Abstract:
Linking healthcare information which nowadays is becoming vastly available with computational physiology models can be instrumental for enabling personalised and predictive clinical decision support systems. In the computational physiology domain semantic interoperability heavily relies on Semantic Web technologies and utilise ontology-based annotations but a wealth of useful information and knowledge sits in EHRs where this technology has limited use. openEHR and ISO13606 provide open standards for the structure, storage and exchange of healthcare data which readily support terminology/ontology based bindings that be exploited to link the two domains. This linkage will be bidirectional which means it will enable data discovery for computational modellers and also model discovery for clinical users. Since the openEHR specifications now underpin many national programs and regional implementations this can unlock unsurmountable potential to create both model and data driven personalised and predictive analytics. Having fit for purpose and standardised ontologies (such as Gene Ontology, Foundational Model of Anatomy) and clinical terminology (such as SNOMED CT, LOINC, ICD) have been a critical first step. However there is still a need to create mappings between these ontologies and develop standard annotation protocols. We describe our high-level methodology with an emphasis on ontology mapping using a crowd-sourcing approach.
