Interactive Visual Exploration of Multivariate Medical Measurements in Diabetes Care

Lead / Contact  Wolfgang Aigner

Team  Wolfgang Aigner, Vienna University of Technology
Michael Atanasov, HTBL Krems
Alexander Rind, Vienna University of Technology
Philipp Schindler, HTBL Krems
Reinhardt Wenzina, HTBL Krems

Partners  Vienna University of Technology, Institute of Software Technology & Interactive Systems
HTBL Krems, Department of Information Technology

Duration  September 2010 – May 2011


Multivariate, time-oriented data plays a central role in many domains such as medicine, finance, or environmental engineering. Information Visualization can be an instrument to make such vast datasets intuitively comprehensible. As the number of time series increases, visualizations need to be space-efficient. Horizon graphs [Reijner, 2008] are an innovative approach to reduce required space by dividing the a chart into bands and layering these. While a first evaluation [Heer, et al., 2009] shows promising results, there is still room for optimization, especially regarding interaction techniques.

In the project HorizonVis, we implemented the horizon graph as component for the VisuExplore framework. HorizonVis was tested with the diabetes dataset from the UCI Machine Learning Repository. In order to support flexible usage, it provides several customization settings:

  • indexing values relative to a specific indexing point versus absolute values
  • number of bands [Heer, et al., 2009]
  • spread of a band, base value
  • mirroring versus offset for values below the base value [Heer, et al., 2009]
  • linear interpolation versus step chart
  • different color schemes
  • showing a legend [Saito et al., 2005]

Future work includes evaluation of HorizonVis in application scenarios (e.g., diabetes care). The applicability of indexing is of special interest. Furthermore, we plan to extend HorizonVis with advanced interaction techniques.

Images  To retrieve an enlarged version of the images, click on them:

HorizonVis with four variables compared to a line plot of the same variables and separate line plots


Customizing HorizonVis: (1) Indexing, (2) absolute values, (3) offset instead of mirroring, and (4) step chart mode


Glucose readings of ten patients: Comparing (1) indexing versus (2) absolute values


Creative Commons License These three screenshots are licensed under a Creative Commons Attribution-NoDerivs 3.0 Unported License.

Related Work  VisuExplore – Gaining New Medical Insights from Visual Exploration

MIDGAARD – Connecting Time-Oriented Data and Information to a Coherent Interactive Visualization

TimeViewer – Visualizing Temporal Data using prefuse

J. Heer, N. Kong, and M. Agrawala, Sizing the Horizon: The Effects of Chart Size and Layering on the Graphical Perception of Time Series Visualizations, in Proc. 27th Int. Conf. Human Factors in Computing Systems (CHI’09), Boston, MA, 2009, p. 1303–1312.

H. Reijner, The Development of the Horizon Graph, in Electronic Proc. Vis08 Workshop From Theory to Practice: Design, Vision and Visualization, 2008.

T. Saito, H.N. Miyamura, M. Yamamoto, H. Saito, Y. Hoshiya, and T. Kaseda, Two-Tone Pseudo Coloring: Compact Visualization for One-Dimensional Data, in Proc. IEEE Symp. Information Visualization (InfoVis 2005), 2005, p. 173–180.

Funding  This project was funded by the Sparkling Science initiative of the Federal Ministry of Science and Research. HorizonVis was awarded € 5000 for a top 5 submissions to the third call.

official project description (in German)