Mining Electronic Health Records using Linked Data

David J Odgers, Michel Dumontier

Research output: Contribution to journalArticleAcademicpeer-review


Meaningful Use guidelines have pushed the United States Healthcare System to adopt electronic health record systems (EHRs) at an unprecedented rate. Hospitals and medical centers are providing access to clinical data via clinical data warehouses such as i2b2, or Stanford's STRIDE database. In order to realize the potential of using these data for translational research, clinical data warehouses must be interoperable with standardized health terminologies, biomedical ontologies, and growing networks of Linked Open Data such as Bio2RDF. Applying the principles of Linked Data, we transformed a de-identified version of the STRIDE into a semantic clinical data warehouse containing visits, labs, diagnoses, prescriptions, and annotated clinical notes. We demonstrate the utility of this system though basic cohort selection, phenotypic profiling, and identification of disease genes. This work is significant in that it demonstrates the feasibility of using semantic web technologies to directly exploit existing biomedical ontologies and Linked Open Data.

Original languageEnglish
Pages (from-to)217-21
Number of pages5
JournalAMIA Joint Summits on Translational Science proceedings. AMIA Joint Summits on Translational Science
Publication statusPublished - 2015
Externally publishedYes


  • Journal Article


Dive into the research topics of 'Mining Electronic Health Records using Linked Data'. Together they form a unique fingerprint.

Cite this