Optimising Antibiotic Treatments with Multi-objective Population-based Algorithms

Citation

Goranova M, Contreras-Cruz MA, Hoyle A & Ochoa G (2020) Optimising Antibiotic Treatments with Multi-objective Population-based Algorithms. In: CEC 2020: Congress on Evolutionary Computation. 2020 IEEE Congress on Evolutionary Computation (CEC), Glasgow, United Kingdom, 19.07.2020-24.07.2020. Piscataway, NJ, USA: IEEE. https://doi.org/10.1109/cec48606.2020.9185489

Abstract
Antibiotic resistance is one of the major challenges that we are facing today. The frequent overuse of antibiotics is one of the main reasons for the development of resistance. A mathematical model of bacterial population dynamics is used, where drug administration and absorption mechanics are implemented to evaluate the fitness of automatically designed treatments. To maximise the probability of curing the host while minimising the total drug used we have explored treatments with different daily dosages and lengths. Two multi-objective population-based methods, a well-known evolutionary algorithm and a particle swarm optimisation algorithm are tuned and contrasted when solving the posed treatment design problem. The best solutions found by our approach suggest treatments ranging from five to seven days with a high initial dose, followed by lower doses, use lower amounts of the drug than the standard common practice of fixed daily dosages over ten days.

Keywords
Antibiotics; Treatment Scheduling and Design; Noisy Multi-Objective optimisation; Stochastic Mathematical Modelling; Pharmacokinetics/Pharmacodynamics Modelling; Evolutionary Algorithms; Particle Swarm optimisation

People

Miss Mila Goranova

PhD Researcher, Computing Science and Mathematics - Division

Dr Andrew Hoyle

Senior Lecturer, Mathematics

Professor Gabriela Ochoa

Professor, Computing Science