Knowledge Discovery in the Form of Prototypical Cases Using Advanced Data Mining Techniques

Knowledge discovery in the form of prototypical cases using advanced data mining techniques Rathnavel Rajagopal Chair of the Supervisory Committee: Asst. Professor Isabelle Bichindaritz Computing and software systems This thesis applies advanced data mining techniques to obtain structured and abstract knowledge structures (or prototypical cases) from clinical data available in a certain medical domain. The research question addressed entails investigating if advanced knowledge structures in the form of prototypical cases can be generated from clinical data using advanced data mining techniques. Relational data mining techniques are applied on clinical data, reverse engineered from expert acquired prototypical cases. Knowledge structures in the form of clinical pathways, or prototypical cases, generated is compared to the expert acquired prototypical cases that is considered as the source for this thesis. Prototypical cases, also called clinical pathways, are the cornerstone of the knowledge acquisition process in the Memoire project, which is a framework for the development of Case-Based Reasoning (CBR) systems in biomedicine. These clinical pathways correspond to clinical diagnostic categories for the most part, some of them corresponding also to essential signs and symptoms requiring specific assessment or treatment actions for a disease. These knowledge structures are represented from ontologies such as the Unified Medical Language System (UMLS) as well as domain specific ontologies. The results obtained from this thesis will be presented and discussed. These results in the form of prototypical cases, which are generated by relational mining techniques, are validated as well.