Presentation on theme: "I2b2 grid integration with Health Ontology Mapper CTSA Informatics All Hands Meeting October 24, 2009 Rob Wynden (UCSF)"— Presentation transcript:
I2b2 grid integration with Health Ontology Mapper CTSA Informatics All Hands Meeting October 24, 2009 Rob Wynden (UCSF)
Interrelated Teams Ontology Mapper: UCSF, UCD, Rochest, UPenn caGrid and TRIAD: OSU I2b2 and SHRINE: Harvard CICTR: UW, UCSF, UCD (SHRINE grid evaluation) HSDB: UCSF, WUSTL, Stanford, UCD, Columbia, Duke, Emory, Johns H, Mayo, OSU, Rockefeller, UW, UTSW, UTHSC (a caGrid based Network) Other grid launching with OntoMapper: DBRD (Dist Biobank for Rare Disease), HOMERUN (Hospital Reengineering) etc.
Project Objectives To create a general purpose instance mapper that can provide a best-fit translation of locally encoded data into multiple standard formats To allow the just-in-time translation of locally encoded data (not handled during ETL) To allow researchers to request data in any standard format that they require To enable the inter-institutional sharing of data encoded in standard formats
I2b2 based initial deployment Many CTSA sites are already using i2b2.org as their integrated data repository (IDR) platform By integrating i2b2 with grid computing environments we can rapidly connect a large number of CTSA sites Integration with sites NOT based on i2b2 is also essential
Ontology Translation The ontology used in the HSDB project is OCRe (the ontology of clinical research) using OWL The OWL ontology must first be translated into ISO 111-79 format ISO 111-79 elements that can carry a data payload are then annotated with Common Data Element (CDE) IDs and checked into either caDSR or OpenMDR
Semantic Interoperability ontology check-in to a Data Standards Repository (DSR)
Instance Mapping The caGrid and TRIAD environments provide semantic interoperability The data exposed on the grid must have the same value-set ranges for query to be possible (we also need syntactic interoperability) Ontology Mapper provides the best-fit instance mapping of that local data for query over caGrid
Syntactic Interoperability instance mapping of local data
Two i2b2 based grid projects SHRINE – a Harvard based project to create a new grid platform (Andy McMurray at Harvard leads that team). The CICTR project is based on this technology I2b2 caGrid connectivity – the existing caBIG grid augmented with the TRIAD platform (Shannon Hastings at OSU leads that team). The initial pilot of the TRIAD project is the HSDB (Human Studies Database) project (Health Ontology Mapper will support both grids)
OntoMapper Encoding Tables Health Ontology Mapper extends the i2b2 Concept Dimension by adding keys to new tables called encoding tables. The Encoding Tables associate caGrid Common Data Elements (CDE’s) with i2b2 Concept Dimension entries. OntoMapper therefore integrates i2b2 with the semantic interoperability components of caBIG for both SHRINE and caGrid.
Summary The Health Ontology Mapper extends i2b2 so that it can contain and reference formally encoded information in a semantically interoperable way. Our initial pilot for the HSDB project has been successful. We have initiated HSDB queries from caGrid of an i2b2 based integrated data repository. We will now seek to integrate OntoMapper more closely with the SHRINE grid and it’s SHRIMP association mapper. We will also now create deployment and user documentation for more wide-spread usage. Contact: Rob Wynden (firstname.lastname@example.org)