Depth of anaesthesia control techniques and human body models

Depth of anaesthesia control techniques and human body models

The objective of this project is to develop patient dose-response models and to provide an adequate drug administration regimen for anaesthesia to avoid under- or over-dosing of patients. The controllers are designed to compensate for patients’ inherent drug response variability, to achieve the best output disturbance rejection, and to maintain optimal set point response.

To address this issue, this project uses four independent methods to investigate the control strategies for the regulation of hypnosis. Two medications are used in a thorough evaluation and comparison of controller performance.

A robust internal model controller (RIMC) based on the Bispectral Index (BIS) is investigated firstly. The controller compares the measured BIS with its input reference to provide the expected Propofol concentration, and then the controller manipulates the anaesthetic Propofol concentration entering the anaesthetic system to achieve the desired BIS value. This study also develops patient dose-response models. The performance of the RIMC is comprehensively compared with that of proportional-integral-derivative (PID) controller for the robustness, set-point changes, disturbances and noise in the measured BIS. Numerical simulations illustrate that the RIMC performed better than the PID controller. The robust performance of the two controllers is evaluated for a wide range of patient models by varying in patient parameters.

The impact of the time-delays of patient and instrumentation on a closed-loop depth of anaesthesia control system was investigated. In this study, the Smith predictive technique is used to identify and compensate for the time-delay problem and improve the overall response of the depth of anaesthesia. The proposed method is validated using measured BIS signals in simulation. The results showed that the proposed procedure improves the performance of the closed-loop system for reference tracking and overall stability. The proposed method also has approximately 15% less overshoot, a two minute shorter settling time, and is more robust to disturbance rejection.

The problem of non-linearity is identified in the depth of anaesthesia model and a deadbeat controller is designed in response to this problem. The proposed system is evaluated in simulation using Matlab and Simulink, and results are compared with a traditional PID control system and with an internal model control (IMC) controller. The results show that the proposed scheme has less over- and under-shoot, shorter settling time and is more robust to depth of anaesthesia disturbances. In addition, the proposed method is easy to implement.

The Model Predictive Control (MPC) technique is also investigated in depth of Anaesthesia (DoA) control. The proposed robust control system with a predictive controller is evaluated in simulation. The result is compared with two control systems. First compared with a conventional PID controller and then with a control system with an Internal Model Controller. The results show that the proposed scheme has a smaller overshoot (by about 10%) and a shorter settling time (by about 2 minutes shorter) and is more robust to disturbances caused by parameter changes.