Interoperability of Swiss Personalized Health Network RDF Schema to FHIR
Overview
EFMI Medical Informatics Europe (MIE) is the flagship conference of the European Federation for Medical Informatics, bringing together clinicians, informaticians, and data engineers working on health data infrastructure, interoperability, and digital health across Europe. The 2025 edition was held in Genoa, Italy.
Together with my collaborator, Dr. Sophie IA Klopfenstein, I had the opportunity to present our work on mapping the Swiss Personalized Health Network (SPHN) RDF Schema to Health Level 7 (HL7) Fast Healthcare Interoperability Resources (FHIR).
SPHN's FAIR Data Team and the Digital Medicine and Interoperability team at the Berlin Institute of Health jointly mapped the SPHN RDF Schema 2025.1 to FHIR Release 4, producing over 700 FHIR artifacts that cover the full breadth of SPHN's semantic model. The result is the first comprehensive FHIR specification built for SPHN semantics. It lays the groundwork for exchanging SPHN-modeled research data through the FHIR ecosystem that Swiss hospitals and international partners already use for clinical and administrative interoperability.
Background
SPHN is a Swiss national initiative that supports FAIR (Findable, Accessible, Interoperable, Reusable), research-oriented use of health data. At its core sits the graph-based SPHN RDF Schema, which leverages standardized ontologies (e.g., SNOMED CT, ICD-10-GM, LOINC) and controlled vocabularies to achieve fine-grained semantic interoperability across Switzerland's decentralized healthcare system.
FHIR, in contrast, is a modular, widely adopted syntactic standard designed for scalable clinical, administrative, and research data exchange. It is set to become increasingly central in Europe as the European Health Data Space (EHDS) regulation and related initiatives take effect. Switzerland, while not part of the EHDS, has its own complementary Swiss Health Data Space, which is expected to align with EHDS principles.
Bridging the two standards lets institutions represent data captured under SPHN's rich semantic model through FHIR's widely supported syntax. FHIR-based tools and infrastructure can then reuse that data directly, without every consumer needing to understand SPHN's native RDF representation.
Representing SPHN semantics in FHIR
An earlier proof-of-concept had already mapped 59 of 63 SPHN concepts (SPHN RDF Schema 2022.1) to 22 FHIR resources. It showed that the approach was feasible but covered only a fraction of the schema. That result motivated the present effort, which targets complete concept coverage of the SPHN RDF Schema 2025.1.
Over roughly a year, FHIR and SPHN modeling experts met bimonthly to perform the mapping. The effort produced 179 logical models, 129 profiles, 55 extensions, 131 ValueSets, 3 CodeSystems, 11 ConceptMaps, and 226 instance examples. We validated these artifacts with the official FHIR Validator and compiled them into a draft SPHN-to-FHIR Implementation Guide using the HL7 IG Publisher.
Because the two models are not isomorphic, we had to decide how to reconcile mismatches rather than simply translate field-for-field. Key challenges included mismatched cardinalities, differing granularity, and structural differences between the models. We also faced ambiguity where a single SPHN concept could correspond to multiple FHIR resources. Where possible, we reused or closely mirrored existing FHIR resources.
The paper Interoperability of Swiss Personalized Health Network RDF Schema with FHIR describes the work in detail, and the draft Implementation Guide is available at sphn.gitlab.io/sphn-on-fhir/sphn-ig.