| Popis: |
Hospitals produce millions of patient records consisting of clinical annotations and there is an extensive usage of abbreviations in these annotations. The data in these clinical annotations are an excellent source for bioinformatics research but the use of abbreviations can create ambiguity. For example the term ‘SMA’ may stand for ‘Sequential Multichannel Autoanalyzer,’ or ‘Smooth Muscle Antibody.’ The main objective of this thesis is the development of a software application that takes a medical acronym as input and accesses medical and pharmaceutical websites to retrieve information from articles containing the acronym along with a user selected full-form of the acronym. The retrieved information consists of article title, authors, publication date, article abstract, journal title and Medical Subject Header (MeSH) details. The tool we develop stores the retrieved information in its internal database to make it available for later use. The MeSH information is used by researchers at the Cincinnati Children’s Hospital Medical Center (CCHMC) as part of a research effort for reducing the ambiguity created by the use of acronyms. Our tools enables CCHMC researchers to investigate hypotheses such as whether a specific MeSH header, occurring along with an acronym/full-form pair in more than 90% of the total number of articles retrieved, supports a conclusion that the MeSH header identifies the pair. Our contribution to this research effort is a framework for disambiguation that accesses online sources and retrieves article and MeSH header information for a user selected acronym/full-form pair. We design and populate an internal database that can be used in future research efforts. |
