| Summary: | In clinical research, knowledge and data are often recorded in free-text format which is difficult to access reliably because the variety of expression is vast. The most comprehensive clinical vocabulary currently available is SNOMED CT (Systematized Nomenclature of Medicine-Clinical Terms) which has become widely used in the clinical field. However, a critical issue that accompanies this extensive clinical terminology; an accurate complete set of concepts for classification purposes is difficult to obtain from the provided clinical information. In addition, clinical notes are provided in free-text format, which may be very difficult to match directly to SNOMED CT concepts. This thesis proposes a novel fuzzy ontology approach that codes clinical notes for SNOMED CT. The aim is to realise effective mapping of clinical notes to SNOMED CT. The fuzzy ontology methodology creates a fuzzy subset that reduces the ontology size and then combined with the fuzzy ontology approach. To render the fuzzy subset of SNOMED CT suitable for professional clinical use, the study incorporates several other tools and methods. The clinical notes are processed by a parser to reduce their length and thereby increase the accuracy of the related concepts found in SNOMED CT. To improve the quality of the fuzzy subset, the similarity is checked by calculating the Levenshtein distance. A future area of the study in crowdsourcing approach which has a potential to increase the accuracy and the usability of the fuzzy subset is also proposed and discussed in the end of this thesis.
|
|---|
