A No-Nonsense Control Engineering Approach to Anaesthesia Control during Induction Phase

This paper introduces a strategy for patient-individualized control of induction phase during general anaesthesia. A down-to-earth, pragmatic (and possibly provocative) approach has been used. The core of the paper is the idea to avoid the direct use of the Bispectral Index signal (BIS) as a feedback signal, since it introduces artificial delays, it is noisy and in the initial stage of the induction it contains no useful information for control purposes because it hardly varies and it introduces nonlinearity. The effect site concentration (Ce) of the drug in the patient is instead used as the controller's feedback signal. Of course the BIS measurement is still indirectly used in our proposed control strategy, in order to adapt the parameters of the Hill curve - which relates BIS to Ce - resulting in a patient-individualized closed loop control of anesthesia. These ideas are validated on 12 virtually generated patients varying significantly in sensitivity to the drug effect, using a simple PID control. The control of the induction phase is split into two phases: the initial phase where no useful BIS feedback signal is available, hence target controlled infusion (TCI) with a fixed setpoint is temporarily applied; and the secondary phase when a useful BIS feedback signal becomes available and the TCI setpoint is adapted based on the estimated Hill-curve steepness. The approach is simple and excellent simulation results combined with theoretical insights indicate that the strategy looks promising for future clinical practice.