SPACE TIME VISUALIZATION FOR EPIDEMIOLOGICAL RESEARCH
M.J. Kraak, P. Madzudzo
ITC – Enschede, the Netherlands
In epidemiological research incorporating the concept of time is very important as it can result in an improved perspective on the progression of epidemic events in relation to space and time. Health scientists would like to understand spatiotemporal relationships and associated patterns of such events to be able to take appropriate actions. The visualization of the distribution and propagation of the epidemics can be very supportive in this process. However, previous attempts to visualize temporal relationships in health studies had been hampered by several limitations. After a study of epidemiological and related geoscience literature it proved that the often erratic and incomplete data collections, the availability of suitable data representations, as well as the problem to disseminate clear results had a negative influence on research activities.
This paper suggests an alternative visualization method for highlighting spatiotemporal relationships in epidemiological research by putting the space-time-cube in the centre of multiple coordinate view environment. In this approach, the traditional space-time-cube, which includes space (x-y plane) and time (the z-axis), is extended with the third spatial dimension to represent the terrain. The other views include the parallel coordinate plot, the scatter plot and bar graphs. Data from the Black Death case study recently made available was used for showing the usability of suggested approach for health studies. This Black Death data has interesting characteristics since at that time, when it occurred in the 14th century, there was no proper recording of events resulting in data characterized by uncertainty and incompleteness. In our approach we tried to incorporate these deficiencies into temporal visualizations which are based on statistical and other computational methods. To allow more insight into the data, the alternative visualization approach provides for user interaction with the data and the views offer brushing functionalities. The studied approach has been compared with existing traditional visual representations. In addition work was done to make the approach also useful for real time data entry to allow health scientist to monitor today’s epidemics as well.