gms | German Medical Science

Introduction: The German Portal for Medical Research Data (FDPG) is the central access point for researchers in the German Medical Informatics Initiative (MII) for performing feasibility queries, creating data usage project requests and enabling data extraction of healthcare data across 39 university hospital sites in Germany. These overarching processes are supported by a federated business process engine based on the data sharing framework, enabling to query and exchange data across individual institutional boundaries. In this abstract, we propose a concept for a distributed FHIR validation process, with guarantees on data integrity for cross-institutional healthcare data exchange.

Methods: Fast Healthcare Interoperability Resources (FHIR) is used as the underlying interoperability standard for the MII core data set (CDS). The data extraction process is used to obtain reduced FHIR instance data only containing the attributes for a specific data use project (DUP). While FHIR validation is commonly employed during development in the context of data extraction and exchange, it is a necessity for information interchange in a federated network where the execution of the data extraction is performed on unknown source data.

Results: We propose an extension to the federated infrastructure for supporting distributed validation at different steps and organizations to ensure uniformity, establish trust and provide guarantees on the instance data that allows researchers to make assumptions on data quality without direct access to the sources. To enable a distributed validation a single source of truth with a synchronization mechanism for all core data set profiles and terminology artefacts is required for all organizations. For validating the data within the data extraction for a specific DUP, a set of DUP profiles should be derived from the CDS, that detail the expected attributes, types, ranges and values, of each data element. For DUP created in the FDPG, the Clinical Cohort Definition Language (CCDL) and the derived Data Extraction Query (DEQ), serve as additional sources of truth during the data sharing process. A data extraction and validation package consisting of the DEQ, the CCDL and the derived DUP profiles should then be distributed to each clinical site and the project specific data management site.

Discussion: We regard the synchronization of single source of truth artefacts for structural and semantical resources with local instances at the clinical sites and an organizational commitment to interoperable FHIR validator implementations to establish a ground truth, as pivotal. The FHIR specification is extensive, and validators do not necessarily cover the full scope of the specification. Additionally, edge cases in the specification leave room for interpretation, which highlights the need for distributed validation against the same artefacts. Others concluded that validation at different organizations during a distributed data sharing process is required to achieve and verify syntactic and semantic uniformity and to provide automated feedback on the success of a data transfer and merging of medical data. The proposed concept needs further discussions before final implementation, nonetheless this concept could ensure structural, semantical and process interoperability by defining trust and uniformity between the participating organizations.

Acknowledgements: This work is supported by grants from the German Federal Ministry of Education and Research, under grants 01ZZ2309A, 01ZZ2309D, 01ZZ2309E (FDPG+), 01ZZ2307A, 01ZZ2307B (DSF-Community), 01ZZ2312A (SU-TermServ).

The authors declare that they have no competing interests.

The authors declare that an ethics committee vote is not required.