Changing landscape

• Clinical decision support,
• surveillance,
• efficacy monitoring,
• integration across national health IT initiatives

requiring continually more cross-disciplinary integration.

Emerging clinical technical requirements

Integration requirements for

• clinical,
• clinical trial/clinical research,
• basic science data

looking more like those for traditional research, e.g.

• drug development,
• GWAS,
• academic biology.

Technical requirements:

• represent networks.
• minimal encoding - closest to perception.
• practical reasoning - write/debug execution once.
• comodity implementations.

off-the-shelf tools for data sharing/interoperability

• perform inference, e.g. CDS rules,
• map ontologies,
• map queries to e.g. SQL,
• federate queries.

Federation

Connects to:

• drug or disease ontologies,
• biological repositories, e.g. uniProt,
• drug interaction DBs, e.g. LODD,
• literature and rules repositories, e.g. Elsevier.

RDF DBs

• Current infrastructure is DB-oriented.
• SPARQL (think SQL) for query and update.
• Operates over web, local and virtual resources.
• Most of LOD cloud is SPARQL endpoints.

Vocabulary interactions

some image capturing long tail of vocabulary interactions for e.g.

• efficacy calculation/adverse events detection
• bio-surveillance
• pharmacogenomics

vs. the more monolothic

• radiology - images
• pathology - bio banking
• billing - patient procedures
• general med - patient procedures

Linked Data Platform

1. IBM's Rational division wrote n² interfaces.
2. Now using
   • REST to access documents,
   • RDF to integrate documents.
3. We can expect commodity libraries.