Adaptive drug interaction model to predict depth of anesthesia in the operating room

A Deep Learning Framework for Anesthesia Depth Prediction from Drug Infusion History

In the target-controlled infusion (TCI) of propofol and remifentanil intravenous anesthesia, accurate prediction of the depth of anesthesia (DOA) is very challenging. Patients with different physiological characteristics have inconsistent pharmacodynamic responses during different stages of anesthesia. For example, in TCI, older adults transition smoothly from the induction period to the maintenance period, while younger adults are more prone to anesthetic awareness, resulting in different DOA data distributions among patients. To address these problems, a deep learning framework that incorporates domain adaptation and knowledge distillation and uses propofol and remifentanil doses at historical moments to continuously predict the bispectral index (BIS) is proposed in this paper. Specifically, a modified adaptive recurrent neural network (AdaRNN) is adopted to address data distribution differences among patients. Moreover, a knowledge distillation pipeline is developed to train the prediction network by enabling it to learn intermediate feature representations of the teacher network. The experimental results show that our method exhibits better performance than existing approaches during all anesthetic phases in the TCI of propofol and remifentanil intravenous anesthesia. In particular, our method outperforms some state-of-the-art methods in terms of root mean square error and mean absolute error by 1 and 0.8, respectively, in the internal dataset as well as in the publicly available dataset.

A Transformer-Based Prediction Method for Depth of Anesthesia During Target-Controlled Infusion of Propofol and Remifentanil

Accurately predicting anesthetic effects is essential for target-controlled infusion systems. The traditional (PK-PD) models for Bispectral index (BIS) prediction require manual selection of model parameters, which can be challenging in clinical settings. Recently proposed deep learning methods can only capture general trends and may not predict abrupt changes in BIS. To address these issues, we propose a transformer-based method for predicting the depth of anesthesia (DOA) using drug infusions of propofol and remifentanil. Our method employs long short-term memory (LSTM) and gate residual network (GRN) networks to improve the efficiency of feature fusion and applies an attention mechanism to discover the interactions between the drugs. We also use label distribution smoothing and reweighting losses to address data imbalance. Experimental results show that our proposed method outperforms traditional PK-PD models and previous deep learning methods, effectively predicting anesthetic depth under sudden and deep anesthesia conditions.

AnesthesiaGUIDE: a MATLAB tool to control the anesthesia

The goals of this paper are: (a) to investigate adaptive and fractional-order adaptive control algorithms for an automatic anesthesia process, using a closed-loop system, and (b) to develop an easy-to-use tool for MATLAB/Simulink to facilitate simulations for users with less knowledge about anesthesia and adaptive control. A model reference adaptive control structure was chosen for the entire system. First of all, to control the patient’s state during the surgery process, the patient mathematical model is useful, or even required for simulation studies. The pharmacokinetic/pharmacodynamics (PK/PD) model was determined using MATLAB’s SimBiology tool, starting from a previously available block diagram, and validated through simulation. Then, to achieve the desired control performances, two controllers are designed: a PI adaptive controller and a PIλ (PI-fractional) adaptive controller, using the MIT algorithm. The time response during anesthetic drug infusion for each patient can be plotted with the AnesthesiaGUIDE tool, which is also designed in MATLAB/Simulink. The tool was tested on data from 12 patients, subjected to general anesthesia, with successful results. Through this tool, the article provides a good opportunity for any user to experience with adaptive control for the anesthesia process.

Performance of blink reflex in patients during anesthesia induction with propofol and remifentanil: prediction probabilities and multinomial logistic analysis

Background

The amount of propofol needed to induce loss of responsiveness varied widely among patients, and they usually required less than the initial dose recommended by the drug package inserts. Identifying precisely the moment of loss of responsiveness will determine the amount of propofol each patient needs. Currently, methods to decide the exact moment of loss of responsiveness are based on subjective analysis, and the monitors that use objective methods fail in precision. Based on previous studies, we believe that the blink reflex can be useful to characterize, more objectively, the transition from responsiveness to unresponsiveness. The purpose of this study is to investigate the relation between the electrically evoked blink reflex and the level of sedation/anesthesia measured with an adapted version of the Richmond Agitation–Sedation Scale, during the induction phase of general anesthesia with propofol and remifentanil. Adding the blink reflex to other variables may allow a more objective assessment of the exact moment of loss of responsiveness and a more personalized approach to anesthesia induction.

Results

The electromyographic-derived features proved to be good predictors to estimate the different levels of sedation/anesthesia. The results of the multinomial analysis showed a reasonable performance of the model, explaining almost 70% of the adapted Richmond Agitation–Sedation Scale variance. The overall predictive accuracy for the model was 73.6%, suggesting that it is useful to predict loss of responsiveness.

Conclusions

Our developed model was based on the information of the electromyographic-derived features from the blink reflex responses. It was able to predict the drug effect in patients undergoing general anesthesia, which can be helpful for the anesthesiologists to reduce the overwhelming variability observed between patients and avoid many cases of overdosing and associated risks. Despite this, future research is needed to account for variabilities in the clinical response of the patients and with the interactions between propofol and remifentanil. Nevertheless, a method that could allow for an automatic prediction/detection of loss of responsiveness is a step forward for personalized medicine.