Controlling the evolution of antibiotic resistance

Modelling antibiotic resistance evolution is inherently a multiscale problem: from the physical interactions between drug molecules and their targets to the epidemiology of drug resistance in clinical settings. Although predicting the evolution of resistance is a difficult and ongoing problem, it is known that pathogens are continually adapting to our drug prescription patterns. For this reason, as well as the continual downturn in the discovery of new drugs, the question of how to best deploy antibiotics has never been more pressing. The purpose of this thesis is to use tools from control and systems theory to ask the following fundamental question: How can we design rational antibiotic deployment strategies that do not promote the evolution of antimicrobial resistance? By re-examining epidemiological models from the literature, in the first part of this thesis we show that the optimal drug deployment protocol has a universal structure not determined by biological detail. This class of epidemiological models, however, provide insight into the underlying mechanisms that influence the spread of disease at the population level but fail to capture the complex molecular interactions between different antibiotics and bacteria, as well as to provide an experimental system to test the efficacy of different treatment protocols. Therefore in the second part of the thesis we pose an evolutionary model of an experimental microbial system that allows us to study drug interactions and the effect that combination treatments have on the evolution of multidrug resistance. Again, using optimal control theory we design drug deployment protocols that minimise conditions promoting the evolution of antimicrobial resistance in a single host. Finally, in the last part of the thesis we propose an epidemiological model where patients are considered as individual agents receiving antimicrobial treatment in a clinical setting. This stochastic and spatially explicit model allows us the possibility to evaluate the efficacy of different drug usage strategies. We conclude with a general principle: the best performing drug usage policies utilise the highest quality of available information.EThOS - Electronic Theses Online ServiceCONACYT and SEPGBUnited Kingdo