Robust adaptive control of hypnosis during anesthesia

A closed-loop controller for hypnosis was designed and validated on humans at our laboratory. The controller aims at regulating the Bispectral Index (BIS) - a surrogate measure of hypnosis derived from the electroencephalogram of the patient - with the volatile anesthetic isoflurane administered with a closed-circuit breathing system. The control algorithm consists of a cascaded Internal Model Controller (IMC) where the master loop aims at regulating BIS. The slave loop tracks endtidal concentration references provided by the master controller. In this paper, a new tuning method is presented. First, a robust design procedure which guarantees stability of the slave controller despite parametric uncertainties is described. Then, we will demonstrate how the estimation of the drug's equilibration constant k(e0) greatly improves performance if the estimated value is used to update the models in the control scheme. In order to do so, an identification scheme for k(e0) is proposed, which requires estimation of the drug's time to peak effect t(peak). The identification algorithm requires few modeling assumptions and guarantees convergence. Simulation results are presented, which quantify both the performance of the identification scheme and the improvement of the closed-loop control performance.