Keywords: Information Visualization, Post-market Surveillance, Network Analysis
The analysis of post-market data at the Food and Drug Administration generally relies on disproportionality methods used to identify statistical associations between medical products and adverse events. These methods can reveal unexpectedly high reporting associations and may trigger a case series review for closer examination of potential safety signals. Case series review is a labor-intensive process for the medical epidemiologists because it requires the detailed review of multiple data sources and often large datasets. Compelling and informative data visualizations allow meaningful explorations and can assist the medical experts in performing such tasks effectively and rigorously. We propose the use of network visualizations to represent the complex relationships in safety data. The application of certain algorithms to the network graphs may reveal hidden patterns and emphasize specific dimensions of these relationships. We further demonstrate the applicability of network visualizations to other subdomains by using the example of gene expression analysis; we discuss a particular network plot that offers an interesting alternative to the traditional heat maps with the gene and array dendrograms.