Transactions on Machine Learning and Data Mining (ISSN: 1865-6781)
Volume 3 - Number 1 - July 2010
An Approach for Data Mining in Drug Development based on Decision Tree Clustering and Classification
S. Schmiedel, H. Zhang, O. Machnow and P. Perner
Institute of Computer Vision and Applied Computer Sciences, Leipzig, Germany
Abstract
Cellular assays are highly recommended scientific tools in modern drug research and examination of pathological processes at the cellular level. Based on a cellular assay, the internal mitochondrial movement of cells is studied. A goal of the current application is to identify “dynamic signatures” involving key parameters of mitochondrial movement and distribution as a rapid screening means for neuronal degeneration or neurodegenerative diseases. The visual appearance of the different transition stages of a cell is not known a-priori. Therefore, we use clustering by decision tree induction in order to discover groups of similar cells. The available data are then labeled with the group name in accordance with their group affiliation. Finally, the classification model is learned based on decision tree induction. In this paper, we present the methodology and the achieved results. Finally, we give an outlook on further work.