Chapter 2: Literature Review
2.4 Industry Domain Specific Interoperability Models
Interoperability issues are of course widely discussed not only within the localisation community, but also within other domains (Naudet et al 2010) such as healthcare, e-governance, networking, military, and construction (Lipman et al 2011). Lessons can be learnt by studying how similar issues have been addressed in other domains. As an example, we will briefly review some of the most important interoperability issues in the healthcare domain, mainly to illustrate approaches taken to achieve semantic and syntactic interoperability issues. We have chosen the healthcare domain due to the critical nature of the interoperability of health records.
2.4.1 Healthcare
Interoperability is a crucial aspect of successful healthcare information systems. For example, the exchange of health records (such as the medical history of a patient, or test results) among different applications as well as institutions provides invaluable benefits.
At the same time, the correct interpretation of information that is being exchanged is vital to avoid life-threatening mistakes. Interoperability remains a huge problem in healthcare. A study by Ray (2009) reported that imperfect interoperability adds as much as $77.8 billion per year to the cost of healthcare in the USA. The main reasons include extensive use of “paperwork” in administrative procedures (Ray 2009) as well as the use of heterogeneous proprietary models for representing Electronic Health Records (EHR) by different medical institutions (Berges et al 2010). Therefore, interoperability is an active research area in the healthcare domain. Several healthcare information interoperability frameworks that are relevant to our study are summarised below.
In the healthcare domain, interoperability has been achieved by means of healthcare standards such as HL7 (v3): a standard for the exchange, management and integration of healthcare information (Sartipi and Yarmand 2008); OpenEHR (Berges et al 2010):
an open standard supporting the specification of better interoperable EHR systems, and SNOMED: a comprehensive clinical terminology system (Sartipi and Yarmand 2008).
Standard-based Framework for Achieving Data and Service Interoperability in eHealth Systems
Sartipi and Yarmand (2008) propose a standard-based framework for achieving data and service interoperability in eHealth systems. The main objective of this research was to migrate data and services used in legacy healthcare systems into standards-based interoperable systems. The methodology adopted in this research involved transformation of data, messages and transaction information used within legacy systems into standard based representations by developing a series of mappings. The semantic interoperability between healthcare applications was achieved by mapping their clinical terms into unique terminology systems. The mapping process was carried out with the help of domain experts. While the meanings of shared data were unified by means of mapping individual terminology systems into unique terminologies, the data transmission aspect of interoperability was achieved by using a proprietary SOA based system (known as Oracle’s Healthcare Transaction Base), due to its compatibility with a selected set of standards. The above framework has been evaluated by means of a case
real-world environment with the help of associated industrial partners and two other research groups. Sartipi and Yarmand (2008) highlight the characteristics of selected healthcare standards towards facilitating interoperability and discuss the applicability of their methodology in other domains.
Semantic Interoperability of EHRs
Kalra et al 2009 (cited in Berges et al 2010) identified three levels of interoperability related to EHRs: level 1 - syntactical interoperability; level 2 – partial semantic interoperability; and level 3 – full semantic interoperability. Berges et al (2010) propose an ontology-driven framework based on semantic web technologies (i.e. OWL2) to achieve full semantic interoperability of EHRs. The core of this methodology is the development of an ontology called EHROnt, which represents definitions of clinical terms in two levels of abstraction: canonical and application. At the canonical level, ontological definitions of EHR statements are represented in a higher application independent manner. This is carried out by the experts in the medical field. At the application level, definitions of EHR statements are represented as they are understood in specific e-health systems. As such, the canonical layer is common, while separate application layers exist for different e-health systems. Each health institution is responsible for creating their own application layer of the ontology. Berges et al (2010) propose a semi-automatic method translating existing proprietary EHR representations into an OWL2 based application layer. Finally, a mapping is constructed between terms found in each application layer with corresponding definitions found in the canonical layer. The proposed methodology is claimed to enable ‘on the fly’ interpretation of exchanged clinical data among different systems regardless of their individual internal clinical data representation schemes (Berges et al 2010). However, evaluation of the proposed methodology was not presented in this study.
Summary and Conclusions
The definition for interoperability by IEEE (see Sect. 1.1.1) focuses on two aspects:
“exchanging data” and “using the exchanged data”. Essentially, these two aspects are related to syntactic interoperability and semantic interoperability. Syntactic interoperability ensures successful data exchange while semantic interoperability ensures proper interpretation of exchanged data (i.e. the usage aspect).
It can be seen that existing research focuses on addressing both these aspects. Sartipi and Yarmand use mappings and SOA to achieve syntactic interoperability while centralised terminologies have been used to enable semantic interoperability. Berges et al mainly focus on addressing semantic interoperability issues by means of developing ontologies.
The main lessons that can be learned from the above studies are two-fold:
importance of representing proprietary data in a standard format;
importance of addressing both syntactic and semantic interoperability issues.
In the next section, we briefly review SOA as a popular mechanism to achieve data-exchange interoperability.