Abstract: Epigrass is a open-source simulation platform created to study epidemics and their spatial (geographic) dynamics. It is licensed under the GPL-v2

Epigrass was developed as a scientific project, by the founders of Metamodellers at the Oswaldo Cruz Foundation. Currently, Metamodellers is the main maintainer of the code, ensuring its continuous improvement while remaining a completely ... free tool.

In Epigrass, space is represented as a network of interconnected localities. Therefore, the concepts of nodes and edges, from mathematical graph theory is used to analyze the properties of the spatio-temporal dynamics of models constructed in Epigrass. The spread of a transmissible disease, for example, becomes a tree-shaped subgraph of the complete spatial graph.

Models in Epigrass can be quickly specified through a small set of text files editable on any standard tex editor. But Epigrass also offers GUI-based model editing tools to help with model definition.

Since space in Epigrass is real geographic space, Epigrass can Import maps in shapefile format which are then used in the models. But usage of such maps is not obligatory as Epigrass also allows the definition of models based on non-georreferenced networks.

However, when maps are provided, the results are generated in shapefile and KML format, which can then be openned and visualized on any GIS tool including Google Earth and Google Maps.

Epigrass has its own built-in map and graph visualizer facilitating the quick inspection of spatio-temporal simulations which Epigrass stores on RDBMS such as SQLite or MySQL.

Epigrass also provides a built-in report generator, which generates LaTeX and PDF output with some basic analyses of the models and that can be easily extended by the user’s own analises.

The spread path of epidemics is output as a standard GraphML format which can then be read an displayed on most graph visualization tools available today.

Epigrass models are a combination of intra and Inter-node dynamics defined by the user.

Intra-node dynamics are determined by a user selected epidemic model, which is replicated at all nodes but parameterized with the unique characteristics of each locality. But the user doesn’t have to limit his/her model to the built-in epidemic models, he/she can write a custom node model, in pure Python, in which its is possible to define any mechanism, thus freeing the model even from representing epidemics. it can represent anything the user conceives.

Inter-node dynamics is based in a flow model between nodes, parameterized individually for every edge of the network. This flow can be of people (in an epidemic model) but can represent any exchange of information between nodes, specified by the user.