Input data files¶
EMOD accepts the following categories of input data files that contain the relatively fixed information about the population within each geographic node. For example, the number of individuals, geographic data, climate, demographics, and migration data. This is in contrast to the demographic, geographic, and migration parameters in the configuration file that control simulation-wide qualities, such as enabling air migration across all nodes in the simulation.
The Institute for Disease Modeling (IDM) provides collections of input data files for many different locations in the world for download on GitHub. Except for the demographics file, you will typically use these input data files in their default state. See Use input data files for more information.
Demographics files are JSON formatted files containing information on the demographics of the population in a geographical region to simulate. For example, the number of individuals and the distribution for age, gender, immunity, risk, and mortality.
In addition, demographics files are useful for creating heterogeneous groups within a population. You can define values for accessibility, age, geography, risk, and other properties and assign individuals to groups based on those property values. For example, you might want to divide a population up into different bins based on age so you can target interventions to individuals in a particular age range. Another common use is to configure treatment coverage to be higher for individuals in easily accessible regions and lower for individuals in areas that are difficult to access.
EMOD assumes homogeneous mixing and disease transmission for the generic simulation type. You can use the HINT feature to add heterogeneous transmission to your generic model. You cannot manually configure heterogeneous transmission using HINT with other simulation types because the heterogeneity in transmission for specific diseases and disease classes is already configured by the simulation type.
You can specify multiple demographics files, which function as a “base layer” file and one or more “overlay” files that override the base layer configuration. Overlay files can change the value of parameters already specified in the base layer or add new parameters. Support for multiple demographics layers allows for the following scenarios:
- Separating different sets of parameters and values into individual layers (for example, to separate those that are useful for specific diseases into different layers)
- Adding new parameters for a simulation into a new layer for easier prototyping
- Overriding certain parameters of interest in a new layer
- Overriding certain parameters for a particular sub-region
- Simulating subsets of a larger region for which input data files have been constructed
Migration files describe the rate of migration of individuals in and out of a geographic node. There are four types of migration files that can be used by EMOD: local migration, regional migration, air migration, and sea migration.
For all types, migration data is contained in a set of two files, a JSON metadata file with header information and a binary data file with the actual migration data. Both files are required. The basic file structure is identical for all types, with the only exception being the number of columns per row allowed to each type.
Local migration describes the foot travel of people into and out of adjacent nodes. A local migration file is required for simulations that support more than one node.
Regional migration describes migration via a road or rail network. If a node is not part of the network, the regional migration of individuals to and from that node considers the closest road hub city. When you create the migration file, you must create a Voronoi tiling based on road hubs of the region, with each non-hub connected to the hub of its tile.
Air migration describes migration via airplane travel. It is usually required for simulations of an entire country or larger geographies.
Sea migration describes migration via ship.
There are two general types of climate files usable by EMOD: climate files generated through actual data, referred to as “climate by data,” and climate files generated from the Koppen classification system, referred to as “climate by Koppen.”
For both types, climate data is contained in a set of two files, a JSON metadata file with header information and a binary file that contains the actual climate data. Both files are required.
Climate by data¶
Climate by data files contain real data gathered from weather stations in the region to be simulated. This includes rainfall, temperature, relative humidity, and so on.
When running a simulation on a multi-core HPC cluster, you can include a load balancing file to control how the computing load is distributed across the cluster. The load balancing file is a binary file that allocates the simulation of each geographic node to cores in the cluster. If no file is submitted, EMOD allocates nodes to cores according to a checkerboard pattern.