RDF in healthcare

As far back as the early days of Artificial Intelligence, healthcare has always been a topic of interest for leading edge ideas in computer science, principally because medical decision making looks like a computable function. But the “return on investment” of such interest has been negligible, and there is no reason to suppose that the application of RDF to healthcare – even in areas such as bioinformatics or genomics - will be different. In fact the main result of computing’s interest in healthcare to date is a mess of unimplemented standards and a lack of basic applications – the New England Journal of Medicine reports that as of 2007 only 4% of US physicians had fully implemented electronic health records. The reality in healthcare is that even basic informatics building blocks such as a unique patient identifier are not sufficiently in place.

Wherever you look in healthcare there are too many elaborate standards (see Eric Browne's recent presentation on the semantic ambiguity of HL7) and too few effective ones. The NHS for example is awash with standard vocabularies – SNOMED, Read Codes, Clinical Terms, ICD 9 and 10, MeSH, OPCS, HRGs, NHS Data Dictionary, BNF Drug Categories, DMD. Yet none of them are used widely or applied consistently, nor are they interoperable – and none of them produce patient benefit. To imagine that this situation might be improved by adding RDF is a complete non-starter, even in areas such as drug discovery. In contrast, modest applications such as Google Flu trends, which links public health surveillance to search trends, might just be useful. There is one area where the idea of linked data ought to be surfacing - PCT data analysis - but there's no sign of any RDF activity in public health informatics.

Related Post:

Widget by [ Iptek-4u ]