Host-specific HCV evolution and response to the combined interferon and ribavirin therapy.
ABSTRACT Machine-learning methods in the form of Bayesian networks (BN), linear projection (LP) and self-organizing tree (SOT) models were used to explore association among polymorphic sites within the HVR1 and NS5a regions of the HCV genome, host demographic factors (ethnicity, gender and age) and response to the combined interferon (IFN) and ribavirin (RBV) therapy. The BN models predicted therapy outcomes, gender and ethnicity with accuracy of 90%, 90% and 88.9%, respectively. The LP and SOT models strongly confirmed associations of the HVR1 and NS5A structures with response to therapy and demographic host factors identified by BN. The data indicate host specificity of HCV evolution and suggest the application of these models to predict outcomes of IFN/RBV therapy.
Conference Paper: The EQ framework for learning equivalence classes of Bayesian networks[Show abstract] [Hide abstract]
ABSTRACT: This paper proposes a theoretical and an algorithmic framework for the analysis and the design of efficient learning algorithms which explore the space of equivalence classes of Bayesian network structures. This framework is composed of a generic learning model which uses essential graphs and more general partially directed graphs in order to represent the equivalence classes evaluated during search, operational characterizations of these graphs, processing procedures and formulas for directly calculating their score. The experimental results of the algorithms designed within this framework show that the space of equivalence classes may be explored efficiently and with better results than the classical search in the space of Bayesian network structuresData Mining, 2001. ICDM 2001, Proceedings IEEE International Conference on; 02/2001
- [Show abstract] [Hide abstract]
ABSTRACT: Within biomedical data analysis, visualization can greatly improve data understanding and support various data mining tasks. The pa-per presents FreeViz, a visualization technique for analysis of class-labelled, multi-dimensional data. FreeViz visualizations can present data on many features in the same graph, but through optimization procedure choose a projection that best separates instances of different class. The paper gives mathematical foundations of Free-Viz, and presents its utility on various biomed-ical data sets, including those with thousands of features from cancer gene expression studies.IDAMAP, Aberdeen, UK; 01/2005