Activities

The research in the past year focused on use of patient data to improve information retrieval. The research goal is to improve information retrieval by automatically creating queries that account for specific characteristics of individual patients. Two main hypotheses are guiding this research:

• Patient information can be automatically extracted and used to improve the retrieval of desired information pertaining to care of individual patients.
• The performance of the knowledge-based approach will be superior as compared to the performance of traditional information retrieval systems.

In the past few months, this part of the research has focused on the possibility of using the co-occurrence of MeSH terms in MEDLINE citations associated with the search strategies optimal for evidence-based medicine to automate construction of a knowledge base. We use the UMLS semantic types in order to analyze search results to determine which semantic types are most relevant for different types of questions (etiology, diagnosis, therapy, and prognosis). A pilot study was performed in order to evaluate the clinical validity of the information retrieved. Current research is focused on the evaluation of the automated knowledge extraction from MEDLINE and the construction of the knowledge base to support the information extraction from the medical record.