Optimization-based design of closed-loop control of anesthesia

Abstract Optimization-based techniques can be usefully employed in the design of control systems in general anesthesia. In particular, in this chapter, the control of the depth-of-hypnosis by using the bispectral index scale signal as process variable is considered. The general approach consists in first optimizing the controller parameters by using an evolutionary algorithm to minimize the worst-case performance index over a reduced set of models of patients. To this end, the integrated absolute error is chosen as optimization functional, and the set of models of patients is rich enough to be representative of a wide population. Then, the robustness to intra- and interpatient variability is verified by testing the devised control system on a wide set of patients, determined, for example, by using a Monte Carlo method. This methodology can be applied to different control schemes and allows the fulfillment of the clinical specifications.