The use of acronyms and abbreviations is increasing profoundly in the clinical domain in large part due to the greater adoption of electronic health record (EHR) systems and increased electronic documentation within healthcare. A single acronym or abbreviation may have multiple different meanings or senses. Comprehending the proper meaning of an acronym or abbreviation using automated machine techniques, known as word sense disambiguation (WSD), in clinical notes is an essential step for medical natural language processing (NLP) systems. While acronym and abbreviation WSD from the biomedical literature is an active area of investigation, little research has been done on this topic with clinical documents.
The purpose of this dissertation is to develop automatic WSD tools for clinical acronyms and abbreviations. A key step toward this end is to build a comprehensive clinical sense inventory based upon the integration of available biomedical resources and upon senses from a large corpus of clinical notes. Another complementary task is the performance maximization of machine learning (ML) algorithms. This includes the development of optimal feature sets for WSD and the exploration of minimum "adequate" sample size for training classifiers. These automatic WSD technologies extend to the complementary problem of symbol disambiguation in clinical texts. Lastly, the anticipated future work will be in developing quality improvement of automatic WSD tools including sense amelioration utilizing biomedical knowledge.
University of Minnesota Ph.D. dissertation. December 2012. Major: Health Informatics. Advisor: Serguei V.S. Pakhomov, PhD. 1 computer file (PDF); vii, 89 pages, appendices p. 89.
Automatic word sense disambiguation of acronyms and abbreviations in clinical texts.
Retrieved from the University of Minnesota Digital Conservancy,
Content distributed via the University of Minnesota's Digital Conservancy may be subject to additional license and use restrictions applied by the depositor.