A GIS is essentially a spatial database that holds specifications of the spatial elements of a population (who lives where), or features of the physical landscape, such as roads or green spaces, or the built environment, such as food outlets (Crooks et al., 2019). Within ABMs, GIS information is often represented using one of two approaches; (i) raster data (that is, a large number of square cells, and then attributes are assigned to them; typically applied to environmental applications) or (ii) vector data (points, lines and polygons). The latter is the more ‘popular’ choice as this format (commonly termed shapefiles) allows the physical environment of roads, buildings and other urban features to be readily represented in fine granularity. With recent developments in popular platforms, such as Netlogo or GeoMason, it is possible to import shapefile layers directly into the platform and ABM simulations run directly within them (Crols et al., 2019).
Integrating GIS within ABMs may be useful for health economic modelling if an intervention being assessed is about access to certain places and there is substantial interaction between an element of the physical environment and behaviour, upon which there is some evidence. It could help to explore where places should be located to maximise cost-effectiveness and/ or reduce inequalities. Note, it is possible for ABMs to incorporate more abstract geographical elements in the absence of GIS to inform such policy decisions (Luke et al., 2017).
Undertaking GIS analysis requires geo-referenced data, many of which are online and open-source. These include OpenStreetMap (contains map data including roads, trails, cafes and railway stations) and Natural Earth Data (contains counties and points of interest) (see Crooks et al., 2019 for a discussion of the different formats). There is also open-source software allowing individuals to create, edit, analyse and visualise the geographical data, which includes Quantum GIS and GRASS software. In addition, many of the platforms available for ABMs have the capability to process GIS data. For example, within NetLogo there is a GIS extension, and it is possible to import both raster and vector data (Crooks et al., 2019).