Saturday, July 23, 2011

HL7 and the Services Aware Interoperability Framework

Jobst Landgrebe is formerly Enterprise Architect at the International Institute for the Safety of Medicines
(ii4sm) and Co-Chair of the HL7 Vocabulary Workgroup. He and I recently collaborated on a paper entitled "The HL7 Approach to Semantic Interoperability" which was presented at the 2nd International Conference on Biomedical Ontology. The abstract is as follows:
Health Level 7 (HL7) is an international standards development organisation in the domain of healthcare information technology. Initially the mission of HL7 was to enable data exchange via the creation of syntactic standards for point-to-point messaging. For some 10 years, however, HL7 has increasingly conceived its mission as one of creating standards for semantic interoperability in healthcare IT on the basis of its 'version 3' (v3) family of standards. Unfortunately, v3 has been marked since its inception by quality and consistency issues, and it has not been able to keep pace with recent developments either in semantics and ontology or in computer science and engineering. To address these problems, HL7 has developed what it calls the 'Services-Aware Interoperability Framework' (SAIF), which is intended to provide a foundation for work on all aspects of standardization in HL7 henceforth and which includes HL7's Reference Information Model as general purpose upper ontology. We here evaluate the SAIF in terms of design principles that must be satisfied by a semantic interoperability framework, principles relating both to ontology (static semantics) and to computational behaviour. We conclude that the SAIF fails to satisfy these principles.
A video of Dr. Landgrebe's presentation is available here, and contains a number of interesting side remarks, including this (at 13:20):
The SAIF does not meet the needs of an interoperability framework. It can’t overcome the crisis of HL7 – and that’s a big problem because for example the IHE, the big organization that the government agencies are looking at to create interoperability in healthcare – it’s called "Integrating the Healthcare Enterprise" – is telling industry that it should use HL7 for everything in healthcare exchange .. , but if industry would do this nothing could work … My theory is that the IHE is set up exactly to make sure that nothing can work (but that’s – you know – conspiracy theory) … 
Update August 6, 2011:
I received the following comment:
Anonymous has left a new comment on your post "HL7 and the Services Aware Interoperability Framew...": 
I am not certain why your paper was accepted, other than the provocative subject matter. Much of what you state as fact is unsubstantiated opinion. Several of your premises are also faulty.
I am concerned, as this paper is quite misleading and/or misguided. 
You make sweeping generalizations without evidence of data or analysis. Many of these read more like bias rather than factual. Your paper contains several dubious citations. Many of the citations of your assertions don't logically support your argument. Some of the authoritative sources you cite are non-retrievable (404) URLs. Others simply not true. 
You often get tied up in knots in your logic, often based on a trifle (or misunderstanding). The line of reasoning is not clear and is encumbered by jargon, which seems to obscure rather than clarify. 
I can understand why correspondents who provide HL7 Watch with evidence of problems with HL7 request that they be allowed to remain anonymous -- criticizing HL7 can still, I am told, cause career problems in the health IT world. But why would someone who is taking the side of HL7 elect to remain anonymous in this way? Jobst Landgrebe and I provided arguments, and we provided an extensive list of references to support these arguments. (We regret that HL7 seems to have removed from the web some of the items for which we cited URLs since our contribution was first circulated.) Anonymous himself, however, provides assertions, which we would be more than happy to discuss. Perhaps he will reveal himself in order to allow such discussion to proceed.

Update August 18, 2011
One intriguing element of the SAIF literature is that it reveals that HL7 is now promoting the use of the word 'ontology' in application to the RIM. While in other circles there is increasing clarity as to the difference between an information model and an ontology (e.g. here and here), HL7 here seems to be once more muddying the waters.

No comments: