Helminths are macroparasites that cause many diseases, in people and animals, with a mode of transmission very different from that of most bacteria and viruses. The severity of the infection in a host is affected by the parasite burden of that specic host. The parasite burdens are extremely variable between individual hosts as a result of the transmission methods. The diseases caused by parasites are often diffcult to treat eectively, and even more dicult to eradicate, largely because of the complex transmission methods. As such, models could be very helpful in determining best practices in the effcient treatment of these diseases.

The research presented here is an investigation into the spread of soil-transmitted helminths, through the use of metapopulation models, and the effects of simple control measures on the infections caused by them. Models often fall into two categories: fully stochastic models, which are computationally costly; or mean-eld models, which cannot capture the detail of this variability. The methodology of this research is to formulate models that incorporate the variance in parasite burdens between hosts to study whether the variance would cause a signicant dierence in the recommended control, based on the optimisation and study of simple controls on the models.

The initial examination is intended to see how the inclusion of additional details specic to populations in multiple environments, such as the inclusion of the variance in parasite burdens between hosts, could be incorporated into the models and the effect this would have on the controls.

In this investigation, we found that simply including the variance in the models did not signicantly improve the predictive power of the model in regards to control. Without additional knowledge of the underlying distribution of the parasite burdens, which the mean and variance serve to model, treatments could only be applied to the whole host population. This means that the effect of more individualised treatments, only on hosts that really need it, cannot be studied and the increased information of the distribution that the variance provides is not utilised. For these treatments on the entirety of the host population, we found that, comparing models which included parasite burden variability and those which did not, did not result in qualitatively different optimal controls. This was the case even when the control was optimised with the intent of minimising the variance as a priority. However, we also took steps to consider how treatments could be applied to only a portion of the overall host population being modelled. This was only possible when the mean and variance of the model were a good t for the underlying distribution of parasite burdens, which was known as a result of empirical means. The hope is that this will provide a potential avenue for further study in this area.

On the other side of the investigation, we studied how the inclusion of parasite resistance to treatment may be modelled and the effects on the controls. In investigating this we found that the total parasite populations could posses properties that the standard model analysis would not show. The most important of these was the potential that the basic model would be unreactive, which would imply that transient growth around an equilibrium was unlikely, yet when the sum of variables was considered transient growth would occur. These properties could then have signicant effects on the control. This is due to the effectiveness of the control being measured against the change to the total population, rather than the individual variables.

It is hoped that this research may further the understanding of how variance may be incorporated
into metapopulation models and the potential diffculties, and how parasite resistance to treatment
Date of Award14 Oct 2020
Original languageEnglish
Awarding Institution
  • University of Bath
SupervisorChristopher Guiver (Supervisor) & Jane White (Supervisor)

Cite this