Skip to content

Latest commit

 

History

History
71 lines (49 loc) · 3.96 KB

model-overview.md

File metadata and controls

71 lines (49 loc) · 3.96 KB

Overview of the model

This is WIP. Anything wrong or missing? Please improve and open a PR!

Basic conceptual structure

Geographical space is divided into cells

This is a spatial model. We divide a country into cells and microcells (9x9 microcells to a cell) which are geolocated.

People live in households located in cells

People are allocated according to population density data (from input files) to cells. People have an age, and other attributes. People's residence location does not change, but they interact with people in other cells via places (see below) and via random social interactions governed by a spatial kernel function.

Each person is modelled using the 'Person' structure. Persons are held in a 1-dimensional array : 'Hosts'.

People are associated with different civil institutions

People are assigned to places (institutions such as households, offices, schools etc.) that have a geographical location. Place groups which divides places into compartments (the intent here is that you're less likely to be infected by someone in the same office but who works on a different floor).

Each place is modelled using the 'Place' structure. Places are held in a 2-dimensional array : 'Places'.

The first index of Places[][] corresponds to the type of place, the second index picks a specific Place of that type. For example if the index of the 'Hospitals' place type were 5, then Places[5][2] would be the third Hospital (array indices start at 0)

People don't move. Instead the simulation employs spatial mixing probability distributions (spatial kernels) that control the probability that people in cell X will infect people in cell Y located in another spatial region.

Infections spread between people

Infections may be initially seeded in different ways. The simplest way is to seed according to population density (but seeds can be from specific places, or randomly etc.)

InfectSweep is the main function where infections spread. It loops over people and transmits infections by calculating a FOI (force of infection). Infection-spreading is divided into 3 transmission mechanisms:

  • household infections (e.g. between family members)
  • place infections (e.g. at work)
  • spatial infections (e.g. when travelling around)

Spatial infection models contacts between individuals which have a frequency which depends upon the distance between home locations (to avoid literally moving people around cells), modelled using a kernel function that weights according to both spatial distance and population densities.

Further information

For more information on the model and associated interventions, please see the following papers:

Please note that some of the above may require a subscription.