Ontologies and ISO-compliant metadata as a sustainable foundation for HL7 FHIR in healthcare

1 Introduction

In healthcare, data is generated in a wide variety of systems: hospital information systems, practice management systems, registries, research infrastructures, and increasingly also in apps and wearables. For care delivery, research, and secondary use, it is crucial that this data remains understandable, interoperable, and reusable in the long term.

With HL7 FHIR (Fast Healthcare Interoperability Resources), a modern, REST-based standard is available that defines data formats and elements as “resources” and enables their exchange via JSON or XML. However, FHIR primarily focuses on structure and transport; the substantive meaning of the data must be derived through terminologies and metadata.

Against this background, ontologies and standardized metadata according to ISO/IEC 11179 and ISO 25964 are gaining importance. They complement FHIR with a semantic and governance-oriented layer that is intended to endure beyond individual projects and technologies.

2 Fundamentals

2.1 Ontologies and terminologies in healthcare

Ontologies formally model a domain of knowledge, typically in RDF/OWL, with clearly defined concepts, relations, and rules. In medicine, they are often used in the form of terminologies or classifications, such as:

• SNOMED CT as a comprehensive clinical reference terminology, represented in OWL and specifically described for use with HL7 standards, in particular FHIR.
• LOINC for laboratory and measurement data, which is increasingly linked to FHIR profiles and implementation guides.

Ontologies make it possible to describe clinical concepts in a machine-readable way, model relationships (e.g., specialization, part–whole), and thus create a basis for semantic interoperability, knowledge graphs, and explainable AI.

2.2 ISO standards for metadata: ISO/IEC 11179 and ISO 25964

ISO/IEC 11179 is the international series of standards for metadata registries (Metadata Registries, MDR). It describes how metadata can be standardized and registered so that data becomes understandable and shareable. Part 1 (Framework) provides the conceptual foundation; further parts define, among other things:

• a metamodel for metadata registries (Part 3),
• principles for naming and identification (Part 5),
• rules for formulating data definitions (Part 4), and
• procedures for registration (Part 6).

ISO 25964 is the international standard for thesauri and interoperability with other vocabularies. Part 1 covers the development and maintenance of thesauri for information retrieval; Part 2 covers interoperability with other vocabularies and provides recommendations for mapping and data exchange.

Both standards are intentionally designed to be technology-independent and formulate concepts such as “data element”, “concept system”, or “thesaurus” that can be used across different exchange formats and protocols.

3 Role of ontologies and metadata in the HL7 FHIR ecosystem

3.1 FHIR as a technical exchange layer

FHIR describes data formats and elements as resources (e.g., Patient, Observation, Condition) and specifies a REST-based interface for their exchange. The standard places great emphasis on ease of implementation and the reusability of building blocks in the form of profiles and value sets.

In this way, FHIR addresses syntactic and structural interoperability. Semantics are connected via the terminology module and external metadata layers.

3.2 Terminology services and metadata repositories

The FHIR terminology module provides mechanisms via resources such as CodeSystem, ValueSet, and ConceptMap to integrate, select, and map internal and external terminologies. Implementations such as Ontoserver or other SNOMED-based terminology servers use FHIR APIs to provide SNOMED CT, LOINC, and other vocabularies as services.

In parallel, metadata repositories are emerging that describe and manage data elements in an ISO/IEC 11179-compliant manner. One example is a FHIR-based metadata repository designed to conform to ISO 11179-3 and using FHIR resources as both input and output format in order to exchange and mediate data elements between partners.

This architecture leads to a layering:

1. FHIR as the transport and structure standard,
2. terminology services (ontologies, terminologies) as the semantic reference,
3. an ISO/IEC 11179 metadata registry as the governance and documentation layer.

4 Comparison with “lighter” modern approaches

4.1 Characteristics of lighter approaches

Modern “lighter” approaches are often understood to mean solutions that are limited to:

• simple JSON schemas (e.g., OpenAPI/Swagger),
• proprietary code lists,
• minimally defined fields with free text

and deliberately avoid extensive semantic modeling.

Such approaches have clear advantages:

• rapid implementation,
• low barrier to entry for developers,
• focus on specific integration scenarios without extensive lead time.

They are suitable for local or short-lived integrations, but are only of limited viability when long-term interoperability, secondary use, or AI applications are the primary focus.

4.2 Assessment with regard to sustainability

In comparison, ontologies and ISO-compliant metadata offer significantly greater sustainability:

• Stable semantics: Terms, data elements, and relations are formally defined and remain constant even when the technical platform changes. ISO/IEC 11179 explicitly aims to standardize metadata so that data remains understandable and shareable in the long term.
• Interoperability and mapping: ISO 25964 and ontologies explicitly address interoperability between different vocabularies and support systematic mapping strategies.
• Reusability: Once defined, data elements and value sets can be referenced in different projects and FHIR profiles; FHIR base and core profiles at the European level demonstrate this reuse approach.
• AI and analytics capability: Ontologies and standardized metadata form the basis for knowledge graphs, semantic queries, and explainable AI, whereas purely syntactic models quickly reach their limits here.

These benefits are offset by higher initial costs (modeling, terminology management, governance). In practice, however, these efforts typically pay off as soon as data is to be reused and analyzed across organizations and projects.

5 Conclusion and outlook

Ontologies and ISO-compliant metadata are far more than a theoretical add-on in the HL7 FHIR environment. They form the sustainable semantics layer required to make FHIR-based interfaces viable beyond the lifespan of individual applications.

Current practice is not developing toward an either/or, but toward hybrid architectures:

• outward-facing “lightweight” FHIR APIs with clear resources, profiles, and JSON schemas,
• inward-facing ontology-based terminologies (SNOMED CT, LOINC), and
• ISO/IEC 11179-compliant metadata registries as a foundation for governance, versioning, and reuse.

For organizations in healthcare, this means that short-term project goals (rapid implementation, easy integration) should be linked to a strategic semantics and metadata concept. Future developments—such as European FHIR base profiles, the European Health Data Space, and AI-supported analyses of routine clinical data—will further reinforce this trend and increase the need for robust ontologies and standards-compliant metadata infrastructures.