Semantic equivalency between concepts of different ontologies, terminologies or standards is determined by the knowledge experts and authors of such concepts as well as the context in which a concept is represented. Interoperability, however, as it applies to electronic data sharing, depends on how technology can represent the structure of the ontologies as well as mappings across terminology sets to be delivered in real-time during a health care encounter. From the starting point of multiple disparate hospital systems with proprietary, non-standardized terms and codes, interoperability depends on the adoption of Standards. Local hospital and proprietary software vendor codes can be migrated over time to standards while immediately benefiting from having access to standards and cross mappings through the use of a language engine, such as Health Language’s Language Engine, known as LE. Other ontologies (genetics, biomedical, etc.) can be integrated with patient health information by being added to the knowledge base as it is “aligned” with other terminologies. Critical is the ability of an ontology to maintain it’s original structure and relationships and also the ability to be mapped to other concepts across different code sets as deemed equivalent by the knowledge experts. Unlike the UMLS which provides a vast resource of mapped ontologies, the knowledge base to be used in an electronic medical record may have code sets that stand alone or ones that may be mapped to cross concepts (reimbursement codes may or may not represent a clinical concept). There also needs to be a knowledge base that can be updated and accessed independently of the vendor application as the various standards bodies and code set experts produce updates and new versions throughout the year.

Such a large scale interoperability application is currently underway throughout the United Kingdom where the National Health Service (NHS) is using LE (www.healthlanguage.co.uk) throughout the country to integrate and share patient data for over 50 million people. The Language Engine will serve as the key component behind all of the interoperable components where content is standardized and shared—across all five LSP’s (Local Service Providers) and NASP (the National Spine) where there will be an estimated 10.4 million transactions/day. The National Electronic Library for Health (in the UK) is already using the HL Language Engine for coding and standardization of their documents. All outputs from LE are in an XML format for easy integration into application and HL7 messages.

In order to achieve a seamless exchange of health information across Europe, the architecture requirements for electronic interoperability need to be similar to those represented by the Health Language model:

• Any code set or language can be represented in the knowledgebase, as well as any corresponding mappings between code sets. Mappings may be those provided by standards bodies or by specialty groups or by hospital systems.
• Interoperability across multiple systems in enabled by providing access to mappings between local code sets and standard code sets as well as mappings between standard code sets that may apply to different contexts. (For example, clinical data can be captured once at the point of care using a standard such as SNOMED CT. The built-in cross mappings would enable the system to generate corresponding standard codes for billing, procedures, ICD, local lab code, etc.
• Terminologies and Standards are to be integrated with applications in run-time mode. A separate terminology server provides a consistent interface with which all applications across the enterprise may receive access to the knowledgebase of ontologies, terminologies or code sets.
• Local concepts can be mapped to Standards ,individual code sets (e.g. SNOMED CT, ICD, etc) can be updated within the knowledgebase on a timeline of their release dates and text documents and text fragments can be parsed, permitting a user to identify entered free-text as concepts/terms/codes from a standard.