Supporting medical decision support process with knowledge-based representation of microrna experiments

Recent technological advancements related with medical informatics has led to progress in both personalized medical support approach and the analysis of gene expression. It has been shown that gene expressions can be used to diagnose disease and identify similar cases. Besides the studies based on the statistical comparison of gene expressions, it is anticipated that the incorporation of the domain knowledge into the analyzes will make a positive contribution to the accuracy of research. Within the scope of this thesis, it is aimed to increase the success of using miRNA gene expressions to diagnose disease and to retrieve relevant cases. For this purpose, a set-based representation of experiments is proposed, which is based on the analysis of miRNA expressions together with miRNA sets obtained from the literature. Within the scope of the proposed approach, a browser-based tool, which is named as miSEA (http://binf.baskent.edu.tr), that is available to the use of researchers has been developed. The proposed approach has been compared with traditional methods and it has been found that performance improvement is achieved in classification by using SVM algorithm. Also, experiment retrieval performance improved by using Euclidean algorithm. In addition, a cross-technology experiment retrieval approach was proposed and performance analysis were conducted to analyze current gene expression profiling technologies. Moreover, a proposal has been made to use the modules including the implemented system and the suggested approaches in decision support systems.