An evaluation of medical knowledge contained in Wikipedia and its use in the LOINC database
|An evaluation of medical knowledge contained in Wikipedia and its use in the LOINC database|
|Authors:||Jeff Friedlin, Clement J. McDonald|
|Citation:||Journal of the American Medical Informatics Association 17 (3): 283-287. 2010 May|
|Database(s):||Google Scholar cites PubMed (PMID/20442145)|
|Web:||Bing Google Yahoo! — Google PDF|
|Article:||BASE Google Scholar PubMed|
|Restricted:||DTU Digital Library|
|Format:||BibTeX Template from PMID|
|Extract:||Talairach coordinates from linked PDF: CSV-formated wiki-formated|
The rearchers describe a Java software tool, Wikipedia to LOINC matcher (WLMA), for mapping Wikipedia articles with LOINC parts.
The tool has a number of components:
- Handling of Wikipedia disambiguiation pages
- Examination of article category
- Use of Unified Medical Language System to find the semantic type of LOINC parts
- Extraction of the introduction sectio of Wikipedia articles
- Exclusion of LOINC parts with "derivatives", such as "AB", "AG", "RNA" and "IgM"
- 1705 LOINC parts were used for the matching
- 1416 matches, 289 no matching
- 1299 complete matches (true positive), 15 partial matches and 102 non-matches
- 15 of the 289 no matching was found to match Wikipedia articles with manual query (false negative)
That is, 1299 +15 +15 = 1329 out of 1705.
The description from Wikipedia was made part of LOINC.