Depth of anaesthesia monitors and the latest algorithms

Harnessing machine learning for EEG signal analysis: Innovations in depth of anaesthesia assessment

Anaesthesia, crucial to surgical practice, is undergoing renewed scrutiny due to the integration of artificial intelligence in its medical use. The precise control over the temporary loss of consciousness is vital to ensure safe, pain-free procedures. Traditional methods of depth of anaesthesia (DoA) assessment, reliant on physical characteristics, have proven inconsistent due to individual variations. In response, electroencephalography (EEG) techniques have emerged, with indices such as the Bispectral Index offering quantifiable assessments. This literature review explores the current scope and frontier of DoA research, emphasising methods utilising EEG signals for effective clinical monitoring. This review offers a critical synthesis of recent advances, specifically focusing on electroencephalography (EEG) techniques and their role in enhancing clinical monitoring. By examining 117 high-impact papers, the review delves into the nuances of feature extraction, model building, and algorithm design in EEG-based DoA analysis. Comparative assessments of these studies highlight their methodological approaches and performance, including clinical correlations with established indices like the Bispectral Index. The review identifies knowledge gaps, particularly the need for improved collaboration for data access, which is essential for developing superior machine learning models and real-time predictive algorithms for patient management. It also calls for refined model evaluation processes to ensure robustness across diverse patient demographics and anaesthetic agents. The review underscores the potential of technological advancements to enhance precision, safety, and patient outcomes in anaesthesia, paving the way for a new standard in anaesthetic care. The findings of this review contribute to the ongoing discourse on the application of EEG in anaesthesia, providing insights into the potential for technological advancement in this critical area of medical practice.

Possibility of Using Surgical Pleth Index in Predicting Postoperative Pain in Patients after Vitrectomy Performed under General Anesthesia

Adequacy of anesthesia concept (AoA) in the guidance of general anesthesia (GA) is based on entropy, and it also reflects the actual depth of anesthesia and the surgical pleth index (SPI). Therefore, this study aimed to analyze the potential existence of relationships between SPI values at certain stages of the AoA-guided GA for vitreoretinal surgeries (VRS) and the incidence of intolerable postoperative pain perception (IPPP). A total of 175 patients were each assigned to one of five groups. In the first, the VRS procedure was performed under GA without premedication; in the second group, patients received metamizole before GA; in the third, patients received acetaminophen before GA; in the fourth group, patients received Alcaine before GA; and, in the peribulbar block group, the patients received a peribulbar block with a mix of the solutions of lignocaine and bupivacaine. Between the patients declaring mild and statistically significant differences in the IPPP in terms of SPI values before induction (52.3 ± 18.8 vs. 63.9 ± 18.1, p < 0.05) and after emergence from GA (51.1 ± 13 vs. 68.1 ± 8.8; p < 0.001), it was observed that the patients postoperatively correlated with heart rate variations despite the group allocation. The current study proves the feasibility that preoperative SPI values help with predicting IPPP immediately after VRS under AoA guidance and discrimination (between mild diagnoses and IPPP when based on postoperative SPI values) as they correlate with heart rate variations. Specifically, this applies when the countermeasures of IPPP and hemodynamic fluctuations are understood to be of importance in reducing unwelcome adverse events.

Changes in bispectral index and patient state index during sugammadex reversal of neuromuscular blockade under steady-state sevoflurane anesthesia

Few studies have investigated the changes in patient state index (PSI) and bispectral index (BIS) in response to abrupt increase in electromyographic (EMG) activity. These were performed using intravenous anesthetics or reversal agents for neuromuscular blockade (NMB) other than sugammadex. We compared the changes in BIS and PSI values caused by the sugammadex reversal of NMB during steady-state sevoflurane anesthesia. We enrolled 50 patients with American Society of Anesthesiologists physical status 1 and 2. At the end of the surgery, we administered 2 mg kg-1 sugammadex while maintaining sevoflurane for a 10-min study period. The changes in BIS and PSI from baseline (T0) to train of four ratio of 90% were not significantly different (median difference 0; 95% CI - 3 to 2; P = 0.83), neither were the changes in BIS and PSI values from T0 to their maximum values (median difference 1; 95% CI - 1 to 4; P = 0.53). Maximum BIS and PSI were significantly higher than their baseline values (median difference 6; 95% CI 4-9; P < 0.001 and median difference 5; 95% CI 3-6; P < 0.001, respectively). We found weak positive correlations between BIS and BIS-EMG (r = 0.12, P = 0.01), as well as PSI and PSI-EMG (r = 0.25, P < 0.001). Both PSI and BIS were affected to some extent by EMG artifacts after sugammadex administration.

Consistency of the anesthesia consciousness index versus the bispectral index during laparoscopic gastrointestinal surgery with sevoflurane anesthesia: A prospective multi-center randomized controlled clinical study

Background

This study aimed to compare the consistency of anesthesia consciousness index (Ai) with that of bispectral index (BIS) in monitoring the depth of anesthesia (DOA) during sevoflurane anesthesia, to reveal the optimal cutoff values in different states of consciousness, and explore the stability of DOA monitoring during intraoperative injurious stimulation.

Methods

We enrolled 145 patients (97 men and 48 women) from 10 medical centers. General anesthesia was induced using intravenous anesthetics and maintained with sevoflurane. Ai and BIS values were recorded.

Results

The mean difference between the Ai and BIS was-0.1747 (95% confidence interval, −0.6660 to 0.3166; p = 0.4857). The regression equation of Ai and BIS from the Deming regression analysis was y = 5.6387 + 0.9067x (y is BIS, x is Ai), and the slope and intercept were statistically significant. Meanwhile, the receiver operating characteristic curve analysis of anesthesia-induced unconsciousness, loss of consciousness, and recovery of consciousness revealed that the accuracy of Ai and BIS were similar. In addition, the optimal cutoff values of the different states of consciousness were not sensitive to age, and both Ai and BIS had no correlation with hemodynamics.

Conclusion

We conclude that Ai and BIS show no systematic deviation in readings with high consistency, similar accuracy, and good stability; these insights provide more data for clinical application.

Frontal electroencephalogram based drug, sex, and age independent sedation level prediction using non-linear machine learning algorithms

Brain monitors which track quantitative electroencephalogram (EEG) signatures to monitor sedation levels are drug and patient specific. There is a need for robust sedation level monitoring systems to accurately track sedation levels across all drug classes, sex and age groups. Forty-four quantitative features estimated from a pooled dataset of 204 EEG recordings from 66 healthy adult volunteers who received either propofol, dexmedetomidine, or sevoflurane (all with and without remifentanil) were used in a machine learning based automated system to estimate the depth of sedation. Model training and evaluation were performed using leave-one-out cross validation methodology. We trained four machine learning models to predict sedation levels and evaluated the influence of remifentanil, age, and sex on the prediction performance. The area under the receiver-operator characteristic curve (AUC) was used to assess the performance of the prediction model. The ensemble tree with bagging outperformed other machine learning models and predicted sedation levels with an AUC = 0.88 (0.81–0.90). There were significant differences in the prediction probability of the automated systems when trained and tested across different age groups and sex. The performance of the EEG based sedation level prediction system is drug, sex, and age specific. Nonlinear machine-learning models using quantitative EEG features can accurately predict sedation levels. The results obtained in this study may provide a useful reference for developing next generation EEG based sedation level prediction systems using advanced machine learning algorithms.

Clinical trial registration: NCT 02043938 and NCT 03143972.

Predicting Deep Hypnotic State From Sleep Brain Rhythms Using Deep Learning: A Data-Repurposing Approach

BACKGROUND:

Brain monitors tracking quantitative brain activities from electroencephalogram (EEG) to predict hypnotic levels have been proposed as a labor-saving alternative to behavioral assessments. Expensive clinical trials are required to validate any newly developed processed EEG monitor for every drug and combinations of drugs due to drug-specific EEG patterns. There is a need for an alternative, efficient, and economical method.

METHODS:

Using deep learning algorithms, we developed a novel data-repurposing framework to predict hypnotic levels from sleep brain rhythms. We used an online large sleep data set (5723 clinical EEGs) for training the deep learning algorithm and a clinical trial hypnotic data set (30 EEGs) for testing during dexmedetomidine infusion. Model performance was evaluated using accuracy and the area under the receiver operator characteristic curve (AUC).

RESULTS:

The deep learning model (a combination of a convolutional neural network and long short-term memory units) trained on sleep EEG predicted deep hypnotic level with an accuracy (95% confidence interval [CI]) = 81 (79.2–88.3)%, AUC (95% CI) = 0.89 (0.82–0.94) using dexmedetomidine as a prototype drug. We also demonstrate that EEG patterns during dexmedetomidine-induced deep hypnotic level are homologous to nonrapid eye movement stage 3 EEG sleep.

CONCLUSIONS:

We propose a novel method to develop hypnotic level monitors using large sleep EEG data, deep learning, and a data-repurposing approach, and for optimizing such a system for monitoring any given individual. We provide a novel data-repurposing framework to predict hypnosis levels using sleep EEG, eliminating the need for new clinical trials to develop hypnosis level monitors.

Impact of General Anesthesia Guided by State Entropy (SE) and Response Entropy (RE) on Perioperative Stability in Elective Laparoscopic Cholecystectomy Patients-A Prospective Observational Randomized Monocentric Study

Laparoscopic cholecystectomy is one of the most frequently performed interventions in general surgery departments. Some of the most important aims in achieving perioperative stability in these patients is diminishing the impact of general anesthesia on the hemodynamic stability and the optimization of anesthetic drug doses based on the individual clinical profile of each patient. The objective of this study is the evaluation of the impact, as monitored through entropy (both state entropy (SE) and response entropy (RE)), that the depth of anesthesia has on the hemodynamic stability, as well as the doses of volatile anesthetic. A prospective, observational, randomized, and monocentric study was carried out between January and December 2019 in the Clinic of Anesthesia and Intensive Care of the “Pius Brînzeu” Emergency County Hospital in Timișoara, Romania. The patients included in the study were divided in two study groups: patients in Group A (target group) received multimodal monitoring, which included monitoring of standard parameters and of entropy (SE and RE); while the patients in Group B (control group) only received standard monitoring. The anesthetic dose in group A was optimized to achieve a target entropy of 40–60. A total of 68 patients met the inclusion criteria and were allocated to one of the two study groups: group A (N = 43) or group B (N = 25). There were no statistically significant differences identified between the two groups for both demographical and clinical characteristics (p > 0.05). Statistically significant differences were identified for the number of hypotensive episodes (p = 0.011, 95% CI: [0.1851, 0.7042]) and for the number of episodes of bradycardia (p < 0.0001, 95% CI: [0.3296, 0.7923]). Moreover, there was a significant difference in the Sevoflurane consumption between the two study groups (p = 0.0498, 95% CI: [−0.3942, 0.9047]). The implementation of the multimodal monitoring protocol, including the standard parameters and the measurement of entropy for determining the depth of anesthesia (SE and RE) led to a considerable improvement in perioperative hemodynamic stability. Furthermore, optimizing the doses of anesthetic drugs based on the individual clinical profile of each patient led to a considerable decrease in drug consumption, as well as to a lower incidence of hemodynamic side-effects.

Comparative Evaluation of a New Depth of Anesthesia Index in ConView® System and the Bispectral Index during Total Intravenous Anesthesia: A Multicenter Clinical Trial

The performance of a new monitor for the depth of anesthesia (DOA), the Depth of Anesthesia Index (Ai) based on sample entropy (SampEn), 95% spectral edge frequency (95%SEF), and burst suppression ratio (BSR) was evaluated compared to Bispectral Index (BIS) during total intravenous anesthesia (TIVA). 144 patients in six medical centers were enrolled. General anesthesia was induced with stepwise-increased target-controlled infusion (TCI) of propofol until loss of consciousness (LOC). During surgery propofol was titrated according to BIS. Both Ai and BIS were recorded. Primary outcomes: the limits of agreement between Ai and BIS were -17.68 and 16.49, which were, respectively, -30.0% and 28.0% of the mean value of BIS. Secondary outcomes: prediction probability (Pk) of BIS and Ai was 0.943 and 0.935 (p=0.102) during LOC and 0.928 and 0.918 (p=0.037) during recovery of consciousness (ROC). And the values of BIS and Ai were 68.19 and 66.44 at 50%LOC, and 76.65 and 78.60 at 50%ROC. A decrease or an increase of Ai was significantly greater than that of BIS when consciousness changes (during LOC: -9.13±10.20 versus -5.83±9.63, p<0.001; during ROC: 10.88±11.51 versus 5.32±7.53, p<0.001). The conclusion is that Ai has similar characteristic of BIS as a DOA monitor and revealed the advantage of SampEn for indicating conscious level. This trial is registered at Chinese Clinical Trial Registry with ChiCTR-IOR-16009471.

Adaptive smith predictor controller for total intravenous anesthesia automation

Anesthetic agent propofol needs to be administered at an appropriate rate to prevent hypotension and postoperative adverse reactions. To comprehend more suitable anesthetic drug rate during surgery is a crucial aspect. The main objective of this proposal is to design robust automated control system that work efficiently in most of the patients with smooth BIS and minimum variations of propofol during surgery to avoid adverse post reactions and instability of anesthetic parameters. And also, to design advanced computer control system that improves the health of patient with short recovery time and less clinical expenditures. Unlike existing research work, this system administrates propofol as a hypnotic drug to regulate BIS, with fast bolus infusion in induction phase and slow continuous infusion in maintenance phase of anesthesia. The novelty of the paper lies in possibility to simplify the drug sensitivity-based adaption with infusion delay approach to achieve closed-loop control of hypnosis during surgery. Proposed work uses a brain concentration as a feedback signal in place of the BIS signal. Regression model based estimated sensitivity parameters are used for adaption to avoid BIS signal based frequent adaption procedure and large offset error. Adaptive smith predictor with lead-lag filter approach is applied on 22 different patients' model identified by actual clinical data. The actual BIS and propofol infusion signals recorded during clinical trials were used to estimate patient's sensitivity parameters EC ₅₀ and λ. Simulation results indicate that patient's drug sensitivity parameters based adaptive strategy facilitates optimal controller performance in most of the patients. Results are obtained with proposed scheme having less settling time, BIS oscillations and small offset error leads to adequate depth of anesthesia. A comparison with manual control mode and previously reported system shows that proposed system achieves reduction in the total variations of the propofol dose. Proposed adaptive scheme provides better performance with less oscillation in spite of computation delay, surgical stimulations and patient variability. Proposed scheme also provides improvement in robustness and may be suitable for clinical practices.

Assessing anesthetic activity through modulation of the membrane dipole potential

There is great individual variation in response to general anesthetics (GAs) leading to difficulties in optimal dosing and sometimes even accidental awareness during general anesthesia (AAGA). AAGA is a rare, but potentially devastating, complication affecting between 0.1% and 2% of patients undergoing surgery. The development of novel personalized screening techniques to accurately predict a patient’s response to GAs and the risk of AAGA remains an unmet clinical need. In the present study, we demonstrate the principle of using a fluorescent reporter of the membrane dipole potential, di-8-ANEPPs, as a novel method to monitor anesthetic activity using a well-described inducer/noninducer pair. The membrane dipole potential has previously been suggested to contribute a novel mechanism of anesthetic action. We show that the fluorescence ratio of di-8-ANEPPs changed in response to physiological concentrations of the anesthetic, 1-chloro-1,2,2-trifluorocyclobutane (F₃), but not the structurally similar noninducer, 1,2-dichlorohexafluorocyclobutane (F₆), to artificial membranes and in vitro retinal cell systems. Modulation of the membrane dipole provides an explanation to overcome the limitations associated with the alternative membrane-mediated mechanisms of GA action. Furthermore, by combining this technique with noninvasive retinal imaging technologies, we propose that this technique could provide a novel and noninvasive technique to monitor GA susceptibility and identify patients at risk of AAGA.