Optimal drug infusion profiles using a Particle Swarm Optimization algorithm

Abstract

The dynamic optimization of the administration of therapeutic drugs in simulated patients is proposed. The approach is based on a non-linear discontinuous cardiorespiratory model, which has been conceived to simulate the effect of inotropic and vasoactive drugs as well as anesthetic agents. A stochastic technique (Particle Swarm Optimization), within the context of the control vector parameterization approach, is adopted to identify the infusion profiles of various drugs in order to track, as close as possible, the set-points on several variables of medical interest. Two different medical procedures are investigated in order to test the efficiency and robustness of the algorithm: a congestive heart failure and the unclamping of an aortic vessel. Due to the conflicting nature of the different objectives, compromise solutions are obtained in all cases.