| Traditionally, physicians rely on medical knowledge learned or acquired through practice to make a correct decision in diagnosis of diseases and affected medications. Progress in knowledge base systems and related Information Technology changed this situation by providing huge amount of medical information that can support physicians in the decision making. Integrating this great amount of information to retrieve the superlative results is however a demanding job. By introducing the concepts of Semantic Web, the sources of medical knowledge have amended to acquire the advantages of this concept and its technologies in gathering and representing the information and recommending superior decision from comprehensive information. Particularly, ontologies are recognized to enhance the efficiency of information management significantly and improve the dependability of communication especially when heterogeneous actors and diverse environments are involved.;In this thesis, various Semantic Web techniques have been employed to support data integration to assist physicians in the process of drug recommendation. The thesis proposes a novel approach for supporting drug recommendation decisions by modeling a Semantic Web-based infrastructure correlating comprehensive medical knowledge which allows making ontology inferences and knowledge discoveries. In this work, we devise a Disease-Drug Ontology (DDO), an ontological model which demonstrates relations between human diseases and their relevant drugs and medications. The DDO, formalized in OWL, allows the integrated representation of various sources of ontologies and data schemas and overcomes the heterogeneity problem among these different sources by applying proper matching techniques. An automated reasoning is performed over the ontology using a Description Logic Reasoner in order to validate the DDO. Our model is also composed of an ontology crawler that provides physicians, by direct queries from DDO, to facilitate the process of making decisions for accurate drug recommendation. More importantly, our system consists of a unique rule-based inferential engine employing drug rules and patient data for the purpose of suitable drug recommendation. In order to prototype the key services of the system and reveal the validity of our semantically integrated Disease-drug knowledge base, some case studies are provided and the obtained results are very promising. |
