Kristian Soltesz

Robust closed-loop control of induction and maintenance of propofol anesthesia in children

During closed‐loop control, a drug infusion is continually adjusted according to a measure of clinical effect (e.g., an electroencephalographic depth of hypnosis (DoH) index). Inconsistency in population‐derived pediatric pharmacokinetic/pharmacodynamic models and the large interpatient variability observed in children suggest a role for closed‐loop control in optimizing the administration of intravenous anesthesia.

Design and Clinical Evaluation of Robust PID Control of Propofol Anesthesia in Children

This paper describes the design of a robust proportional-integral derivative (PID) controller for propofol infusion in children and presents the results of clinical evaluation of this closed-loop system during endoscopic investigations in children age 6-17. The controller design is based on a set of models that describes the interpatient variability in the response to propofol infusion in the study population. The PID controller is tuned to achieve sufficient robustness margins for the identified uncertainty. 108 children were enrolled in this study, and anesthesia was closed-loop controlled in 102 of these cases. Clinical evaluation of the system shows that closed-loop control of both induction and maintenance of anesthesia in children based on the WAVCNS index as a measure of clinical effect is feasible. A robustly tuned PID controller can accommodate the interpatient variability in children and spontaneous breathing can be maintained in most subjects.

Quantification of the Variability in Response to Propofol Administration in Children

Closed-loop control of anesthesia is expected to decrease drug dosage and wake up time while increasing patient safety and decreasing the work load of the anesthesiologist. The potential of closed-loop control in anesthesia has been demonstrated in several clinical studies. One of the challenges in the development of a closed-loop system that can be widely accepted by clinicians and regulatory authorities is the effect of interpatient variability in drug sensitivity. This system uncertainty may lead to unacceptable performance, or even instability of the closed-loop system for some individuals. The development of reliable models of the effect of anesthetic drugs and characterization of the uncertainty is, therefore, an important step in the development of a closed-loop system. Model identification from clinical data is challenging due to limited excitation and the lack of validation data. In this paper, approximate models are validated for controller design by evaluating the predictive accuracy of the closed-loop behavior. A set of 47 validated models that describe the interpatient variability in the response to propofol in children is presented. This model set can be used for robust linear controller design provided that the experimental conditions are similar to the conditions during data collection.

Nonlinear lateral control strategy for nonholonomic vehicles

This paper proposes an intuitive nonlinear lateral control strategy for trajectory tracking in autonomous nonholonomic vehicles. The controller has been implemented and verified in Alice, Team Caltechs contribution to the 2007 DARPA Urban Challenge competition for autonomous motorcars. A kinematic model is derived. The control law is described and analyzed. Results from simulations and field tests are given and evaluated. Finally, the key features of the proposed controller are reviewed, followed by a discussion of some limitations of the proposed strategy.

Closed-Loop Anesthesia in Children using a PID Controller: A Pilot Study

8 Measuring adeQuacy of analgesia with cardiorespiratory coherence Chris Brouse, Walter Karlen, Guy Dumont, Dorothy Myers, Erin Cooke, Jonathan Stinson, Joanne Lim, J. Mark Ansermino The University of British Columbia, Vancouver, Canada Introduction: An automated nociception monitor would be very useful in general anesthesia, providing anesthesiologists with real-time feedback about the adequacy of analgesia. We have developed an algorithm to measure nociception using respiratory sinus arrhythmia (RSA) in heart rate variability (HRV). We have previously shown that this algorithm can detect patient movement (strongly nociceptive events) during general anesthesia 1. We will now attempt to determine if the algorithm responds to boluses of anesthetic drugs (strongly anti-nociceptive events). Method: Algorithm: The algorithm estimates cardiorespiratory coherence, which is the strength of linear coupling between HR and respiration (one measure of RSA). It measures and combines the spectral power in both signals using wavelet analysis. Coherence is dimensionless, and ranges from 0 (no coherence, strong nociception) to 1 (perfect coherence, no nociception). Data Analysis: Following ethics approval and informed consent, 60 drug bolus events (excluding induction of anesthesia) were recorded in 47 pediatric patients receiving general anesthesia during dental surgery. In post hoc analysis, coherence was averaged over the 60s immediately preceding the bolus dose of drug (nociceptive period). The bolus was given 30s to take effect, after which the coherence was averaged over the following 60s (anti-nociceptive period). The change in average coherence between the two periods was calculated. The change in average HR was also calculated, for comparison. Results: Coherence increased by an average of 0.14 (32%) in response to the bolus dose of anesthetic drug. HR decreased by an average of 4.1 beats/min (3.9%). Discussion: Cardiorespiratory coherence responded much more strongly to the anesthetic boluses than did HR alone. This result, combined with previous work showing that coherence is low during periods of nociception [1], demonstrates that cardiorespiratory coherence can be used to measure the adequacy of analgesia during general anesthesia. We are currently adapting the algorithm so that it can be used in real-time.

Individualized PID control of depth of anesthesia based on patient model identification during the induction phase of anesthesia

This paper proposes a closed-loop propofol admission strategy for depth of hypnosis control in anesthesia. A population-based, robustly tuned controller brings the patient to a desired level of hypnosis. The novelty lies in individualizing the controller once a stable level of hypnosis is reached. This is based on the identified patient parameters and enhances suppression of output disturbances, representing surgical stimuli. The system was evaluated in simulation on models of 44 patients obtained from clinical trials. A large amount of improvement (20 – 30%) in load suppression performance is obtained by the proposed individualized control.

Transfer function parameter identification by modified relay feedback

This paper proposes a transfer function parameter identification method, applicable to SISO systems of any order. Parameter identification is posed as a (non-convex) squared output error minimization problem, numerically solved utilizing Newton-Raphson iteration with back tracking line search. Focus lies on computing the cost function gradient and Hessian with respect to the parameter vector and on finding a feasible start point. The method is demonstrated for FOTD model identification. A modified relay non-linearity is utilized in order to obtain most input signal energy at a predefined phase, without a priori system information. The identification method is evaluated on a batch of common process industry processes. Finally, conclusions and suggestions on future work are provided.

Assessing control performance in closed-loop anesthesia

Recently, several control systems for closed-loop anesthesia have been demonstrated both in simulation and clinical studies. A set of performance measures, proposed by Varvel et al., have constituted the standard means of comparing such systems. This paper debates the adequacy of the Varvel measures, as applied to closed-loop anesthesia, and proposes an alternative set of measures. Key features of the proposed measures are: wide acceptance within the control community; reflection of clinical feasibility; separate measures for induction and maintenance of anesthesia; separation of outlier detection and performance evaluation. The proposed measures are descriptive, few, and easy to compute.

Extending the Relay Feedback Experiment

An augmented version of the traditional relay feedback experiment is proposed. It aims at producing an input with energy concentrated to a frequency band, corresponding to a certain phase sector of the Nyquist curve of the process to be identified. A non-convex problem is formulated. Sub-optimal, but efficient, algorithms are developed.

Initialization of the Kalman filter without assumptions on the initial state

In absence of covariance data, Kalman filters are usually initialized by guessing the initial state. Making the variance of the initial state estimate large makes sure that the estimate converges quickly and that the influence of the initial guess soon will be negligible. If, however, only very few measurements are available during the estimation process and an estimate is wanted as soon as possible, this might not be enough. This paper presents a method to initialize the Kalman filter without any knowledge about the distribution of the initial state and without making any guesses.