Vis-Saúde - Uma metodologia para visualização e análise de dados de saúde pública

Abstract

Computer systems provide a set of tools that allow the analysis and extraction of knowledge in a large and complex set of variables. The set of data that form an event, in a first analysis under numerical form can not show any important information to support a deep study of the data. Although the use of visualization techniques with data analysis methodologies can allow a best comprehension about an event. Possible correlations between variables can be evidenced by the use of methodologies of analysis that open a set of new possibilities of research. The use of data visualization techniques according to the context, moment and set of variables can show important information about an event that was hidden by numbers. Vis-Health shown in this paper used open source technologies to develop an analysis and visualization methodology of data in an web environment, where if the data are available, the registers can be processed and the result shown in real time. Efforts to utilize analysis and visualization methodologies in a set of public health data are relevant to extract knowledge of the complex data set available with the goal of guiding the population and public health managers. The methodology analysis used in this paper is based on principal component analysis with the aim to find the dependency relationship between an event of interest and other related variables available for study. The use of data visualization techniques enhances the possibilities of comprehension of the information that is being studied. Vis-Health provided two distinct georeferenced views of health records, where both are based in the region where the event occurred and displayed on a geographical map for easy identification of the region of interest. The first visualization is showing the result of the dependency analysis between the variables of study. The second one has the aim to make more dynamic the comprehension about the incidence of a disease through the presentation of concentration of events occurred using heat map. In this visualization the data are grouped in temporal series and displayed with an animation with temporal passage. The public health is a very dynamic segment of government administration that is always threatened with endemic or epidemic situations. This behavior requires a constant monitoring of managers to identify serious situations that can require more attention and resources. With the goal to demonstrate the operation of the methodology with real data, the registers about dengue incidence in Brazil were used. Dengue is a disease present in all Brazil’s states and represents a serious risk to the collective health considering that the incidence has increasing in the latest years.