Research topics

Research into rare types of cancer

Within the framework of the FFG COIN project VISIOMICS, a web-based platform for integrated exploration of multi-omics and clinical data was created by VRVis in cooperation with St. Anna Kinderkrebsforschung (Paediatric Cancer Research) and other partners in Vienna, which combines visual and automatic data analysis. A research objective is the classification of tumours as well as patient-specific risk stratification and survival prediction. Full integration of multi-scalar genome-wide (multi-omics) data sets, such as cellular features, genomics, gene expression and/or epigenetic data with clinical data, supports exploration into types of cancer in the patient context, therefore helping to improve individual cancer diagnosis precisely in rare types of cancer. While machine and deep learning approaches depend on a larger data stock to learn models, the platform developed by VRVis for visual data exploration represents an alternative especially for rare types of cancer (for which there are only few data sets available) so that researchers can analyse complex relationships and develop new hypotheses.

  • S. Alemzadeh, F. Kromp, B. Preim, S. Taschner-Mandl, and K. Bühler, “A Visual Analytics Approach for Patient Stratification and Biomarker Discovery,” presented at the Eurographics Workshop on Visual Computing for Biology and Medicine, 2019.

Improvement in the statistical significance of long-term studies

Medical long-term studies are often concerned with the relationships between risk factors and incidence of diseases. Large volumes of data are accumulated yet, at the same time, one of the biggest challenges is the lack of data. The dropout of patients and non-recorded values are among the main statistical problems in analysing long-term studies. If they are not correctly handled, they cause a bias in data analysis, thus reducing the statistical significance and therefore the value of the results. Together with epidemiologists at the University of Greifswald, a visual analytics application was developed with VIVID, and this allows missing values to be traced in clinical cohort studies and their character determined. VIVID also integrates methods to replace missing data and to evaluate its validity. VIVID was developed by our scientific partners at the University of Magdeburg in collaboration with VRVis.

  • S. Alemzadeh, U. Niemann, T. Ittermann, H. Völzke, D. Schneider, M. Spiliopoulou, K. Bühler, and B. Preim, “Visual Analysis of Missing Values in Longitudinal Cohort Study Data,” Computer Graphics Forum, vol. 45, no. 3, pp. 1–13, May 2019.