Rob J. Roy

Multiple-model adaptive predictive control of mean arterial pressure and cardiac output

A multiple-model adaptive predictive controller has been designed to simultaneously regulate mean arterial pressure and cardiac output in congestive heart failure subjects by adjusting the infusion rates of nitroprusside and dopamine. The algorithm is based on the multiple-model adaptive controller and utilizes model predictive controllers to provide reliable control in each model subspace. A total of 36 linear small-signal models were needed to span the entire space of anticipated responses. To reduce computation time, only the six models with the highest probabilities were used in the control calculations. The controller was evaluated on laboratory animals that were either surgically or pharmacologically altered to exhibit symptoms of congestive heart failure. During trials, the controller performance was robust with respect to excessive switching between models and nonconvergence to a single dominant model. A comparison with a previous multiple-drug controller design is made.<<ETX>>

Multiple Model Adaptive Control Procedure for Blood Pressure Control

Multiple model adaptive control procedures have been considered for a computer-based feedback system which regulates the infusion rate of a drug (nitroprusside) in order to maintain desired blood pressure. Because the transfer function parameters are different for each patient, and furthermore are time variant, such an algorithm is desirable for maintaining both steady-state and transient specifications. To this effect, computer simulation has shown that multiple model adaptive control procedures might be successfully applied to the control of blood pressure despite the uncertainty in the delays, time constant, and gains. Additional efforts concerned with the actual demonstration of these concepts on dogs have further supported the role of adaptive control for blood pressure regulation.

Derived fuzzy knowledge model for estimating the depth of anesthesia

Reliable and noninvasive monitoring of the depth of anesthesia (DOA) is highly desirable. Based on adaptive network-based fuzzy inference system (ANFIS) modeling, a derived fuzzy knowledge model is proposed for quantitatively estimating the DOA and validate it by 30 experiments using 15 dogs undergoing anesthesia with three different anesthetic regimens (propofol, isoflurane, and halothane). By eliciting fuzzy if-then rules, the model provides a way to address the DOA estimation problem by using electroencephalogram-derived parameters. The parameters Include two new measures (complexity and regularity) extracted by nonlinear quantitative analyses, as well as spectral entropy. The model demonstrates good performance in discriminating awake and asleep states for three common anesthetic regimens (accuracy 90.3% for propofol, 92.7%, for isoflurane, and 89.1% for halothane), real-time feasibility, and generalization ability (accuracy 85.9% across the three regimens). The proposed fuzzy knowledge model is a promising candidate as an effective tool for continuous assessment of the DOA.

Brief paper: Model reference adaptive control of drug infusion rate

The role of model reference adaptive control is under consideration for a computer based feedback system which will regulate the infusion rate of a drug (nitroprusside) in order to maintain desired blood pressure. Because the transfer function parameters are different for each patient and furthermore are variable with time, an adaptive algorithm is desirable for maintaining both steady-state and transient specifications. To this effect, computer simulation has shown that direct (or implicit) model reference procedures might be successfully applied to the control of blood pressure despite the uncertainty in the delays, time constant and gains. Additional efforts concerned with the actual demonstration of these concepts on dogs have further supported the role of adaptive control for blood pressure regulation.

Depth of anesthesia estimation and control [using auditory evoked potentials]

A fully automated system was developed for the depth of anesthesia estimation and control with the intravenous anesthetic, Propofol. The system determines the anesthesia depth by assessing the characteristics of the mid-latency auditory evoked potentials (MLAEP). The discrete time wavelet transformation was used for compacting the MLAEP which localizes the time and the frequency of the waveform. Feature reduction utilizing step discriminant analysis selected those wavelet coefficients which best distinguish the waveforms of those responders from the nonresponders. A total of four features chosen by such analysis coupled with the Propofol effect-site concentration were used to train a four-layer artificial neural network for classifying between the responders and the nonresponders. The Propofol is delivered by a mechanical syringe infusion pump controlled by Stanpump which also estimates the Propofol effect-site and plasma concentrations using a three-compartment pharmacokinetic model with the Tackley parameter set. In the animal experiments on dogs, the system achieved a 89.2% accuracy rate for classifying anesthesia depth. This result was further improved when running in real-time with a confidence level estimator which evaluates the reliability of each neural network output. The anesthesia level is adjusted by scheduled incrementation and a fuzzy-logic based controller which assesses the mean arterial pressure and/or the heart rate for decrementation as necessary. Various safety mechanisms are implemented to safeguard the patient from erratic controller actions caused by external disturbances. This system completed with a friendly interface has shown satisfactory performance in estimating and controlling the depth of anesthesia.

The use of fuzzy integrals and bispectral analysis of the electroencephalogram to predict movement under anesthesia

The objective of this study was to design and evaluate a methodology for estimating the depth of anesthesia in a canine model that integrates electroencephalogram (EEG)-derived autoregressive (AR) parameters, hemodynamic parameters, and the alveolar anesthetic concentration. Using a parametric approach, two separate AR models of order ten were derived for the EEG, one from the third-order cumulant sequence and the other from the autocorrelation lags of the EEG. Since the anesthetic dose versus depth of anesthesia curve is highly nonlinear, a neural network (NN) was chosen as the basic estimator and a multiple NN approach was conceived which took hemodynamic parameters, EEG derived parameters, and anesthetic concentration as input feature vectors. Since the estimation of the depth of anesthesia involves cognitive as well as statistical uncertainties, a fuzzy integral was used to integrate the individual estimates of the various networks and to arrive at the final estimate of the depth of anesthesia. Data from 11 experiments were used to train the NNs which were then tested on nine other experiments. The fuzzy integral of the individual NN estimates (when tested on 43 feature vectors from seven of the nine test experiments) classified 40 (93%) of them correctly, offering a substantial improvement over the individual NN estimates.

Improvement in Arteral Oxygen Control Using Multiple-Model Adaptive Control Procedures

A computer-based proportional-integral (PI) controller has been developed to control arterial oxygen levels in mechanically ventilated animals. Arterial oxygen saturation is monitored using a noninvasive oximeter and control is effected by adjusting the inspired oxygen fraction. The performance of the feedback system is sensitive to the open-loop gain so that the desired transient specifications can be achieved only by empirical adjustments of the PI controller. Because the open-loop gain includes the animals response, it may vary with time and with the administration of positive end-expiratory pressure. Multiple-model adaptive control procedures were therefore used to desensitize the system to these variable gains. Computer simulations demonstrated the effectiveness of the algorithm over a wide variation of plant parameters. A comparison to a fixed, well-tuned proportional-integral controller showed an improvement in the regulatory response to a step disturbance. Animal experiments confirmed the feasibility of using multiple-model adaptive control to regulate arterial oxygen saturation.

A feedback controller for ventilatory therapy

A computerized system that uses feedback of end-tidal CO2 fraction (FETCO2) to adjust minute volume of a ventilator has been developed and tested. The effectiveness and robustness of the controller were evaluated in five anesthetized dogs. The controlled responded to step-changes in the set-point for FETCO2 by adjusting minute volume so that the FETCO2 settled to the new set-point in less than 60 sec with less than 20% overshoot. The system exhibited suitable dynamic response to step-changes in set-point with loop gains as large as two times and as small as one-half the optimal value. The breath-to-breath variation in FETCO2 values during prolonged periods of closed-loop controlled ventilation was smaller than the variation during periods of constant minute volume ventilation in three of five experiments. The controller generally maintained FETCO2 within ±0.1 vol% of the set-point. A disturbance to the controlled system was produced by releasing an occlusion of a branch of the pulmonary artery. The controller always responded to this disturbance in a stable manner, returning the FETCO2 to its desired value within 30 sec. Accurate control of arterial partial pressure of CO2(PaCO2) will require modifications enabling the system to determine the relationship between FETCO2 and PaCO2.

Anesthesia control using midlatency auditory evoked potentials

This paper shows the development of a system to control inhalation anaesthetic concentration delivered to a patient based upon that patients midlatency auditory evoked potentials (MLAEPs). It was developed and tested in dogs by determining response to the supramaximal stimulus of tail clamping. Prior to tail clamp, the MLAEP was recorded along with inhalational anaesthetic concentration and classified as responders or nonresponders as determined by tail clamping. This was performed at a number of different anaesthetic levels to obtain a data training set. The MLAEPs were compacted by means of discrete time wavelet transform (DTWT), and together with anaesthetic concentration value, a stepwise discriminant analysis (SDA) was performed to determine those features which could separate responders from nonresponders. It was determined that only 3 features were necessary for this recognition. These features were then used to train a 4-layer artificial neural network (ANN) to separate the responders from nonresponders. The network was tested using a separate set of data, resulting in a 93% recognition rate in the anaesthetic transition zone between responders and nonresponders, and 100% recognition rate outside this zone. The anaesthetic controller used this ANN combined with fuzzy logic and rule-based control. A set of 10 animal experiments were performed to test the robustness of this controller. Acceptable clinical performance was obtained, showing the feasibility of this approach.

Multiple-drug hemodynamic control using fuzzy decision theory

A fuzzy-logic-based, automated drug-delivery system has been developed and validated on a nonlinear canine circulatory model for managing hemodynamic states. This controller features: (1) a fuzzy decision analysis module for patient status determination by assessing cardiac index, systemic vascular resistance index, and pulmonary vascular resistance index and (2) a fuzzy hemodynamic management module utilizing dopamine, phenylephrine, nitroprusside, and nitroglycerin for regulating mean arterial pressure, mean pulmonary arterial pressure, and cardiac output. A rule-based drug delivery scheduling program has been devised and incorporated to execute the therapeutic strategy as recommended by the decision-analysis module. Compared to the existing controllers, this system is able to achieve a faster response time with a more secured and effective regulation. The simulation results have demonstrated the feasibility of the decision analysis process for automated management of the arterial and venous circulation with an expanded arsenal of pharmacological agents.