Date of Award
Master of Science
This thesis presents the design and implementation of a system to discover the semantically related networks of drug-disease associations, called DDNet, from medical literature. A fully functional DDNet can be transformative in identification of "drug targets" and may new avenues for "drug repositioning" in clinical and translational research. In particular, a Local Latent Semantic Analysis (LLSA) was introduced to implement a system that is efficient, scalalble and relatively free from systemi bias. In addition, a query-based sampling was introduced to find representative samples from the "ocean of data" to build model that is relatively free from "garbage-in garbage-out syndrome. Also, the concept of mapping ontologies was adopted to determine the relevant results and reverse ontology mapping were used to create a network of associations. In addition, a web service application was developed to query the system and visualize the computed network of associations in a form that is easy to interact. A pilot study was conducted to evaluate the performance of the system using both subjective and objective measures. The PahrmGKB was used as the gold standard and the PR curve was obtained from a large number of queries at different recall points. Empirical analyses suggest that DDNet is robust, relatively stable and scalable over traditional Global LSA model.
dissertation or thesis originally submitted to the local University of Memphis Electronic Theses & dissertation (ETD) Repository.
Muthukuri, Karththikka Ramani, "Query Based Sampling and Multi-Layered Semantic Analysis to find Robust Network of Drug-Disease Associations" (2013). Electronic Theses and Dissertations. 812.