Document Detail


Truecasing clinical narratives.
MedLine Citation:
PMID:  21893817     Owner:  NLM     Status:  In-Data-Review    
Abstract/OtherAbstract:
Truecasing, or capitalization, is the rewriting of each word of an input text with its proper case information. Many medical texts, especially those from legacy systems, are still written entirely in capitalized letters, hampering their readability. We present a pilot study that uses the World Wide Web as a corpus in order to support automatic truecasing. The texts under scrutiny were German-language pathology reports. By submitting token bigrams to the Google Web search engine we collected enough case information so that we achieved 81.3% accuracy for acronyms and 98.5% accuracy for normal words. This is all the more impressive as only half of the words used in this corpus existed in a standard medical dictionary due to the excessive use of ad-hoc single-word nominal compounds in German. Our system performed less satisfactory for spelling correction, and in three cases the proposed word substitutions altered the meaning of the input sentence. For the routine deployment of this method the dependency on a (black box) search engine must be overcome, for example by using cloud-based Web n-gram services.
Authors:
Markus Kreuzthaler; Stefan Schulz
Related Documents :
1859037 - Effectiveness of an acaricide in management of house dust mite allergy.
22195097 - An evaluation of the umls in representing corpus derived clinical concepts.
16039107 - Comparative improvement of asthma symptoms and expiratory flows after corticosteroid tr...
16140227 - Adherence to asthma controller medication regimens.
18387707 - Medical response to a radiologic/nuclear event: integrated plan from the office of the ...
17578687 - Plumage colour and feather pecking in laying hens, a chicken perspective?
Publication Detail:
Type:  Journal Article    
Journal Detail:
Title:  Studies in health technology and informatics     Volume:  169     ISSN:  0926-9630     ISO Abbreviation:  Stud Health Technol Inform     Publication Date:  2011  
Date Detail:
Created Date:  2011-09-06     Completed Date:  -     Revised Date:  -    
Medline Journal Info:
Nlm Unique ID:  9214582     Medline TA:  Stud Health Technol Inform     Country:  Netherlands    
Other Details:
Languages:  eng     Pagination:  589-93     Citation Subset:  T    
Affiliation:
Institute for Medical Informatics, Statistics and Documentation, Medical University of Graz, Austria.
Export Citation:
APA/MLA Format     Download EndNote     Download BibTex
MeSH Terms
Descriptor/Qualifier:

From MEDLINE®/PubMed®, a database of the U.S. National Library of Medicine


Previous Document:  Roogle: an information retrieval engine for clinical data warehouse.
Next Document:  Checking Coding Completeness by Mining Discharge Summaries.