Spatial Analysis for Epidemiology. Methods and Tools

This article details the different methods used in spatial analysis for epidemiology. It follows on from [MED 4 001] , which sets out the conceptual foundations. The two-dimensional nature of geographical location means that standard representation methods and statistical techniques for dealing with essentially univariate sets of variables need to be supplemented by more sophisticated methods. The term "spatial analysis" is thus used to describe analysis techniques, essentially statistical, that make use of geographical location. Spatial analysis of a health phenomenon mainly involves analyzing and characterizing the spatial distribution of the phenomenon, in order to identify risk factors and, if possible, characterize the processes that determine this spatial distribution, so that they can be modeled. Spatial analysis also involves introducing spatial relationships (and in particular spatial auto-correlation) into statistical risk models.

Descriptive spatial analysis includes cartographic analysis, detection of geometric and spatio-temporal characteristics, analysis of the spatial variability of a value, detection of aggregates (clusters), search for analysis and synthesis scales, analysis of environmental correlations. Explanatory spatial analysis is essentially statistical, with the identification of statistical models involving spatial relationships between individuals and between individuals and the characteristics of their environment.