Interactive Exploration of Medical Data Sets Heimo Müller1, Kurt Zatloukal1, Marc Streit2, Dieter Schmalstieg2 (1) Medical University Graz, (2) Graz University of Technology Abstract This paper describes an interactive data exploration system for molecular and clinical data in the field of personalized medicine. It addresses the essential but to date unsolved problem of how to identify connections between genetic variants and their corresponding diseases or the response to certain drugs and treatments, respectively. It is, therefore, necessary to connect genetic data and clinical data in order to categorise specific subgroups of patients with certain disease features. The huge amount of data provided by molecular analytical methods (e.g. data on genetic alterations, proteomic or metabolomic data) can only be analysed by applying statistical methods and bioinformatics. However, even standard methods of statistics and bioinformatics fail when the data is inhomogeneous – as is the case with clinical data – and when data structures are obscured by noise and dominant patterns. The structures of large medical data sets is made visible by using so called object and attribute glyph, which can be arranged in a two dimensional space and synchronised with a set of visualization views. 1. Introduction We developed a data exploration system for the “visualization of” and “navigation in” huge molecular and medical data spaces using a specifically designed physical workplace for collaborative analysis of huge inhomogeneous data sets in the application field of personalized medicine. Our system aims for • support hypothesis driven data analysis, • support different contextual views on the data, and • identify hidden correlation in unconnected databases In order to fulfil these goals, the main challenges in this undertaking were the provision of a set of 3D glyphs for the medical problem domain and methods for interactive exploration of medical databases. With the ability to arrange the glyphs in a two-dimensional space utilizing different spatial grammars and to synchronise different visualization through linked views, a medical expert can in the truest sense of the word, travel through his data space.
2. Related Work Due to the huge number and different structures of molecular and medical parameters (e.g. mutations, genetic polymorphisms, epigenetic alterations, gene expression data, data on protein expression and protein modifications, data on metabolites, diagnosis of disease, laboratory parameters, imaging data, treatment, outcome, accompanying diseases, life style etc.) the coupling of clinical metadata and molecular data sets is still an unsolved problem. While research on efficient visual data mining of very large data sets is a current topic of research, the particular approach of simultaneous visualization of molecular and patient-specific clinical data and the use of high-throughput low-latency user interface techniques are novel. Basic research on information visualization and user interfaces typically focuses on isolated aspects of the visual appearance of the user interface, but many of the approaches face problems when attempting to apply these techniques to real world problems. This has led to a stronger demand of research in visualization, which is perfectly addressed by the literally huge (in terms of data size) problems of medical data analysis. Conversely, medical research is concerned with the comprehension of the hidden meaning of medical data, by any means available. The efficiency of a new analysis tool can only be assessed by the expert working with this tool. As the research question is strongly determined by the requirements of medical experts and the visualization algorithms rely on real world data, a workflow was developed within an interdisciplinary approach. A method for the integrated visualization of microarray data is presented by Grinstein and Smrtic [1][2]. Our approach in focus and context interfaces and large screen displays will be particularly based on the research of Lamping and Rao [3] Baudisch et al. [4] Kosara et al., 2002 [5]. Visual data exploration methods on large data sets, especially hierarchical data structures are described by Hege et al [6] Keim and Kirgel [7], Grinstein and Meneses [8]. Hinneburg, Keim and Wawryniuk from the University of Konstanz developed a software solution (HD-Eye) for the visualisation of high-dimensional data