 |
We have implemented an influenza epidemic simulation tool and have made the source code available. The model is based closely on Germann et al 2006. FluTE differs in its implementation (C/C++ instead of C/Python/SPaSM) and several model details to make it more applicable to the novel H1N1 2009 pandemic. |
| FluTE generates text output files, which can be easily processed with use-supplied scripts. In this example, we import the results from a US simulation into ArcGIS to generate a heatmap showing illness prevalence. |  ArcGIS map generated from FluTE output |
A C++ compiler is required to compile the source code. Binaries are not included with the download.
We have found that about 80Mb of RAM are required per 1,000,000 individuals in the simulation. Therefore, to simulate a region with 10 million people, 800Mb are required. Simulating an unmitigated epidemic in a population of 10 million takes about 2 hours. Smaller epidemics (with smaller populations, low R0, and/or effective intervention) can take much less time.
OpenMPI is required to compile the parallel version.
The parallel version is only useful for simulating large populations
(more than one county). In the current version, if you allocate
mpiflute more processors than it can use, it will terminate with
an error. To simulate the entire continental US (about 280 million
people), I usually use 32 processors.
Download