Towards Achieving Semantic Interoperability in eHealth Services - E-Healthcare Systems and Wireless Communications - page 395

OUR APPROACH • Enable transparent or seamless integration

or interrogation of data sources.

The HSs in many countries are often made up
of independent organizations and institutions
providing health care services to its patients or
customers, forming often isolated yet loosely in-
terconnected organizations within the enterprise -
Health Service (HS). Across the domain, they have
their own rich and complex data representations,
terminologies, acronyms that are represented in
real terms by the health systems that support these
organizations. While services provide a potential
solution to enable interoperability between the
organizations and their systems, the focus needs
to be on sufficiently describing the data, and to a
lesser extent the systems themselves, opposed to
the functional services that enable access to them.
This is the approach taken in the ePCRN
project. The ePCRN (web or grid) services com-
position architecture includes a rich semantically
annotated representation of the 'problem' or the
'user data requirement' model and an enactment
broker that interoperates with health systems us-
ing information from a meta registry that holds
the domain ontology and mapping to health sys-
tems. As illustrated in Figure 1 , the architecture,
referred to as ePCRN-IIA (Interoperable Integra-
tion Architecture), includes five main semantic
elements to achieve this objective: 'Problem'
models referred to as Extensible Query Abstrac-
tion Model (EQAM); terminology and knowledge
services that provide rich domain terminologies,
referred to as Knowledge Services ; domain and
data governance ontologies referred to as Exten-
sible Domain Data Abstraction Ontology (EDAO)
and Extensible Data Governance Abstraction
Ontology (EDGAO) and mapping between them;
and a Meta Registry that holds rich metadata
descriptions of services (i.e. physical resources
or health systems). The general approach is to
capture the 'problem' in a semantically annotated
extensible model, which can be passed across the
infrastructure and translated to or by individual
services or EHRs using the domain and mapping

These capabilities require a flexible frame-
work that enables dynamic binding or access-on-
demand to the data sources or systems through an
interoperable integration framework. The need is
not only to enable configurable yet rich semantic
descriptions of these systems, but also the under-
lying data that they serve. Such descriptions are
not currently supported in current web services
standards or approaches.
Ontologies can be used to describe the seman-
tics of health systems or resources and make the
content explicit, and thus can be used to discover
semantic equivalence between information con-
cepts. The use of shared or standard ontologies
as a possible solution to the semantic interoper-
ability problem has been studied for over a decade.
They have also been found to provide reasonable
results for semantic interoperability [Wache, H.
et al (2001), Cruz, I. F. et al (2009), Wu, M. et
al (2009)]. The second issue that web semantic
approaches may encounter is the need to provide
users with a freedom of choice in terms of how
they structure their concepts or semantics and
their preferred representation language. This is
may be essential or at least difficult to enforce
for the wide use of web services in the market
place. This, however, will in turn lead to what is
often referred to as the ontology heterogeneity
problem [Visser, P. R. S., et al (1997), Farooq,
A. et al (2010)]. This may make it difficult to
achieve semantic interoperability as intended.
However, given the closed domain need, this can
be avoided by standardizing the ontology that can
be used to describe data as well as the service.
Equally, this domain ontology needs to be clear
enough and to employ an agreed format to be
used by data providers to describe their data in a
meaningful way. In the ePCRN project, we built
our ontology based on the CCR (a variation of the
widely adopted HL7) standard to represent the
core medical domain concepts for primary care.