Work package 4: Standardisation, semantic interoperability and data integration
Brief description and aims of work
The objective of WP 4 is to identify in a functionality driven and user-oriented way the semantic resources needed, identify the candidates for re-use (by evaluation), and provide a common structure to unify the semantics for the project or the sub-task at hand. p-medicine’s approach to semantic interoperability and data integration is an ontology-based one.
The research focus of WP4 is not to create new ontologies, but to find technical solutions aimed at the specific needs of the project in identifying case-relevant ontologies and re-using or merging them. Nevertheless we need a semantic basis for the data integration and sharing in p-medicine which will be provided by the ACGT Master Ontology and its derivates, most importantly the Middle Layer Ontology for Clinical Care (MLOCC) developed in FP7 project CHRONIOUS will be used and further developed for leukaemias. Our aim is bringing together experience from ontology-driven clinical research (ACGT, @neurist , CHRONIOUS) with biomedical ontologies for the more theoretical disciplines which are collaboratively developed and stored in the OBO Foundry , e.g. Gene Ontology, Foundational Model of Anatomy (FMA), Ontology of Biomedical Investigation (OBI), is of key importance. We need to address ontology-specific variations of the bench-to-bed problems to bridge that gap for the data within p-medicine.
With respect to this the challenge of p-medicine regarding semantic interoperability revolves around two key issues: mapping of data in pre-existent HIS and extension and optimization of semantic resources against the background of evolving data that is pushed into the system. Automated ways to merge one semantic resource from a selection of pre-existing ontologies is a key strategy that needs to be developed within p-medicine. Similar approaches have already been taken, but the challenge is to achieve this without compromising the high standards of quality in the selected resources.
Regardless of our focus on re-use of ontologies, we will need to provide an extension of an Upper Ontology representing entities in the clinical domain. First steps into this direction have already been taken BioTop , but we need to evaluate their coverage with respect to the p-medicine scenarios and MLOCC and extend them where necessary. In doing so, we will provide a Common Biomedical Schema that rests on ontologies that already exist and consists of non-domain-specific theoretical classes and of high level representation of the domain. The sources for this deliverable are Basic Formal Ontology (BFO) , Ontology of Biomedical Reality (OBR) (Rosse, 2005 ) and the Middle Layer for Clinical Disease Management from the CHRONIOUS project (MLOCC).
The objective is to explore possibilities to automatically search and assess existing ontologies, not only with respect to common quality criteria, but also regarding specific issues in biomedical semantics, like granularity. In order to support the semantic integration of data as required within the p-medicine environment the following questions need to be answered:
- What are the entities necessary to represent clinical data in a patient-oriented way? How can we provide a consistent formalization of these? Which parts of pre-existent ontologies can be re-used in order to avoid development from scratch?
- How to get seamlessly integration of data from the hospital information system into ObTiMA and the data warehouse using the push- and sync- services developed in WP8? How can we overcome the gap between clinical terminologies and messaging systems in use (ICD-10, HL7) and the ontology?
- How can we seamlessly and ad hoc exploit individual ontologies that are built in accordance with the same quality criteria into a functionality-oriented semantic resource? How can we setup this process, maintain it and store its results in ways that support the domain experts to not only guide it, but actively participate in it?