gms | German Medical Science

Introduction: In medical research, efficient collection and completeness of data are crucial. The Data Integration Center (DIC) of the Jena University Hospital (JUH) is developing a hybrid data collection strategy for this reason. Hybrid data collection is used in clinical research for supplementing classical documentation data from electronic health records (EHRs) through an electronic data capture (EDC) system. While clinical systems are the primary data source, EDC adds flexibility to address use-case data requests and some of the data quality challenges related to EHRs. The combined data allows research through a centralized source, which has to adhere to interoperability standards suitable for healthcare data [1].

Methods: We utilize REDCap [2] as EDC system at the DIC Jena, where FHIR resources generation is necessary according to the Core Data Set profiles of the Medical Informatics Initiative (MII-CDS) [3]. The FHIR Services Module [4] of the Vanderbilt REDCap Group offers the foundation for the direct creation of FHIR resources from REDCap data input. The module works as follows: first, annotations are added to the REDCap form fields. For example, the resource type and attribute name that correspond to the field are included in this annotation, so that they will have a structure comparable to EHR data. Metadata is created to clarify the origin of the resource. Next, FHIR resources are generated based on the annotations and output as bundle resources.

Results: We developed an extended prototype of the FHIR Services Module. To generate MII-CDS-conform FHIR from REDCap we modified the FHIR Services Module. The output is validated to ensure MII-CDS conformity, by installing the published MII-CDS profiles on a FHIR validator. The updated FHIR Services Module was successfully tested with a REDCap form (including data such as birthdate, gender, body weight, etc.) and checked for the functionality of field annotation, profile and provenance addition, FHIR bundle generation and MII-CDS validation. Upcoming tests will also include data from the HELP study [5], where previously collected questionnaires were post-processed into FHIR resources with great effort. The current prototype proved that such complicated post-processing might become obsolete.

Discussion: We are able to enhance the process to generate FHIR from EDC systems such as REDCap. One caveat is that the MII-CDS is only one among several specific FHIR profiles obligatory in Germany. Furthermore, our current implementation is based on our own metadata model, so establishing a cross-DIC metadata model for data provenance is essential. This will eventually enable communication with FHIR servers from different DIC.

Conclusion: The modified workflow of generated FHIR currently focuses only on specific profiles but prospectively will conform to more profiles that are relevant. Creation of MII-CDS-conform FHIR data through EDC underscores the potential for streamlined integration into DIC. This offers pathways for enabling clinical research through hybrid data collection relying on modern interoperability standards for healthcare data. Contribution of our code to the official development of the FHIR Services Module is planned.

The authors declare that they have no competing interests.

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