Malaria is a disease that aects millions of people each year, with 90% of
deaths occurring in Africa alone. The subject of this dissertation is the
agent based modelling of malaria in populations that are suering mass
migration, living in unsanitary conditions or undergoing other eects of
impoverished circumstances. The research is important due to the large
numbers of people aected by malaria globally, with about 3.3 billion at
risk. A large proportion of those exposed to the risk live in Sub-Saharan
African countries.
Agent based modelling is a type of computational modelling which is com-
monly used for the simulation of interacting, autonomous agents. Using
agent based modelling it is possible to assess the eects of interactions be-
tween individual agents and populations of agents on the whole system.
There is a limited availability of associated documentation and quantiable
research data in many areas of malaria spread research. To address the
problem, three models have been built that investigate dierent aspects of
malaria transmission. The models are developed with
exibility and adapt-ability as important factors in their use, so that they can provide veriable
results with potentially limited availability of data.
The three models produced are as follows.
Malaria in Displaced Populations.
Malaria in Peri-Urban Settlements.
Malaria and Human Immunodeciency Virus (HIV) Dual Infection.
The rst model simulates malaria spread amongst a migratory population of
agents. The second model simulates malaria spread amongst a settled pop-
ulation of agents living in peri-urban conditions with an associated mapping
model used to create custom environments based on real world settlements.
The third model visualises the spread of both malaria and HIV through
a population of agents that form partnerships to simulate sexual activity.
The dual infection of malaria and HIV eects the rate of infection for HIV
and the parasite burden of malaria in the hosts.
A further problem to be addressed is that agent based modelling for malaria
has lacked a combination of sophistication and
exibility in past work. Mod-elling that inputs a strictly rigid set of variables, in the case of a complex
system like disease spread, does not lend itself well to the simulation of sce-
narios where some data is unavailable. The models documented here were
built with
exibility as an important factor in their design.
The results show that the models produce output that is consistent with
real world malaria statistics. They demonstrate both qualitative validation,
in the demonstration of realistic trends and also quantitative validation, in
output of percentages of infected individuals and other disease related phe-
nomena. There is some evidence of emergence in the Malaria in Displaced
Populations model which is of particular interest. The conclusion is that
agent based modelling shows promise as a method of evaluating the eects
of malaria. The models and the research that accompanies them are im-
portant steps toward an understanding of what agent based modelling may
contribute to malariology.