Cancer is a large and growing health problem in the developed world and chemotherapy is one of the main modes of treatment.
Improved cancer chemotherapy will reduce mortality rates and prolong survival times even when the disease is incurable.
This work uses Evolutionary Algorithms to explore the range of possible treatments available to clinicians "in silico" rather
than in vivo. Mathematical models of tumour response to chemotherapy are used to evaluate the therapies generated by the EA. The
EA is able to evolve optimised treatments that, in simulation, outperform chemotherapies established by clinical use. This strongly
suggests that improvements to chemotherapy can be gained by moving beyond conventional treatment approaches.
Over twelve years, a series of projects has applied a range of EA approaches to cancer chemotherapy optimisation, principally
Genetic Algorithms, Particle Swarm Optimisation and Estimation of Distribution Algorithms. These methods have proved superior to
traditional optimisation techniques in optimising treatments. In particular, they can handle the inclusion of drug toxicity
constraints without penalty in performance. During this process, much knowledge has been gained about the design of these
algorithms leading to efficient and effective computation.