Delivery propofol fault detection in the automation process of general anesthesia

Given the danger associated with the delivery of anesthesia during general surgical procedures, research in this field has been extensive. The design of controllers to automate the process aims to assist the anesthesiologist in precisely administering the drug dose. Based on the mathematical model of the patient, which is composed of pharmacokinetics and pharmacodynamics models, the automation challenge lies in administering the appropriate drug dose during general surgery, ensuring that the depth of hypnosis remains at a desired value. On the other hand, failures in actuators and monitors are present in the automation process. This article presents an algorithm for detecting a failure in drug infusion pumps. The algorithm for fault detection presented is based on the so-called interval observers and the positivity property of the mathematical dynamic model. The presented model predictive control (MPC) algorithm guarantees the system’s positivity. The proposed algorithm is proven in five in-silico patients where a failure in the infusion pump is simulated and detected.