Health Informatics

There are many fascinating natural language processing problems to be found in the medical domain, which has led to our growing interest in the area of Health Informatics. Much of our research in this area takes advantage of the ontologies, hierarchies, and terminologies found in the Unified Medical Language System (UMLS).

We have developed measures of semantic similarity and relatedness for the medical domain. These are found in UMLS::Similarity, which also relies on UMLS::Interface. You can try out a web interface for UMLS::Similarity on our servers atlas or maraca.

We have also developed an unsupervised knowledge based method of word sense disambiguation UMLS::SenseRelate. Previously we developed a supervised word sense disambiguation system (CuiTools) . Both of the disambiguate medical terms relative to senses found in the UMLS.

A number of UMD graduate students have held summer internships at the Mayo Clinic, Division of Biomedical Informatics, in Rochester, Minnesota. Bridget McInnes participated in the National Library of Medicine (NLM) Medical and Graduate Student Training Program (March - September 2008).


Current Health Informatics Collaborators

By: Ted Pedersen - tpederse AT d umn edu