FAIRification of health-related data using semantic web technologies in the Swiss Personalized Health Network

Abstract

The Swiss Personalized Health Network (SPHN) is a government-funded initiative developing federated infrastructures for a responsible and efficient secondary use of health data for research purposes in compliance with the FAIR principles (Findable, Accessible, Interoperable and Reusable). We built a common standard infrastructure with a fit-for-purpose strategy to bring together health-related data and ease the work of both data providers to supply data in a standard manner and researchers by enhancing the quality of the collected data. As a result, the SPHN Resource Description Framework (RDF) schema was implemented together with a data ecosystem that encompasses data integration, validation tools, analysis helpers, training and documentation for representing health metadata and data in a consistent manner and reaching nationwide data interoperability goals. Data providers can now efficiently deliver several types of health data in a standardised and interoperable way while a high degree of flexibility is granted for the various demands of individual research projects. Researchers in Switzerland have access to FAIR health data for further use in RDF triplestores.

Fig. 1

Screenshot of the class Heart Rate from the SPHN RDF Schema visualised in pyLODE. The class Heart Rate is represented with its label, URI (which stands for Uniform Resource Identifier and is a type of IRI that does not support Universal Coded Character Set), description, meaning binding, parent class and connected properties and their restrictions. In addition, a visual schema representation of the Heart Rate class with its possible properties is given.

Fig. 2

Connection between a subset of the SPHN RDF Schema with a focus on the “HeartRate” concept and an example of data instantiation. In the schema, the top part of the figure shows the SPHN classes and properties in the schema and how they are interconnected. The class “HeartRate” is connected to the class “Quantity” through the property “hasQuantity”, meaning that the heart rate has annotated quantity information, which in turn contains the numerical value (through the property “hasValue”) and the “Unit” of the value (through the property “hasUnit”). In addition, the SPHN RDF Schema links measurements to the patient (i.e., the class “SubjectPseudoIdentifier”) via the property “hasSubjectPseudoIdentifier”. The bottom part of the figure depicts an example of an instance of a heart rate measurement. The heart rate measured (“heartRate1”) from the patient (“patient1”) has a quantity of 75 beats per minute (represented with “beats/min”, as per the UCUM notation). These data instances are connected to the classes in the schema via the property “rdf:type”, which allows annotating the resource with the intended meaning (e.g., patient1 is an instance of SubjectPseudoIdentifier as defined in the SPHN RDF Schema).

Fig. 3

(a) SPARQL query and (b) visual representation of the related classes. The query retrieves the list of distinct patients (sphn:SubjectPseudoIdentifier) who had an allergic reaction (sphn:AllergyEpisode) to any (including a descendance of) “Pulse Vegetable” substance (snomed:227313005) with RDFS inference enabled.

Fig. 4

(a) Excerpt of the oxygen saturation from the SPHN RDF Schema 2022.2 Turtle (.ttl) format and (b) its visual representation. The excerpt shows the implementation of an owl:Restriction for the code allowed for annotating the unit of an oxygen saturation. The only value allowed is “percent” from the UCUM notation, which is stated with the property owl:hasValue.

Fig. 5

Translation of the SPHN Dataset content into the SPHN RDF Schema. Concepts from the SPHN Dataset are translated into classes in the SPHN RDF Schema. The composedOfs, depending on the type of value, are either translated into object properties (i.e., for instances of concepts and qualitative elements) or data properties (i.e., for quantitative, string and temporal elements). Value sets and subsets defined with the SPHN Dataset are translated into either classes (e.g., SNOMED CT, CHOP, LOINC values) or named individuals (i.e., SPHN and UCUM values). Meaning binding to SNOMED CT and LOINC are interpreted as equivalent classes in the SPHN RDF Schema.