHIR Path expressions.

Application Example

public class Application {


  public static void main(String[] args) {
    var bluePrints = BluePrintLoader.loadBluePrints(
            "src/main/resources/BluePrint/TestDataResourceBluePrint.json");
    var resources = bluePrints.stream().map(FhirResourceFactory::createTestResourceFromBluePrint)
            .filter(Objects::nonNull).flatMap(
                    Collection::stream)
            .toList();
    var bundle = FhirTransactionBundleConverter.convertToFhirTransactionBundle(resources);
    FhirResourceFactory.writeResource(bundle, "src/main/resources/Bundle/GeneratedBundle.json");
    var params = FhirResourceFactory.writeNDJsonByResourceType(resources,
            "src/main/resources/NDJson/test_resources");
    FhirResourceFactory.writeResource(params,
            "src/main/resources/Parameters/GeneratedParameters.json");
  }

}

Upload to FHIR Server

In Application.java you can find an example of how to use the program. Based on the BluePrints the test resources are generated. You can convert them two a FHIR Transaction Bundle to upload them to Standard FHIR Server like Hapi or blaze.

Upload to Pathling

To upload the generated resources to Pathling we need to create a NDJSON file for each resource type. writeNDJsonByResourceType does exactly that. And returns a Parameter Resource that can be used to import the resources into Pathling. Using pathlings .../fhir/$import operation. Make sure you mount the generated NDJSON files into the pathling container and eventually update the valueUrl in the Parameter Resource to match the path of the mounted files.

{
  "parameter": [
    {
      "name": "source",
      "part": [
        {
          "name": "resourceType",
          "valueCode": "Condition"
        },
        {
          "name": "url",
          "valueUrl": "file:///src/main/resources/NDJson/test_resources-Condition.ndjson"
        }
      ]
    }
  ]
}