Dynamic maps: a visual-analytic methodology for exploring spatio-temporal disease patterns

Abstract

Background: Epidemiologic studies are often confounded by the human and environmental interactions that are complex and dynamic spatio-temporal processes. Hence, it is difficult to discover nuances in the data and generate pertinent hypotheses. Dynamic mapping, a method to simultaneously visualize temporal and spatial information, was introduced to elucidate such complexities. A conceptual framework for dynamic mapping regarding principles and implementation methods was proposed.

Methods: The spatio-temporal dynamics of Salmonella infections for 2002 in the U.S. elderly were depicted via dynamic mapping. Hospitalization records were obtained from the Centers of Medicare and Medicaid Services. To visualize the spatial relationship, hospitalization rates were computed and superimposed onto maps of environmental exposure factors including livestock densities and ambient temperatures. To visualize the temporal relationship, the resultant maps were composed into a movie.

Results: The dynamic maps revealed that the Salmonella infections peaked at specific spatio-temporal loci: more clusters were observed in the summer months and higher density of such clusters in the South. The peaks were reached when the average temperatures were greater than 83.4 degrees F (28.6 degrees C). Although the relationship of salmonellosis rates and occurrence of temperature anomalies was non-uniform, a strong synchronization was found between high broiler chicken sales and dense clusters of cases in the summer.

Conclusions: Dynamic mapping is a practical visual-analytic technique for public health practitioners and has an outstanding potential in providing insights into spatio-temporal processes such as revealing outbreak origins, percolation and travelling waves of the diseases, peak timing of seasonal outbreaks, and persistence of disease clusters.

Figure 1

The re-aggregation scheme. The chart explains the chain of decisions made when aggregating counties with low elderly population in this study. Arrows with "Y" indicates "Yes"; "N" indicates "No".

Figure 2

Natural breaks classification using logarithmic transformation. A. Natural breaks classification on untransformed Salmonella hospitalization rate data, B. Natural breaks classification on log transformed Salmonella hospitalization rate data.