Data Aggregation and Analysis for Cancer Statistics - A Visual Analytics Approach

The disparity between data collected in rural and urban counties is often detrimental in the appropriate analysis of cancer care statistics. Low counts drastically affect the incidence and mortality rates of the data, leading to skewed statistics. In order to more accurately report the data, various levels of aggregation have been used (grouping counties by population, age percentages, etc.); however, such data aggregation methods have often been ad hoc and/or time consuming. Such groupings are performed on a user defined basis; however, grouping based purely on population demographics does not take into account the spatial relationships between data. Furthermore, researchers want to search for spatiotemporal correlations within their data domain. In this work, we introduce a visual analytics system for exploring cancer care statistics in a series of linked views and interactive user interface queries. We also apply the AMOEBA algorithm [1] for clustering counties based on population demographics in a visual analytics environment. Users select the population demographics field on which they wish to cluster, and these county clusters then form the basis for the data aggregation. Such a system allows the user to group their data by fields (age, gender, income) while maintaining spatial structure and provides and interactive mapping system in which to compare and explore such groupings. By utilizing such geographical groupings, we hope to better enhance the underlying structure of the data and help alleviate reporting problems associated with small area statistics.