A combined estimation and multi-parametric model predictive control approach for intravenous anaesthesia

This paper describes a strategy for the control of intravenous depth of anaesthesia (DOA). Based on a mathematical model of the system, global sensitivity analysis is first presented to determine the relative influence of the uncertain pharmacokinetic and pharmacodynamic parameters and variables. Then estimation techniques are applied for the key parameters that cannot be measured directly, combined with a multi-parametric/explicit model predictive control strategy. The two estimation strategies: a Kalman filter and multi-parametric moving horizon estimation are employed and tested on a set of twelve patients.