FAIR data principles and open science are globally endorsed as beneficial for healthcare. As co-founders of the FAIR principles we investigate implementation of FAIR principles, particularly interoperability for machines and interdisciplinary FAIRification. The Covid-19 pandemic emphasized that FAIRification ‘at source’ is vital: observational data are first collected in hospitals and should become FAIR for researchers as quickly as possible, inside and outside of the hospital. However, multiple information systems are used inside hospitals that are not directly interoperable. At the same time, existing systems have their own value such that replacing them is not desirable.

Here, we present a strategy to implement FAIR principles that complements existing hospital systems. We coordinated the FAIRification of observational data of hospitalised patients within an interdisciplinary collaboration that was organised within the hospital to face the Covid-19 challenges. We defined an architecture around ontological models that link data in existing systems. Guided by research questions of the medical doctors, we transformed data into machine actionable digital objects, and developed ontological models for data and metadata, including investigational parameters. DCAT2-structured metadata was exposed by FAIR Data Points. We demonstrated machine actionability by (i) federated queries across hospital data and existing Linked Data-based knowledge sources, (ii) Web APIs for querying Linked Data, (iii) hypothesis-support applications built on top of FAIR patient data.

Our work demonstrates that a FAIR research data management plan based on interdisciplinary collaboration and ontological models for data and metadata, Semantic Web technologies, and FAIR Data Points can complement hospital infrastructure to make machine-actionable FAIR digital objects available for integrative analysis. This prepares hospital systems for federated analysis (e.g. as part of the European Open Science Cloud), linking to other FAIR data such as Linked Open Data, and reuse in software applications.