Models of Acquired Immunity to Malaria: A Review

■95

and virulence) relating to malaria transmission and interactive variability in parasite-host

dynamics can easily be modelled. Another strength of this modeling approach is that it en-

ables the simulation of the interaction of individuals and vectors in the domain of interest,

which can in itself be heterogeneous, and stores information about each agent (mosquito

and/or human). Protective efficacy of immunity varies between individual humans [42], so

it may be misleading to ignore individual variations in biologically important parameters

while this can be efficiently addressed by ABM simulations. Decay rates of Immunity, for

instance, varies in humans and should be sampled from a suitable distribution with an as-

sumed mean. The flexibility of agent-based approaches in modifying model attributes to

reflect local individual features and geographical factors, allows the construction of models

that can handle realistic questions relating to malaria control in specific local contexts. A

considerable disadvantage is the higher computational burden, especially with increasing

population size.

Most individual-based models focus more on increasing the immune status of an in-

dividual as they age and get more exposed, without considering how parasite densities

changes. However, the models which meet the kind of realistic expectation of the dynam-

ics and complex structure of NAI to malaria are those suggesting that immunity indicators

should be functions of parasite density and diagnostic detection sensitivity rather than bi-

nary factors. Most of these models are those included in the OpenMalaria ensemble [159],

[194] which are the advanced, quite realistic individual-based models we rely on today. The

programming of the models was done in C++ as part of the open source software platform

(http://code.google.com/p/openmalaria/), hence the name, OpenMalaria.

A compelling strength of this model is that it couples an ensemble of 14 associated sub-

models that have been validated with real field data from various African settings. The

integrated mathematical models are used for predicting the epidemiologic and economic

effects of malaria vaccines both at the individual and population level. The model simu-

lations use a function that reduces asexual parasite densities to model important concepts

such as: the transmission to mosquito [160], [114], the degree of severity of episodes based

on the parasite load [115], [134] and how the distribution of parasite densities is modified

in the semi-immune host [76]. Additionally, the model considers the effects of factors

such as heterogeneity in transmission [198], body surface area [205], [114], [160], ma-

ternal antibodies [76], stage-specificity of immunity and superinfection [115], [76], [205].

However the aspect of the decay of NAI in the absence of exposure is yet to be properly

addressed. This is because there is limited quantitative data from which to directly estimate

this rate of decay of immune control of parasite densities. Overall, there still remain other

unclear mechanisms of malaria immunity which have not yet been properly explained (by

the aforementioned model), mainly due to limited field data, but they remain a reliable ba-

sis for further research on this topic. Closing these gaps so as to facilitate the development

of malaria vaccines remains the main target of tropical diseases research.

Most of the models discussed in this review were approximately fitted to available

epidemiological data, where the infection incidence determines the status of either stable

endemicity and widespread immunity or an unpredictable pattern of sporadic epidemics in

a population lacking sufficient regular exposure to maintain immunity. Moreso, while some