Leveraging Biolink as a “Rosetta Stone” Between C-Path and EJP-RD Semantic Models Provides Emergent Interoperability

Abstract

Interoperability between clinical datasets is challenging due to, in part, the number of data models and vocabularies in use and the variety of implementations. Here we describe the first steps in an ongoing effort to achieve interoperability between two clinical datasets currently being constructed within independent international projects. Both are utilizing the FAIR Principles but have constructed their data models independently and have selected different ontologies. In this initial exploratory experiment, we examined the degree to which a mapping of both models into an independent schema, Biolink, can increase interoperability. Mapping was achieved by categorizing the key nodes in both data models as “types” of concepts in the Biolink schema. We found that with this very thin mapping in place, and without changing either model, queries could be constructed that extracted data from both datasets, demonstrating that at least some degree of interoperability had been achieved. Our results support the use of FAIR-compliant data representations, which are, by nature, more interoperable than legacy clinical data representations, even when the models have not been coordinated upfront.