An optimal control problem for in vitro virus competition

Authors:

Abstract
Competition between organisms that occupy the same physical space is an important phenomenon that has been studied extensively through mathematical models, primarily in ecology. Using a relatively benign pathogen with a competitive advantage to contain a more harmful one is an important and relatively new concept in infectious disease management. But many questions remain about effective strategies for the implementation of this concept. This paper presents a method to formulate instances of these questions in the language of dynamical systems theory, amenable to methods from optimal control theory. Specifically, the method begins with a stochastic simulation of in vitro competition of two different virus strains that reproduces competition dynamics observed in experiments. After transforming this simulation into a deterministic simulation, the method proceeds by interpolating a state space model of this dynamical system from time series of system states. That is, a closed form model is constructed from "experiments" performed on the simulation. The interpolation is done using methods from computational algebraic geometry, resulting in a (typically nonlinear) polynomial dynamical system over a finite field. Based on available biological methods, a cost function is defined and a control component is introduced, which provides input to enhance the competitive fitness of one of the two strains. The resulting system can be expressed as a discrete event dynamical system. One can now employ methods developed for optimal control of polynomial dynamical systems over a finite field to solve the control problem so defined. An example of a controller is given, that has been implemented in the laboratory and successfully verified.