Related Experiment Videos
A large-scale evaluation of terminology integration characteristics
F S McDonald1, C G Chute, P V Ogren
1Mayo Foundation, Rochester, MN, USA.
Proceedings. AMIA Symposium
|November 24, 1999
Summary
Automated Term Composition effectively maps local medical terms to large vocabularies like UMLS. Semantic type filtering enhances accuracy, ensuring practical integration of specialty and general terms into health systems.
Area of Science:
- Medical Informatics
- Natural Language Processing
- Health Terminology
Background:
- Local terminologies present challenges for integration into large-scale health vocabularies.
- Standardized mapping is crucial for data interoperability and analysis in healthcare.
Purpose of the Study:
- To evaluate the effectiveness of Automated Term Composition (ATC) for mapping local specialty-specific and general terms to the Unified Medical Language System (UMLS).
- To assess the impact of semantic type filtering on the accuracy of terminology mapping.
Main Methods:
- Compared sensitivity, specificity, and predictive values of ATC using Metaphrase for 9050 dermatology and 4994 general terms against UMLS.
- Analyzed results from exact matches, semantic type filtered matches, and non-filtered matches.
- Evaluated the effect of semantic type filtering on general terms.
Main Results:
- Exact matches showed 51% sensitivity and 86% specificity for dermatology terms.
- Semantic type filtering improved sensitivity to 75% (dermatology) and 88% (general) while maintaining high positive predictive values (95.1% dermatology, 98.4% general).
- Non-filtered matches increased sensitivity but significantly decreased specificity.
Conclusions:
- Automated methods, particularly with semantic type filtering, are practical for mapping local terminologies to controlled health vocabularies.
- Semantic type filtering balances sensitivity and specificity, achieving high positive predictive values for accurate term integration.