Graph-based creation of IDMP-conformant FHIR Medication instances using Neo4j: An evaluation study

Introduction: The availability of syntactically and semantically interoperable data is a critical prerequisite to high-quality research in the medical field. For medication data, the FHIR Core Data Set of the Medical Informatics Initiative Germany provides a powerful data model based on the IDMP standards that supports high levels of structure and quality. However, populating this model requires detailed medication information, which is often not available at the data integration centers. Objective: To address this issue, this study aims to implement a software to generate MII CDS-compliant FHIR Medication objects based on pharmaceutical databases and to assess their conformity to standards. Methods: Medication data from the MMI Pharmindex was loaded into a Neo4j graph database. Both manual and automatic mapping techniques were used to enrich the data with various terminologies. Finally, the resulting graph contents were exported into FHIR instances. Results: The export generated approximately 70,000 FHIR Medication instances conforming to the MII Core Data Set. The inclusion of terminologies and IDMP compliance were partially successful, with missing data in the source datasets proving to be the primary issue. Conclusion: Our results highlight the need for higher-quality data on pharmaceuticals available in Germany. Nevertheless, the generated Medication instances can accelerate the provision of a complete medication dataset in the MII data integration centers, marking a significant step toward future structured, detailed medication documentation.