Towards Achieving Semantic Interoperability in eHealth Services - E-Healthcare Systems and Wireless Communications - page 401

Figure 6. Mapping Route (Example)

relevant components and services of ePCRN-IIA
and does not include all implemented services.
The described prototype has been implemented
and is being used in three clinical research centres.
The architecture of the current prototype has
followed, at its core, the general implementation
of grid-based applications for the distributed
middleware. Although, grid-based services as-
sume more static binding behaviour, which often
and lack semantic support, they provide an ideal
distributed infrastructure. Thus the decision is
to use a grid infrastructure but modify or add
components to employ semantically rich problem
models combined with domain ontologies, sup-
ported by knowledge services.
The problem models, EQAM, combined with
the terminology and knowledge services are the
backbone of this approach to enable the semantic
interoperability in the current prototype. They en-
abled semantic-awareness through the architecture
to individual data sources. However, the richness
of the semantic-awareness in the architecture
depends, to a great extend, on the richness of the
terminologies and knowledge services. While
there always will potentially be limitations of the
richness of terminologies and their mappings, the

key is to enable observing and recognising it in
the respective data services. There needs to exist
a feedback mechanism to recognise and quantify
semantic shortcomings where they exist. Given
the legacy nature of electronic health records, it
would be difficult to assume completeness in the
knowledge services that include all terminologies.
However, the key is that this approach enables
extensibility, allowing the addition of terminolo-
gies while maintaining the integrity of the results
of problem model instances, which is critical in
the clinical domain.
We found only very minimal changes needed
to the domain and data governance ontologies
across services deployed at the different clinical
centres. While the prototype was implemented
only in three different clinical centres, they
represented a reasonable sample of the primary
care domain in the UK and USA. However, it
can be a time consuming task to create domain
ontologies for a given domain, this was noted
for the first prototype given the lack of knowl-
edge of the interdisciplinary collaboration and
evaluation of potential contributing standards.
However, this can be improved once these steps
are overcome especially for other domains. The