Summer Scholarship Programme

	SSP Project Summary: Spatial Interaction Modelling of Epidemics

[EPCC home] [SSP home] [2001 projects] [2000 projects] [1999 projects] [1998 projects] [1997 projects] [1996 projects] [1995 projects] [1994 projects] [1993 projects]

Student

Jennifer Market, Texas A & M University, USA

Supervisor

Elspeth Minty, EPCC

Simulating the dynamics of diseases provides a means of understanding their spread through a population. Whilst the model to describe the effect of diseases on isolated populations is generally accepted, such types of populations no longer exist in the modern world. This has lead to a growing interest in the spatial aspects of disease dynamics.

Models have been developed to look at the effect of small numbers of infectious individuals moving between sub-populations of the main population group. However due to computational limitations, the number of sub-populations which it is possible to look at is small (~10). With this small number of sub-populations, unrealistic features are observed, e.g., synchronisation between sub-populations leads to the disease dying out, which does not occur in the real world.

The aim of this project is to develop an MPI version of one of the simpler models of disease dynamics, with the aim of modelling a larger number of sub-populations. The model to be used is the SEIR (Susceptable, Exposed, Infectious, Recovered) model. This model divides the population into 4 groups:

susceptible: people who can catch the disease.
exposed: people who are infected but not yet infectious.
infectious: people who are capable of spreading the disease.
recovered: people who are immune to the disease.

Members of the population enter into and leave the simulation (i.e., are born and die), and also move between the different groups. By modelling the flow of the population between the different groups over a period of time, the spread of disease can be monitored. To simplify the simulation, it will be assumed that the disease in the simulation is not fatal, so that the disease does not influence the death rates. In addition, we assume that the birth and death rates are equal so that the population size remains the same.

The population is divided into a number of sub-populations which live on the computation grid. A stochastic model will be used to implement SEIR by generating events probabilistically across the grid. The result of these events are used to update the values of the different groups. For example, if a birth occurs the number of people susceptible to the disease would increase by one.

In part this project is a reworking of a previous SSP project (EPCC-SS95-05) implemented on the CM-200. While the program developed in the previous project is unlikely to help with the proposed project, the student considered many aspects of the parallelisation which are relevent to this project. Thus an MPI implentation should be relatively straight forward, and it is expected that the student will implement additional features which were not included in the original project such as extending the sub-population interactions beyond nearest neighbour.

In addition to developing the an MPI implentation of the stochastic SEIR model, DIVE will be used to visualise the results, and hopefully a video produced of the simulation.

The work proposed in this project will form the basis of an application demonstrator to illustrate the application of HPC in this area, and is part of the EPCC-TEC project CACHE.

The final report for this project is available here.

Webpage maintained by mario@epcc.ed.ac.uk