We are using individual-based stochastic simulation models of the epidemiology of Plasmodium falciparum to make predictions of the epidemiological effects and cost-effectiveness of a comprehensive set of malaria control strategies, including integrated control. We have started with a focus on predicting the epidemiological effects of potential malaria vaccines (asexual blood stage, transmission-blocking, and pre-erythrocytic) delivered via the Expanded Program on Immunization, and have also modeled intermittent preventive treatment. Models for vector control can be integrated into the same simulation platform and hence will give directly comparable quantitative predictions of effects and cost-effectiveness. We thus plan to simulate a comprehensive set of malaria transmission settings, with realistic assumptions about different health systems.
We currently use underlying within-host models based on descriptions of the course of parasitaemia in malariatherapy patients. For parameter optimisation we currently use a set of 61 field scenarios from sub-Saharan Africa, comprising data on seasonality, age- patterns of infection, parasite density, clinical episodes, severe malaria and mortality, and fit models to these data using a genetic algorithm. The software is implemented using the Berkeley Open Interface for Network Computing (BOINC) which enables volunteer members of the public to run the simulations, allowing parallel processing of many different computer intensive tasks. The robustness of the conclusions is evaluated using probabilistic sensitivity analysis, and we also plan comparisons of ensembles of different models.
The models will be useful to all those who need to understand the implications for malaria burden of changes in interventions and health systems.
