Service Unit Terminological Services (SU-TermServ)

Motivation

As healthcare becomes increasingly digital, the semantically interoperable exchange of medical data is gaining critical importance. The use of standardized, international terminologies and classifications is essential to ensure consistent interpretation and reliable data exchange across different IT systems. To technically support this interoperability, the Service Unit “Terminological Services” (SU-TermServ) project establishes a central terminology server for the national research IT infrastructure within the Medical Informatics Initiative (MII) and the Network University Medicine (NUM). This server serves as a cross-institutional infrastructure component, aiming to provide standardized terminological resources and thereby enable cross-site collaboration and the reusability of medical data. The infrastructure need for such solutions is becoming increasingly evident and has been reflected in legislation: With the Hospital Care Relief Act (KHPflEG) passed at the end of 2022, the mandatory development of a national terminology server by the Federal Institute for Drugs and Medical Devices (BfArM) was anchored in § 355 of the German Social Code (§ 355 SGB V). However, the requirements within the MII and NUM projects go well beyond the scope of the BfArM, with a strong demand for central terminology services to ensure unified semantic standards and nationwide interoperability. The primary target groups include the developers of the MII core dataset, as well as the data integration centers (DIZ) at university hospitals.

FHIR Terminology Server & Services

Terminology servers are specialized software systems that provide access to terminological content through standardized APIs. Within the distributed infrastructure of MII and NUM, the HL7 FHIR Terminology Module plays a central role. It supports the management and usage of medical terminologies and includes three key resources: CodeSystem, ValueSet, and ConceptMap. These resources are referenced in FHIR profiles, which typically reflect specific clinical requirements. They define the scope of the data to be processed and set standards for the creation of medical content, such as in the core dataset module “Lab Results.” Through FHIR packages, these resources are bundled and made available, forming a foundational basis for the standardization of health data. As the terminology server, the project uses Ontoserver, developed by the Australian E-Health Research Centre at CSIRO. Ontoserver not only manages resources but also offers a syndication function to distribute terminology content to local servers. Via FHIR operations such as $lookup, $expand, and $translate, the server enables flexible use of various coding systems, fostering interoperability in healthcare.

By combining a robust software platform with expert support services and defined maintenance processes, FHIR-based terminology services offer substantial value for addressing challenges in the ongoing development of Germany’s digital health ecosystem.

Objectives and Approach

The SU-TermServ project aims to provide a central infrastructure to make medical terminologies accessible, manageable, and interoperable for the MII and NUM projects. A central terminology service based on the CSIRO Ontoserver has been implemented. The service focuses on the key FHIR resources – CodeSystem, ValueSet, and ConceptMap – required for the MII core dataset (KDS). These resources may be defined directly in the KDS or originate from HL7 base modules. In addition, other essential terminological resources, such as those from BfArM, are also made available as FHIR packages. All created packages are made transparently available via GitLab. For coding systems not yet available in FHIR, powerful and innovative tooling is being developed to support their representation in FHIR.

Results and Innovations

Since November 2024, SU-TermServ has operated a central terminology server for the MII and NUM. The server currently hosts around 1,800 CodeSystems, 5,200 ValueSets, and approximately 100 ConceptMaps. As a core infrastructure component of the MII, this terminology server also forms the basis for the German Health Research Data Portal (FDPG), which enables researchers to request access to data for distributed research projects.

The services and objectives of MII and NUM are highly innovative on a global scale. Likewise, the delivery of terminology services through SU-TermServ is both unprecedented and forward-thinking.

By leveraging emerging standards driven by the dynamic and diverse international HL7 FHIR standards community, this infrastructure significantly contributes to biomedical research.

Continuous Updates and Expansion

The server is continuously updated with new resources to provide up-to-date terminological content. Coding systems not yet in FHIR can be converted using the powerful BableFSH  tool and made available as terminology resources. Thus, HL7 FHIR terminology resources from various origins are aggregated and made centrally accessible. Users are also empowered to manage and propose their own resources for inclusion, providing value that extends beyond the MII core dataset. Thanks to the syndication feature of Ontoserver, these resources can also be deployed to local terminology servers for decentralized use.

Visualization and Management of Complex Resources

To improve the visualization and management of complex resources, we are developing platforms like the CRMI tool. Based on a new implementation guide from the FHIR community, it provides a robust infrastructure for organizing and distributing medical terminologies. Among other features, it visualizes dependencies between packages as interactive graphs, improving understanding of the relationships between resources and making access to relevant content easier. Our implementation of core aspects of the CRMI implementation guide is particularly innovative – to our knowledge, it is the first publicly available open-source implementation of this specification worldwide.

Support for Complex Terminologies such as SNOMED CT

A central focus is the support for complex terminologies like SNOMED CT, especially regarding postcoordination. This method enables the precise representation of specific clinical facts but is also highly complex, requiring deep knowledge of the SNOMED CT Concept Model and the Compositional Grammar. Due to its complexity and lack of software support, postcoordination is often avoided in practice. Therefore, projects like WASP aim to provide assisted authoring of postcoordinated SNOMED CT expressions. Additionally, the VaPCE project supports the validation of such expressions and is directly integrated into WASP. Only correctly structured expressions can be reliably processed. The focus is not only on verifying correctness, but also on generating precise correction suggestions. Both functionalities – assisted authoring and intelligent validation – represent major innovations, as no comparable tools existed previously.

Further Information

https://mii-termserv.de/

https://mii-termserv.de/assets/pdf/paper/20230505-MIRACUM_Journal-SU-TermServ.pdf

https://www.gesundheitsforschung-bmbf.de/de/distribution-von-ressourcen-und-diensten-16145.php

Theses and Internships

If you are interested in completing an internship or writing your thesis within this project, feel free to contact a member of the Lübeck project team.

Funding / Project Team / Collaborations

The Service Unit “Terminological Services” is a 2B project within the funding structure of the Medical Informatics Initiative, part of its new consolidation and expansion phase. The funding began in 2023 and will run through 2026, supported by the Federal Ministry of Research, Technology, and Space.

Project partners include University of Lübeck, University Hospital Cologne and Hannover Medical School

Lübeck Project Team:

  • M.Sc. Joshua Wiedekopf
  • M.Sc. Tessa Ohlsen
  • Dr. Ann-Kristin Kock-Schoppenhauer
  • Prof. Dr. Josef Ingenerf