Contributions to the initialization of online identification algorithms for anaesthesia: The NeuroMuscular Blockade case study

The initialization of identification algorithms in anaesthesia is addressed in this paper. The two parameters of a minimally parameterized Wiener model for the NeuroMuscular Blockade (NMB) are identified online. A linear correlation was found between the parameter in the linear part of the model and the time the NMB response takes to drop from its initial value of 100% to 50% after the initial bolus drug administration. When applied to data collected from 43 real patients undergoing general surgery, the adjusted squared correlation coefficient was 0.744. The parameter in the nonlinear block was estimated by inversion of the Hill equation at the time instant where the recovery after the initial bolus is assumed to start. This point of recovery is case-dependent and it is obtained through an online adaptive algorithm: OnLine tuned Algorithm for Recovery Detection. A quadratic relationship was found between the true value of this parameter and the identified one (with an adjusted squared correlation coefficient of 0.754). These methods provide good estimations for the model parameters that describe each individual patient response to the administration of drugs. The results strongly support the incorporation of these real signal-based strategies in online identification algorithms for the control of drug delivery systems in anaesthesia.