Initializing the VA medication reference terminology using UMLS metathesaurus co-occurrences.

University of Utah, USA.
Proceedings / AMIA ... Annual Symposium. AMIA Symposium 02/2002;
Source: PubMed

ABSTRACT We developed and evaluated a UMLS Metathesaurus Co-occurrence mining algorithm to connect medications and diseases they may treat. Based on 16 years of co-occurrence data, we created 977 candidate drug-disease pairs for a sample of 100 ingredients (50 commonly prescribed and 50 selected at random). Our evaluation showed that more than 80% of the candidate drug-disease pairs were rated "APPROPRIATE" by physician raters. Additionally, there was a highly significant correlation between the overall frequency of citation and the likelihood that the connection was rated "APPROPRIATE." The drug-disease pairs were used to initialize term definitions in an ongoing effort to build a medication reference terminology for the Veterans Health Administration. Co-occurrence mining is a valuable technique for initializing term definitions in a large-scale reference terminology creation project.

  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: The Strategic Health IT Advanced Research Projects (SHARP) program seeks to conquer well-understood challenges in medical informatics through breakthrough research. Two SHARP centers have found alignment in their methodological needs: (1) members of the National Center for Cognitive Informatics and Decision-making (NCCD) have developed knowledge bases to support problem-oriented summarizations of patient data, and (2) Substitutable Medical Apps, Reusable Technologies (SMART), which is a platform for reusable medical apps that can run on participating platforms connected to various electronic health records (EHR). Combining the work of these two centers will ensure wide dissemination of new methods for synthesized views of patient data. Informatics for Integrating Biology and the Bedside (i2b2) is an NIH-funded clinical research data repository platform in use at over 100 sites worldwide. By also working with a co-occurring initiative to SMART-enabling i2b2, we can confidently write one app that can be used extremely broadly. Our goal was to facilitate development of intuitive, problem-oriented views of the patient record using NCCD knowledge bases that would run in any EHR. To do this, we developed a collaboration between the two SHARPs and an NIH center, i2b2. First, we implemented collaborative tools to connect researchers at three institutions. Next, we developed a patient summarization app using the SMART platform and a previously validated NCCD problem-medication linkage knowledge base derived from the National Drug File-Reference Terminology (NDF-RT). Finally, to SMART-enable i2b2, we implemented two new Web service "cells" that expose the SMART application programming interface (API), and we made changes to the Web interface of i2b2 to host a "carousel" of SMART apps. We deployed our SMART-based, NDF-RT-derived patient summarization app in this SMART-i2b2 container. It displays a problem-oriented view of medications and presents a line-graph display of laboratory results. This summarization app can be run in any EHR environment that either supports SMART or runs SMART-enabled i2b2. This i2b2 "clinical bridge" demonstrates a pathway for reusable app development that does not require EHR vendors to immediately adopt the SMART API. Apps can be developed in SMART and run by clinicians in the i2b2 repository, reusing clinical data extracted from EHRs. This may encourage the adoption of SMART by supporting SMART app development until EHRs adopt the platform. It also allows a new variety of clinical SMART apps, fueled by the broad aggregation of data types available in research repositories. The app (including its knowledge base) and SMART-i2b2 are open-source and freely available for download.
    01/2013; 2(1):e11. DOI:10.2196/ijmr.2454
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: In this article, we focus on the may_treat predicate linking drugs and disorders. Such predicate is expressed in RDF format in the VHA National Drug File Reference Terminology (NDF-RT), a specialized medical resource. The DailyMed dataset also contains this predicate, but only in textual form: for each drug there is an indication field that links the drug’s URI to a literal that is a long description (more than 100 words on the average). We show that natural language processing (NLP) techniques can be used to further distil the indication field to extract may_treat predicates. We then move to Web exploration and show how we can apply similar NLP techniques to find may_treat predicates. The diversity in natural language expressions and the embedding of good information among noisy and often redundant data make it a challenge to exploit the Web. Still, we show that it can be used for finding NDF-RT may_treat predicates with comparable success to using DailyMed.
    Workshop on Web Scale Knowledge Extraction, 10th International Semantic Web Conference, Bonn, Germany; 09/2011
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: UMLS (Unified Medical Language System) is a large medical resource which over the years has merged multiple smaller medical resources, mapping different terminologies onto each other, with an ontological backbone supporting it. But as with any large resource, consistency is difficult to maintain as the resource expands. Lack of consistency is problematic if text mining software relies on UMLS knowledge as background resource. Our study focuses on UMLS network of relations to analyse consistency at the boundary between terminology and ontology. We also look into relation signatures, as defined by the semantic groups assigned to its subject and object. Our approach is a simple statistical approach to determine polysemy and ambiguity in the way relations are defined and used. Although simple, we show that such approach is powerful for finding where inconsistency lies within the resource.
    Terminology and Knowledge Engineering, Madrid; 06/2012

Full-text (2 Sources)

Available from
May 20, 2014