Closed-loop feedback control of methohexital anesthesia by quantitative EEG analysis in humans

Closed-loop anesthesia: foundations and applications in contemporary perioperative medicine

A closed-loop automatically controls a variable using the principle of feedback. Automation within anesthesia typically aims to improve the stability of a controlled variable and reduce workload associated with simple repetitive tasks. This approach attempts to limit errors due to distractions or fatigue while simultaneously increasing compliance to evidence based perioperative protocols. The ultimate goal is to use these advantages over manual care to improve patient outcome. For more than twenty years, clinical studies in anesthesia have demonstrated the superiority of closed-loop systems compared to manual control for stabilizing a single variable, reducing practitioner workload, and safely administering therapies. This research has focused on various closed-loops that coupled inputs and outputs such as the processed electroencephalogram with propofol, blood pressure with vasopressors, and dynamic predictors of fluid responsiveness with fluid therapy. Recently, multiple simultaneous independent closed-loop systems have been tested in practice and one study has demonstrated a clinical benefit on postoperative cognitive dysfunction. Despite their advantages, these tools still require that a well-trained practitioner maintains situation awareness, understands how closed-loop systems react to each variable, and is ready to retake control if the closed-loop systems fail. In the future, multiple input multiple output closed-loop systems will control anesthetic, fluid and vasopressor titration and may perhaps integrate other key systems, such as the anesthesia machine. Human supervision will nonetheless always be indispensable as situation awareness, communication, and prediction of events remain irreplaceable human factors.

Application and Development of EEG Acquisition and Feedback Technology: A Review

This review focuses on electroencephalogram (EEG) acquisition and feedback technology and its core elements, including the composition and principles of the acquisition devices, a wide range of applications, and commonly used EEG signal classification algorithms. First, we describe the construction of EEG acquisition and feedback devices encompassing EEG electrodes, signal processing, and control and feedback systems, which collaborate to measure faint EEG signals from the scalp, convert them into interpretable data, and accomplish practical applications using control feedback systems. Subsequently, we examine the diverse applications of EEG acquisition and feedback across various domains. In the medical field, EEG signals are employed for epilepsy diagnosis, brain injury monitoring, and sleep disorder research. EEG acquisition has revealed associations between brain functionality, cognition, and emotions, providing essential insights for psychologists and neuroscientists. Brain-computer interface technology utilizes EEG signals for human-computer interaction, driving innovation in the medical, engineering, and rehabilitation domains. Finally, we introduce commonly used EEG signal classification algorithms. These classification tasks can identify different cognitive states, emotional states, brain disorders, and brain-computer interface control and promote further development and application of EEG technology. In conclusion, EEG acquisition technology can deepen the understanding of EEG signals while simultaneously promoting developments across multiple domains, such as medicine, science, and engineering.

Machine Learning, Deep Learning, and Closed Loop Devices-Anesthesia Delivery

With the tremendous volume of data captured during surgeries and procedures, critical care, and pain management, the field of anesthesiology is uniquely suited for the application of machine learning, neural networks, and closed loop technologies. In the past several years, this area has expanded immensely in both interest and clinical applications. This article provides an overview of the basic tenets of machine learning, neural networks, and closed loop devices, with emphasis on the clinical applications of these technologies.

High-Impact Clinical Studies That Fomented New Developments in Anesthesia: History of Achievements, 1966-2015

The aim of this work is to identify the most influential initial clinical studies that fomented important developments in anesthesiology over the past 50 years. Studies fomenting new development can be selected using vastly different approaches and, therefore, might provide diverse outcomes. In the present work, two basic aspects of study assessments – the stage of development (eg, generation of idea, preclinical studies, clinical trials) and the method of selection (eg, committee vote, various types of citation analysis, method of finding the invention disclosure) – were chosen according to the following model. The stage of development: the initial clinical studies demonstrating the basic advantage of an innovation for providing anesthesia. The method: a combination of two factors – the study priority in terms of the time of its publication and the degree of its acknowledgement in the form of citation impact; the time of study publication was regarded as a primary factor, but only if the study’s citation count was =/>20. The initial high-impact studies were selected for 16 drug-related topics (ketamine, isoflurane, etomidate, propofol, midazolam in anesthesia, vecuronium, alfentanil, atracurium, sevoflurane, sufentanil, rocuronium, desflurane, ropivacaine, remifentanil, dexmedetomidine in anesthesia, and sugammadex), and 9 technique-related topics (ultrasound-guided peripheral nerve block, capnography in anesthesia, target-controlled intravenous anesthesia, pulse oximetry in anesthesia, total intravenous anesthesia, transesophageal echocardiography in anesthesia, combined spinal-epidural anesthesia, and bispectral index). Twenty-five studies were designated the first high-impact studies (one for each topic); 16 are drug-related and 9 are technique-related. Half of the first high-impact studies had a citation count of =/>100, (range: 100 to 555). The citation count of the other half of high-impact studies did not reach the 100-citation threshold (range: 41 to 97). If a selected first high-impact study had a citation count <100, a next-on-timeline, additional study with citation count =/>100 was also selected; (range: 100 to 344). The present results show that an initial high-impact clinical study on a new development in anesthesiology can be determined and that related citations usually vary from one hundred to five hundred.

Closed-Loop Delivery Systems Versus Manually Controlled Administration of Total IV Anesthesia: A Meta-analysis of Randomized Clinical Trials

Bispectral Index Scale (BIS)-guided closed-loop delivery of anesthetics has been extensively studied. We performed a meta-analysis of all the randomized clinical trials comparing efficacy and performance between BIS-guided closed-loop delivery and manually controlled administration of total IV anesthesia. Scopus, PubMed, EMBASE, and the Cochrane Central Register of clinical trials were searched for pertinent studies. Inclusion criteria were random allocation to treatment and closed-loop delivery systems versus manually controlled administration of total IV anesthesia in any surgical setting. Exclusion criteria were duplicate publications and nonadult studies. Twelve studies were included, randomly allocating 1284 patients. Use of closed-loop anesthetic delivery systems was associated with a significant reduction in the dose of propofol administered for induction of anesthesia (mean difference [MD] = 0.37 [0.17-0.57], P for effect <0.00001, P for heterogeneity = 0.001, I = 74%) and a significant reduction in recovery time (MD = 1.62 [0.60-2.64], P for effect <0.0001, P for heterogeneity = 0.06, I = 47%). The target depth of anesthesia was preserved more frequently with closed-loop anesthetic delivery than with manual control (MD = -15.17 [-23.11 to -7.24], P for effect <0.00001, P for heterogeneity <0.00001, I = 83%). There were no differences in the time required to induce anesthesia and the total propofol dose. Closed-loop anesthetic delivery performed better than manual-control delivery. Both median absolute performance error and wobble index were significantly lower in closed-loop anesthetic delivery systems group (MD = 5.82 [3.17-8.46], P for effect <0.00001, P for heterogeneity <0.00001, I = 90% and MD = 0.92 [0.13-1.72], P for effect = 0.003, P for heterogeneity = 0.07, I = 45%). When compared with manual control, BIS-guided anesthetic delivery of total IV anesthesia reduces propofol requirements during induction, better maintains a target depth of anesthesia, and reduces recovery time.

A Comparison of Multiscale Permutation Entropy Measures in On-Line Depth of Anesthesia Monitoring

OBJECTIVE

Multiscale permutation entropy (MSPE) is becoming an interesting tool to explore neurophysiological mechanisms in recent years. In this study, six MSPE measures were proposed for on-line depth of anesthesia (DoA) monitoring to quantify the anesthetic effect on the real-time EEG recordings. The performance of these measures in describing the transient characters of simulated neural populations and clinical anesthesia EEG were evaluated and compared.

METHODS

Six MSPE algorithms-derived from Shannon permutation entropy (SPE), Renyi permutation entropy (RPE) and Tsallis permutation entropy (TPE) combined with the decomposition procedures of coarse-graining (CG) method and moving average (MA) analysis-were studied. A thalamo-cortical neural mass model (TCNMM) was used to generate noise-free EEG under anesthesia to quantitatively assess the robustness of each MSPE measure against noise. Then, the clinical anesthesia EEG recordings from 20 patients were analyzed with these measures. To validate their effectiveness, the ability of six measures were compared in terms of tracking the dynamical changes in EEG data and the performance in state discrimination. The Pearson correlation coefficient (R) was used to assess the relationship among MSPE measures.

RESULTS

CG-based MSPEs failed in on-line DoA monitoring at multiscale analysis. In on-line EEG analysis, the MA-based MSPE measures at 5 decomposed scales could track the transient changes of EEG recordings and statistically distinguish the awake state, unconsciousness and recovery of consciousness (RoC) state significantly. Compared to single-scale SPE and RPE, MSPEs had better anti-noise ability and MA-RPE at scale 5 performed best in this aspect. MA-TPE outperformed other measures with faster tracking speed of the loss of unconsciousness.

CONCLUSIONS

MA-based multiscale permutation entropies have the potential for on-line anesthesia EEG analysis with its simple computation and sensitivity to drug effect changes. CG-based multiscale permutation entropies may fail to describe the characteristics of EEG at high decomposition scales.

Clinical Electroencephalography for Anesthesiologists: Part I: Background and Basic Signatures

The widely used electroencephalogram-based indices for depth-of-anesthesia monitoring assume that the same index value defines the same level of unconsciousness for all anesthetics. In contrast, we show that different anesthetics act at different molecular targets and neural circuits to produce distinct brain states that are readily visible in the electroencephalogram. We present a two-part review to educate anesthesiologists on use of the unprocessed electroencephalogram and its spectrogram to track the brain states of patients receiving anesthesia care. Here in part I, we review the biophysics of the electroencephalogram and the neurophysiology of the electroencephalogram signatures of three intravenous anesthetics: propofol, dexmedetomidine, and ketamine, and four inhaled anesthetics: sevoflurane, isoflurane, desflurane, and nitrous oxide. Later in part II, we discuss patient management using these electroencephalogram signatures. Use of these electroencephalogram signatures suggests a neurophysiologically based paradigm for brain state monitoring of patients receiving anesthesia care.

Identification of scalp EEG circadian variation using a novel correlation sum measure

OBJECTIVE

In this paper, we propose a novel method to determine the circadian variation of scalp electroencephalogram (EEG) in both individual and group levels using a correlation sum measure, quantifying self-similarity of the EEG relative energy across waking epochs.

APPROACH

We analysed EEG recordings from central-parietal and occipito-parietal montages in nine healthy subjects undergoing a 72 h ultradian sleep-wake cycle protocol. Each waking epoch (∼ 1 s) of every nap opportunity was decomposed using the wavelet packet transform, and the relative energy for that epoch was calculated in the desired frequency band using the corresponding wavelet coefficients. Then, the resulting set of energy values was resampled randomly to generate different subsets with equal number of elements. The correlation sum of each subset was then calculated over a range of distance thresholds, and the average over all subsets was computed. This average value was finally scaled for each nap opportunity and considered as a new circadian measure.

MAIN RESULTS

According to the evaluation results, a clear circadian rhythm was identified in some EEG frequency ranges, particularly in 4-8 Hz and 10-12 Hz. The correlation sum measure not only was able to disclose the circadian rhythm on the group data but also revealed significant circadian variations in most individual cases, as opposed to previous studies only reporting the circadian rhythms on a population of subjects. Compared to a naive measure based on the EEG absolute energy in the frequency band of interest, the proposed measure showed a clear superiority using both individual and group data. Results also suggested that the acrophase (i.e., the peak) of the circadian rhythm in 10-12 Hz occurs close to the core body temperature minimum.

SIGNIFICANCE

These results confirm the potential usefulness of the proposed EEG-based measure as a non-invasive circadian marker.

A generalizable adaptive brain-machine interface design for control of anesthesia

Brain-machine interfaces (BMIs) for closed-loop control of anesthesia have the potential to automatically monitor and control brain states under anesthesia. Since a variety of anesthetic states are needed in different clinical scenarios, designing a generalizable BMI architecture that can control a wide range of anesthetic states is essential. In addition, drug dynamics are non-stationary over time and could change with the depth of anesthesia. Hence for precise control, a BMI needs to track these non-stationarities online. Here we design a BMI architecture that generalizes to control of various anesthetic states and their associated neural signatures, and is adaptive to time-varying drug dynamics. We provide a systematic approach to build general parametric models that quantify the anesthetic state and describe the drug dynamics. Based on these models, we develop an adaptive closed-loop controller within the framework of stochastic optimal feedback control. This controller tracks the non-stationarities in drug dynamics, achieves tight control in a time-varying environment, and removes the need for an offline system identification session. For robustness, the BMI also ensures small drug infusion rate variations at steady state. We test the BMI architecture for control of two common anesthetic states, i.e., burst suppression in medically-induced coma and unconsciousness in general anesthesia. Using numerical experiments, we find that the BMI generalizes to control of both these anesthetic states; in a time-varying environment, even without initial knowledge of model parameters, the BMI accurately controls these two different anesthetic states, reducing bias and error more than 70 times and 9 times, respectively, compared with a non-adaptive system.

A brain-machine interface for control of medically-induced coma

Medically-induced coma is a drug-induced state of profound brain inactivation and unconsciousness used to treat refractory intracranial hypertension and to manage treatment-resistant epilepsy. The state of coma is achieved by continually monitoring the patient's brain activity with an electroencephalogram (EEG) and manually titrating the anesthetic infusion rate to maintain a specified level of burst suppression, an EEG marker of profound brain inactivation in which bursts of electrical activity alternate with periods of quiescence or suppression. The medical coma is often required for several days. A more rational approach would be to implement a brain-machine interface (BMI) that monitors the EEG and adjusts the anesthetic infusion rate in real time to maintain the specified target level of burst suppression. We used a stochastic control framework to develop a BMI to control medically-induced coma in a rodent model. The BMI controlled an EEG-guided closed-loop infusion of the anesthetic propofol to maintain precisely specified dynamic target levels of burst suppression. We used as the control signal the burst suppression probability (BSP), the brain's instantaneous probability of being in the suppressed state. We characterized the EEG response to propofol using a two-dimensional linear compartment model and estimated the model parameters specific to each animal prior to initiating control. We derived a recursive Bayesian binary filter algorithm to compute the BSP from the EEG and controllers using a linear-quadratic-regulator and a model-predictive control strategy. Both controllers used the estimated BSP as feedback. The BMI accurately controlled burst suppression in individual rodents across dynamic target trajectories, and enabled prompt transitions between target levels while avoiding both undershoot and overshoot. The median performance error for the BMI was 3.6%, the median bias was -1.4% and the overall posterior probability of reliable control was 1 (95% Bayesian credibility interval of [0.87, 1.0]). A BMI can maintain reliable and accurate real-time control of medically-induced coma in a rodent model suggesting this strategy could be applied in patient care.

Brain-machine interfaces (BMI) for closed-loop control of anesthesia have the potential to enable fully automated and precise control of brain states in patients requiring anesthesia care. Medically-induced coma is one such drug-induced state in which the brain is profoundly inactivated and unconscious and the electroencephalogram (EEG) pattern consists of bursts of electrical activity alternating with periods of suppression, termed burst suppression. Medical coma is induced to treat refractory intracranial hypertension and uncontrollable seizures. The state of coma is often required for days, making accurate manual control infeasible. We develop a BMI that can automatically and precisely control the level of burst suppression in real time in individual rodents. The BMI consists of novel estimation and control algorithms that take as input the EEG activity, estimate the burst suppression level based on this activity, and use this estimate as feedback to control the drug infusion rate in real time. The BMI maintains precise control and promptly changes the level of burst suppression while avoiding overshoot or undershoot. Our work demonstrates the feasibility of automatic reliable and accurate control of medical coma that can provide considerable therapeutic benefits.

Real-time closed-loop control in a rodent model of medically induced coma using burst suppression

BACKGROUND

A medically induced coma is an anesthetic state of profound brain inactivation created to treat status epilepticus and to provide cerebral protection after traumatic brain injuries. The authors hypothesized that a closed-loop anesthetic delivery system could automatically and precisely control the electroencephalogram state of burst suppression and efficiently maintain a medically induced coma.

METHODS

In six rats, the authors implemented a closed-loop anesthetic delivery system for propofol consisting of: a computer-controlled pump infusion, a two-compartment pharmacokinetics model defining propofol's electroencephalogram effects, the burst-suppression probability algorithm to compute in real time from the electroencephalogram the brain's burst-suppression state, an online parameter-estimation procedure and a proportional-integral controller. In the control experiment each rat was randomly assigned to one of the six burst-suppression probability target trajectories constructed by permuting the burst-suppression probability levels of 0.4, 0.65, and 0.9 with linear transitions between levels.

RESULTS

In each animal the controller maintained approximately 60 min of tight, real-time control of burst suppression by tracking each burst-suppression probability target level for 15 min and two between-level transitions for 5-10 min. The posterior probability that the closed-loop anesthetic delivery system was reliable across all levels was 0.94 (95% CI, 0.77-1.00; n = 18) and that the system was accurate across all levels was 1.00 (95% CI, 0.84-1.00; n = 18).

CONCLUSION

The findings of this study establish the feasibility of using a closed-loop anesthetic delivery systems to achieve in real time reliable and accurate control of burst suppression in rodents and suggest a paradigm to precisely control medically induced coma in patients.

A closed-loop anesthetic delivery system for real-time control of burst suppression

OBJECTIVE

There is growing interest in using closed-loop anesthetic delivery (CLAD) systems to automate control of brain states (sedation, unconsciousness and antinociception) in patients receiving anesthesia care. The accuracy and reliability of these systems can be improved by using as control signals electroencephalogram (EEG) markers for which the neurophysiological links to the anesthetic-induced brain states are well established. Burst suppression, in which bursts of electrical activity alternate with periods of quiescence or suppression, is a well-known, readily discernible EEG marker of profound brain inactivation and unconsciousness. This pattern is commonly maintained when anesthetics are administered to produce a medically-induced coma for cerebral protection in patients suffering from brain injuries or to arrest brain activity in patients having uncontrollable seizures. Although the coma may be required for several hours or days, drug infusion rates are managed inefficiently by manual adjustment. Our objective is to design a CLAD system for burst suppression control to automate management of medically-induced coma.

APPROACH

We establish a CLAD system to control burst suppression consisting of: a two-dimensional linear system model relating the anesthetic brain level to the EEG dynamics; a new control signal, the burst suppression probability (BSP) defining the instantaneous probability of suppression; the BSP filter, a state-space algorithm to estimate the BSP from EEG recordings; a proportional-integral controller; and a system identification procedure to estimate the model and controller parameters.

MAIN RESULTS

We demonstrate reliable performance of our system in simulation studies of burst suppression control using both propofol and etomidate in rodent experiments based on Vijn and Sneyd, and in human experiments based on the Schnider pharmacokinetic model for propofol. Using propofol, we further demonstrate that our control system reliably tracks changing target levels of burst suppression in simulated human subjects across different epidemiological profiles.

SIGNIFICANCE

Our results give new insights into CLAD system design and suggest a control-theory framework to automate second-to-second control of burst suppression for management of medically-induced coma.

A brain-machine interface for control of burst suppression in medical coma

Burst suppression is an electroencephalogram (EEG) marker of profound brain inactivation and unconsciousness and consists of bursts of electrical activity alternating with periods of isoelectricity called suppression. Burst suppression is the EEG pattern targeted in medical coma, a drug-induced brain state used to help recovery after brain injuries and to treat epilepsy that is refractory to conventional drug therapies. The state of coma is maintained manually by administering an intravenous infusion of an anesthetic, such as propofol, to target a pattern of burst suppression on the EEG. The coma often needs to be maintained for several hours or days, and hence an automated system would offer significant benefit for tight control. Here we present a brain-machine interface (BMI) for automatic control of burst suppression in medical coma that selects the real-time drug infusion rate based on EEG observations and can precisely control the burst suppression level in real time in rodents. We quantify the burst suppression level using the burst suppression probability (BSP), the brain's instantaneous probability of being in the suppressed state, and represent the effect of the anesthetic propofol on the BSP using a two-dimensional linear compartment model that we fit in experiments. We compute the BSP in real time from the EEG segmented into a binary time-series by deriving a two-dimensional state-space algorithm. We then derive a stochastic controller using both a linear-quadratic-regulator strategy and a model predictive control strategy. The BMI can promptly change the level of burst suppression without overshoot or undershoot and maintains precise control of time-varying target levels of burst suppression in individual rodents in real time.

Influence of real-time Bayesian forecasting of pharmacokinetic parameters on the precision of a rocuronium target-controlled infusion

INTRODUCTION

Bayesian forecasting has been shown to improve the accuracy of pharmacokinetic/pharmacodynamic (PK/PD) models by adding measured values to a population model. It could be done in real time for neuromuscular blockers (NMB) using measured values of effect. This study was designed to assess feasibility and benefit of Bayesian forecasting during a rocuronium target-controlled infusion (TCI).

METHODS

After internal review board (IRB) approval and informed consent, 21 women scheduled for breast plastic surgery were included. Anesthesia was maintained with propofol, alfentanil, and controlled ventilation through a laryngeal mask. Rocuronium was delivered in TCI with Stanpump software and the Plaud population model. The target effect was 50% blockade until insertion of breast prosthesis; thereafter it was set to 0%. Response to train of four (TOF) at adductor pollicis was recorded using a force transducer. In ten patients, drug delivery was based on the population model. In the others, repeated measures values were entered in the software, and the PK model was adjusted to minimize the error in predicted effect. Model precision was compared between groups using mean prediction error and mean absolute prediction error.

RESULTS

At target 50%, model accuracy was not improved with Bayesian adjustments; conversely, post-infusion errors were significantly decreased. The first two measures had the most influence on the model changes.

DISCUSSION

Below clinical utility, such adjustments may be used to explore cofactors influencing interindividual and intraindividual variability in NMB dose-response relationship. Similar tools may also be developed for drugs in which a quantitative effect is available, such as electroencephalography (EEG) for hypnotics.

IMPLICATION

Real-time Bayesian forecasting combining measured values of effect with a population model is suitable to guide NMB-agent delivery using Stanpump software.

Review article: closed-loop systems in anesthesia: is there a potential for closed-loop fluid management and hemodynamic optimization?

Closed-loop (automated) controllers are encountered in all aspects of modern life in applications ranging from air-conditioning to spaceflight. Although these systems are virtually ubiquitous, they are infrequently used in anesthesiology because of the complexity of physiologic systems and the difficulty in obtaining reliable and valid feedback data from the patient. Despite these challenges, closed-loop systems are being increasingly studied and improved for medical use. Two recent developments have made fluid administration a candidate for closed-loop control. First, the further description and development of dynamic predictors of fluid responsiveness provides a strong parameter for use as a control variable to guide fluid administration. Second, rapid advances in noninvasive monitoring of cardiac output and other hemodynamic variables make goal-directed therapy applicable for a wide range of patients in a variety of clinical care settings. In this article, we review the history of closed-loop controllers in clinical care, discuss the current understanding and limitations of the dynamic predictors of fluid responsiveness, and examine how these variables might be incorporated into a closed-loop fluid administration system.

Electroencephalographic Narcotrend index, spectral edge frequency and median power frequency as guide to anaesthetic depth for cardiac surgery in laboratory sheep

In order to provide objective measures of anaesthetic depth for periods without clinical signs (e.g., cardiopulmonary bypass [CPB]), the correlation between the electroencephalographic Narcotrend index (NI), 95% spectral edge frequency (SEF95), median power frequency (MPF) and clinical stages of anaesthesia was investigated in 16 juvenile sheep. Data were recorded during recovery from anaesthesia for pulmonary or aortic valve replacement. A significant (P<0.05) negative correlation was found between clinical stages of anaesthesia and NI (r(s) = -0.534) and SEF95 (r(s) = -0.543). No significant correlation existed between anaesthetic stages and MPF (r(s) = -0.292, P>0.05). The sensitivity of NI, SEF95 and MPF to assure an adequate level of anaesthesia was 71.43%, 66.67% and 44.44%, respectively, while the specificity of the descriptors ranged between 97.44 and 92.31%. No significant age-related effect on EEG data and stages of anaesthesia was detected when data from sheep between 4 and 6 months of age were compared with data obtained from 8- to 12-month old sheep. In conclusion, NI seems to be the most appropriate EEG descriptor to assure adequate depth of anaesthesia in juvenile, isoflurane-anaesthetized sheep.

Reinforcement learning versus proportional-integral-derivative control of hypnosis in a simulated intraoperative patient

BACKGROUND

Research has demonstrated the efficacy of closed-loop control of anesthesia using bispectral index (BIS) as the controlled variable. Model-based and proportional-integral-derivative (PID) controllers outperform manual control. We investigated the application of reinforcement learning (RL), an intelligent systems control method, to closed-loop BIS-guided, propofol-induced hypnosis in simulated intraoperative patients. We also compared the performance of the RL agent against that of a conventional PID controller.

METHODS

The RL and PID controllers were evaluated during propofol induction and maintenance of hypnosis. The patient-hypnotic episodes were designed to challenge both controllers with varying degrees of interindividual variation and noxious surgical stimulation. Each controller was tested in 1000 simulated patients, and control performance was assessed by calculating the median performance error (MDPE), median absolute performance error (MDAPE), Wobble, and Divergence for each controller group. A separate analysis was performed for the induction and maintenance phases of hypnosis.

RESULTS

During maintenance, RL control demonstrated an MDPE of -1% and an MDAPE of 3.75%, with 80% of the time at BIS(target) ± 5. The PID controller yielded a MDPE of -8.5% and an MDAPE of 8.6%, with 57% of the time at BIS(target) ± 5. In comparison, the MDAPE in the worst-controlled patient of the RL group was observed to be almost half that of the worst-controlled patient in the PID group.

CONCLUSIONS

When compared with the PID controller, RL control resulted in slower induction but less overshoot and faster attainment of steady state. No difference in interindividual patient variation and noxious destabilizing challenge on control performance was observed between the 2 patient groups.

Closed-loop control for intensive care unit sedation

The potential clinical applications of active control for pharmacology in general, and anesthesia and critical care unit medicine in particular, are clearly apparent. Specifically, monitoring and controlling the depth of anesthesia in surgery and the intensive care unit is of particular importance. Nonnegative and compartmental models provide a broad framework for biological and physiological systems, including clinical pharmacology, and are well suited for developing models for closed-loop control for drug administration. These models are derived from mass and energy balance considerations that involve dynamic states whose values are nonnegative and are characterized by conservation laws (e.g., mass, energy, fluid, etc.) capturing the exchange of material between kinetically homogenous entities called compartments. Compartmental models have been particularly important for understanding pharmacokinetics and pharmacodynamics. One of the basic motivations for pharmacokinetic/pharmacodynamic research is to improve drug delivery. In critical care medicine it is current clinical practice to administer potent drugs that profoundly influence levels of consciousness, respiratory, and cardiovascular function by manual control based on the clinician's experience and intuition. Open-loop control (manual control) by clinical personnel can be tedious, imprecise, time-consuming, and sometimes of poor quality, depending on the skills and judgement of the clinician. Closed-loop control based on appropriate dynamical systems models merits investigation as a means of improving drug delivery in the intensive care unit. In this article, we discuss the challenges and opportunities of feedback control using nonnegative and compartmental system theory for the specific problem of closed-loop control of intensive care unit sedation. Several closed-loop control paradigms are investigated including adaptive control, neural network adaptive control, optimal control, and hybrid adaptive control algorithms for intensive care unit sedation.

Contributions of PK/PD modeling to intravenous anesthesia

Pharmacokinetic (PK)/pharmacodynamic (PD) modeling has made an enormous contribution to intravenous anesthesia. PK/PD models have provided us with insight into the factors affecting the onset and offset of drug effect. For example, we are now able to describe the influence of cardiac output on the disposition of intravenous drugs within the first few minutes after administration of the drug. We are able to calculate intravenous loading doses that allow for the delay between the concentration of the drug in the plasma and the rising concentration at the site of drug effect. We are able to achieve and maintain a stable level of anesthetic effect using computerized infusion pumps that target the site of drug effect rather than the plasma. Importantly, on the basis of models of drug interaction and an understanding of how drug offset varies with duration of administration, we are now able to rationally combine hypnotics and opioids.

Titration automatisée du propofol guidée par l'index bispectral

This review analyzes the clinical studies concerning the automated perfusion, or closed-loop, of propofol guided by the bispectral index (BIS). To carry out the maintenance of general anaesthesia by a closed loop propofol-BIS is feasible as shown by studies comprising few low risk patients. We showed that induction of anaesthesia is feasible with a closed loop, haemodynamic stability being similar to a manual titration. A second study, bearing on the whole of the anaesthesia of patients ASA I to III undergoing very diverse surgical acts, showed that the closed loop propofol-BIS was more precise than a manual perfusion. This confirms that the closed loop propofol-BIS is not an esoteric research and that it represents a tool with a future for the clinician.

Neural network adaptive output feedback control for intensive care unit sedation and intraoperative anesthesia

The potential applications of neural adaptive control for pharmacology, in general, and anesthesia and critical care unit medicine, in particular, are clearly apparent. Specifically, monitoring and controlling the depth of anesthesia in surgery is of particular importance. Nonnegative and compartmental models provide a broad framework for biological and physiological systems, including clinical pharmacology, and are well suited for developing models for closed-loop control of drug administration. In this paper, we develop a neural adaptive output feedback control framework for nonlinear uncertain nonnegative and compartmental systems with nonnegative control inputs. The proposed framework is Lyapunov-based and guarantees ultimate boundedness of the error signals. In addition, the neural adaptive controller guarantees that the physical system states remain in the nonnegative orthant of the state space. Finally, the proposed approach is used to control the infusion of the anesthetic drug propofol for maintaining a desired constant level of depth of anesthesia for noncardiac surgery.

Pharmakokinetische/pharmakodynamische Modelle für Inhalationsanästhetika

Pharmacokinetic models can be differentiated into two groups: physiological-based models and empirical models. Traditionally the pharmacokinetics of volatile anaesthetics are described using physiological-based models together with the respective tissue-blood distribution coefficients. The compartments of the empirical model have no anatomical equivalents and are merely the product of the mathematical procedure for parameter estimation. The end expiratory concentration of volatile anaesthetics is approximately equal to the arterial concentration and, therefore, the description of the transition between plasma and effect site for volatile anaesthetics plays a central role. The most important parameter here is the k(e0) value which is a time constant and describes the time delay for the transition from the central compartment to the calculated effect compartment. The k(e0) values for sevoflurane and isoflurane are the same but the concentration balance between the end-tidal concentration and the effect compartment occurs twice as quickly with desflurane. In clinical practice volatile anaesthetics are normally combined with N(2)O and/or opioids. This results in an additive interaction between volatile anaesthetics and N(2)O but a synergistic interaction of volatile anaesthetics with opioids. However, there are relatively few investigations on the interactions between the clinically widely used combination of volatile anaesthetics, N(2)O and opioids.

Bispectral index, spectral edge frequency 95% and median frequency recorded at varying desflurane concentrations in pigs

The objective of this study is to evaluate the usefulness of bispectral index (BIS), spectral edge frequency 95% (SEF) and median frequency (MED) in relation to a simple descriptive scale (SDS) as indicators of anaesthetic depth at different desflurane concentrations in swine. Sixteen pigs were randomly allocated to four groups. Electroencephalograms (EEG) were recorded during desflurane anaesthesia, and BIS, SEF and MED were calculated from the EEG. The agent was administered in pure oxygen at 1, 1.25, 1.5 and 1.7 MAC in randomized order. Anaesthetic depth was evaluated on a SDS. BIS decreased significantly (P<0.001) at the different anaesthetic dosages used. SEF decreased significantly (P<0.001) from basal to 1 MAC of desflurane. MED decreased significantly (P<0.001) from basal to 1 MAC and from 1 to 1.75 MAC. Good correlation was seen between SDS scores and BIS values and between SDS scores and MED values. BIS appeared to be useful to predict changes in anaesthetic depth at clinically used dosages of inhalant anaesthesia.

Quantifying cortical activity during general anesthesia using wavelet analysis

This paper reports on a novel method for quantifying the cortical activity of a patient during general anesthesia as a surrogate measure of the patient's level of consciousness. The proposed technique is based on the analysis of a single-channel (frontal) electroencephalogram (EEG) signal using stationary wavelet transform (SWT). The wavelet coefficients calculated from the EEG are pooled into a statistical representation, which is then compared to two well-defined states: the awake state with normal EEG activity, and the isoelectric state with maximal cortical depression. The resulting index, referred to as the wavelet-based anesthetic value for central nervous system monitoring (WAV(CNS)), quantifies the depth of consciousness between these two extremes. To validate the proposed technique, we present a clinical study which explores the advantages of the WAV(CNS) in comparison with the BIS monitor (Aspect Medical Systems, MA), currently a reference in consciousness monitoring. Results show that the WAV(CNS) and BIS are well correlated (r = 0.969) during periods of steady-state despite fundamental algorithmic differences. However, in terms of dynamic behavior, the WAV(CNS) offers faster tracking of transitory changes at induction and emergence, with an average lead of 15-30 s. Furthermore, and conversely to the BIS, the WAV(CNS) regains its preinduction baseline value when patients are responding to verbal command after emergence from anesthesia. We conclude that the proposed analysis technique is an attractive alternative to BIS monitoring. In addition, we show that the WAV(CNS) dynamics can be modeled as a linear time invariant transfer function. This index is, therefore, well suited for use as a feedback sensor in advisory systems, closed-loop control schemes, and for the identification of the pharmacodynamic models of anesthetic drugs.

Classic electroencephalographic parameters: Median frequency, spectral edge frequency etc

Even today many anaesthesiologists rely on parameters of the autonomic nervous system, such as blood pressure and heart rate to decide if a patient is adequately anaesthetized. It is thought that the electroencephalogram (EEG) may provide more information on the state of anaesthesia. Because full EEG analysis is not possible in the operating room, processed EEG parameters have been developed comprising complex information into a single value. Time and frequency domain parameters are calculated. The power spectrum results from a Fourier analysis and can be described by parameters such as median frequency, spectral edge frequency and others. It was noted, however, that anaesthetics at low doses increase frequency of the EEG, whereas at high doses the EEG is depressed. This biphasic response makes it difficult to clearly distinguish the exact anaesthetic state of a patient. Median frequency and spectral edge frequency have been studied in numerous studies. However, no sole indicator has been derived from the EEG that could serve as a descriptor of anaesthetic depth.

Closed loops in anaesthesia

Closed-loop systems are able to make their own decisions and to try to reach and maintain a preset target. As a result, they might help the anaesthetist to optimise the titration of drug administration without any overshoot, controlling physiological functions and guiding monitoring variables. Thanks to the development of fast computer technology and more reliable pharmacological effect measures, the study of automation in anaesthesia has regained popularity. This short review focuses on the most recently developed and tested feedback systems in anaesthesia. Various new approaches for controlling the administration of intravenous and inhaled hypnotic-anaesthetic drugs have recently been published. For analgesics, a framework for further research has been presented in the literature. For other drugs, such as muscle relaxants and haemodynamic agents, only short reviews can be found. Until now, most of these systems have had to be under development. The challenge is now fully to establish the safety, efficacy, reliability and utility of closed-loop anaesthesia so that it can be adopted in the clinical setting. Besides, their role in optimising the controlled variables and control models, these systems have to be tested in extreme circumstances in order to test their robustness.

Passivity-Based Neural Network Adaptive Output Feedback Control for Nonlinear Nonnegative Dynamical Systems

The potential clinical applications of adaptive neural network control for pharmacology in general, and anesthesia and critical care unit medicine in particular, are clearly apparent. Specifically, monitoring and controlling the depth of anesthesia in surgery is of particular importance. Nonnegative and compartmental models provide a broad framework for biological and physiological systems, including clinical pharmacology, and are well suited for developing models for closed-loop control of drug administration. In this paper, we develop a neural adaptive output feedback control framework for adaptive set-point regulation of nonlinear uncertain nonnegative and compartmental systems. The proposed framework is Lyapunov-based and guarantees ultimate boundedness of the error signals corresponding to the physical system states and the neural network weighting gains. The approach is applicable to nonlinear nonnegative systems with unmodeled dynamics of unknown dimension and guarantees that the physical system states remain in the nonnegatiye orthant of the state-space for nonnegative initial conditions. Finally, a numerical example involving the infusion of the anesthetic drug midazolam for maintaining a desired constant level of depth of anesthesia for noncardiac surgery is provided to demonstrate the efficacy of the proposed approach.

Noxious stimulation during orthopaedic surgery results in EEG 'arousal' or 'paradoxical arousal' reaction in isoflurane-anaesthetised sheep

The effects of surgical stimuli on haemodynamic and electroencephalographic (EEG) variables were determined in 25 adult ewes undergoing an experimental orthopaedic procedure in isoflurane anaesthesia. Data were recorded after 15 min of constant end-tidal concentration of approximately 2.2% isoflurane (SS: steady state=baseline), during skin disinfection (DIS), incision (INC), drilling of the first hole through the tibia (DRI) and insertion of a threaded pin (PIN) for external fixation. Stimulation resulted in a significant increase in systolic and mean arterial pressure above SS at INC, DRI and PIN. Haemodynamic changes were accompanied by either significant increases or decreases in EEG median frequency (MF) and 80% spectral edge frequency (SEF80) above or below SS at all four stimulation time points suggesting 'arousal' or 'paradoxical arousal' reaction, respectively. We conclude, that either type of EEG activation pattern could be elicited dependent on stimulation intensity and level of anaesthetic depth.

Bispectral index, spectral edge frequency 95%, and median frequency recorded for various concentrations of isoflurane and sevoflurane in pigs

OBJECTIVE

To evaluate bispectral index (BIS), spectral edge frequency 95% (SEF), and median frequency (MED) in relation to a visual analogue scale (VAS) as indicators of anesthetic depth for various concentrations of sevoflurane and isoflurane in pigs.

ANIMALS

32 pigs.

PROCEDURE

Pigs were randomly allocated to 8 groups (4 pigs/group). An electroencephalogram (EEG) was recorded in each conscious pig. Pigs were then anesthetized by use of sevoflurane (n = 16) or isoflurane (16). Agents were administered in oxygen at minimum alveolar concentrations (MACs) of 1, 1.25, 1.5, and 1.75 MAC in a randomized order. End-tidal sevoflurane and isoflurane concentrations were maintained for 30 minutes, after which an EEG was recorded for 5 minutes; BIS, SEF, and MED were then calculated. Anesthetic depth was evaluated by use of the VAS. Cardiovascular and EEG responses to nociceptive stimuli were evaluated for each anesthetic agent.

RESULTS

BIS decreased significantly for the various concentrations of each anesthetic. At equivalent MACs, BIS values were significantly higher during sevoflurane-induced anesthesia than during isoflurane-induced anesthesia. Values of MED and SEF decreased significantly from basal values to 1 MAC of sevoflurane and isoflurane. For both agents, there was good correlation between VAS scores and BIS values and between VAS scores and SEF values.

CONCLUSIONS AND CLINICAL RELEVANCE

BIS was useful for predicting changes in anesthetic depth at clinical dosages of inhalant anesthetics. Values of BIS, SEF, and MED were significantly higher during anesthesia induced by administration of sevoflurane than during anesthesia induced by administration of isoflurance at equivalent MACs.

Isoflurane dosage for equivalent intraoperative electroencephalographic suppression in patients with and without epidural blockade

We conducted a prospective, randomized, controlled trial to establish the effect of epidural blockade on isoflurane requirements for equivalent intraoperative electroencephalographic (EEG) suppression. Fifty patients undergoing abdominal hysterectomy received combined epidural and general anesthesia or general anesthesia alone with isoflurane and alfentanil. Isoflurane was administered by computer-controlled closed-loop feedback to maintain an EEG 95% spectral edge frequency of 17.5 Hz, a target chosen on the basis of a pilot study. In epidural patients, end-tidal isoflurane concentration (FE'(ISO)) was 0.19% smaller (95% confidence interval [CI], -0.32% to -0.06%; P < 0.01), mean arterial blood pressure was 17 mm Hg lower (95% CI, -24 to -9 mm Hg; P < 0.0001), and body temperature was 0.4 degrees C lower (95% CI, -0.7 to 0 degrees C; P < 0.05) than in controls. EEG bispectral index (BIS) was 4 points higher (95% CI, 1 to 7; P < 0.05). EEG median frequency and heart rate were similar in both groups. Epidural patients were 76% more likely (95% CI, 58% to 94%; P < 0.001) to require metaraminol for hypotension and were 28% more likely (95% CI, 3% to 53%; P < 0.05) to require glycopyrrolate for bradycardia. After surgery, the time to eye opening in epidural patients was 2.3 min shorter (95% CI, -4.2 to -0.5 min; P < 0.05). Time to eye opening correlated better with FE'(ISO) in the last 30 s of anesthesia (FE'(ISO) = 0.07 x time to eye opening + 0.31; r(2) = 0.59; P < 0.0001) than with BIS from the same period (BIS = 64 - 1.25 x time to eye opening; r(2) = 0.22; P < 0.001) (P < 0.0001). To maintain similar intraoperative spectral edge frequency, patients receiving combined epidural and general anesthesia require 21% less isoflurane than those receiving general anesthesia alone. This smaller isoflurane dose is associated with faster emergence from anesthesia.

IMPLICATIONS

The dose of general anesthetic required to maintain similar intraoperative suppression of brain electrical activity is 21% less in patients with nerve blockade than in those without. This dose reduction results in faster waking times in patients with nerve blockade, which may reflect lighter intraoperative anesthesia. The dose of general anesthetic required to maintain similar intraoperative suppression of brain electrical activity is 21% less in patients with nerve blockade than in those without. This dose reduction results in faster waking times in patients with nerve blockade, which may reflect lighter intraoperative anesthesia.

[Automatic analysis of electroencephalogram: what is its value in the year 2000 for monitoring anesthesia depth?]

Spontaneous EEG has been proposed for monitoring depth of anaesthesia and adjusting anesthetic drugs doses. This review describes the main features of spontaneous EEG, the principles of EEG signal analysis used in anaesthesia, and the EEG effects of the different anesthetic drugs in adults and children. Then, the correlations between EEG parameters changes and clinical signs of anesthesia (loss of consciousness and memory, lack of movement and haemodynamic stability despite noxious stimulations) are analyzed. The best signal analysis technique available today for routine use seems to be bispectral analysis, which returns, in the available monitors, a single number called bispectral index or BIS. Based upon the recent literature, clinical uses, performances and limits of use of BIS are described and discussed.

Modeling and closed-loop control of hypnosis by means of bispectral index (BIS) with isoflurane

A model-based closed-loop control system is presented to regulate hypnosis with the volatile anesthetic isoflurane. Hypnosis is assessed by means of the bispectral index (BIS), a processed parameter derived from the electroencephalogram. Isoflurane is administered through a closed-circuit respiratory system. The model for control was identified on a population of 20 healthy volunteers. It consists of three parts: a model for the respiratory system, a pharmacokinetic model and a pharmacodynamic model to predict BIS at the effect compartment. A cascaded internal model controller is employed. The master controller compares the actual BIS and the reference value set by the anesthesiologist and provides expired isoflurane concentration references to the slave controller. The slave controller maneuvers the fresh gas anesthetic concentration entering the respiratory system. The controller is designed to adapt to different respiratory conditions. Anti-windup measures protect against performance degradation in the event of saturation of the input signal. Fault detection schemes in the controller cope with BIS and expired concentration measurement artifacts. The results of clinical studies on humans are presented.

Exploring the IBIS data library contents: tools for data visualisation, (pre-) processing and screening. Improved Monitoring for Brain Dysfunction in Intensive Care and Surgery

During the IBIS project a high-quality data library of continuous and intermittent physiological signals and variables from patients during intensive care and surgery has been collected. To facilitate exploration of the full content of this data library a data browser was developed, which offers a flexible graphical display of the collection of multivariate data. To supplement the functionality of the display of the 'raw' data, a set of screening and pre-processing tools has been developed. A separate trend analysis tool offers a convenient overview of an entire recording focusing on the slow changes in the general state of the patient and the interaction between different physiological subsystems seen from a long-term perspective. A frequency analysis tool for processing the electroencephalography (EEG) signals has been integrated in the data browser to facilitate a quick screening of the cerebral function. The data library is the foundation of the development and validation of biosignal interpretation methods. This process can potentially be more productive using the described tool for algorithm prototyping based on a graphical network specifying the interaction between data processing primitives.

Closed loop control of anaesthesia: an assessment of the bispectral index as the target of control

We investigated the performance of a closed-loop system for administration of general anaesthesia, using the bispectral index as a target for control. One hundred patients undergoing gynaecological or general surgery were studied. In 60 patients, anaesthesia was maintained by intravenous infusion of a propofol/alfentanil mixture. In 40, an isoflurane/nitrous oxide based technique was used. For each technique, patients were randomly allocated to receive either closed-loop or manually controlled administration of the relevant agents (propofol/alfentanil or isoflurane), with an intra-operative target bispectral index of 50 in all cases. Closed-loop and manually controlled administration of anaesthesia resulted in similar intra-operative conditions and initial recovery characteristics. During maintenance of anaesthesia, cardiovascular and electro-encephalographic variables did not differ between closed-loop and manual control groups and deviation of bispectral index from the target value was similar. Intra-operative concentrations of propofol, alfentanil and isoflurane were within normal clinical ranges. Episodes of light anaesthesia were more common in the closed-loop group for patients receiving propofol/alfentanil anaesthesia and in the manual group for patients receiving isoflurane/nitrous oxide anaesthesia. Convenience aside, the closed-loop system showed no clinical advantage over conventional, manually adjusted techniques of anaesthetic administration.

Power spectral analysis of the electroencephalogram during increasing end-expiratory concentrations of isoflurane, desflurane and sevoflurane

We studied the effects of increasing end-expiratory concentrations of isoflurane (0.3, 0.6, 0.9, 1.2 vol.%) (n = 12 patients), desflurane (1.5, 3.0, 4.5, 6.0 vol.%) (n = 12 patients) and sevoflurane (0.5, 1.0, 1.5, 2.0 vol.%) (n = 12 patients) on power spectral analysis of the electroencephalogram (EEG). Spectral edge frequency (SEF), total power (TP) and relative power in the delta, theta, alpha and beta band were calculated. EEG changes were very similar within the three groups. SEF decreased, TP and relative power in the delta and theta band increased, power in the beta band decreased in a dose-dependent fashion with comparable regression lines. This indicates that MAC equivalent administration of isoflurane, desflurane and sevoflurane in clinically applied dose ranges is associated with equipotent EEG suppression.

Clinical utility of EEG parameters to predict loss of consciousness and response to skin incision during total intravenous anaesthesia

We studied 30 female patients undergoing elective surgery, to assess the reliability of electroencephalogram spectral edge frequency and median frequency to predict loss of consciousness and movement in response to skin incision during total intravenous anaesthesia. Each patient received a different combination of propofol (1, 2, 3, 4, 5 or 6 micrograms.ml-1) and sufentanil (0.1, 0.2, 0.3, 0.5 or 1.0 ng.ml-1) target concentrations for induction of anaesthesia using target controlled infusions, assigned randomly. In a logistic regression model, spectral edge frequency was a significant determinant of both loss of consciousness (p = 0.0006) and movement to skin incision (p = 0.0044), whereas for median frequency no significant prediction model could be established. The probabilities of 50% and 95% no response for spectral edge frequency were 13.4 Hz and 6.8 Hz, respectively. The variability of the data limited the predictive value, so that spectral edge frequency was a poor predictor and median frequency was no predictor of response in the individual patient during total intravenous propofol/sufentanil anaesthesia.