Anesthesia depth monitoring system based on electroencephalogram signals

By using an EEG-based anesthesia depth monitoring system, which combines high-precision sensors and dynamic network analysis with machine learning algorithms, the accuracy and real-time performance issues of anesthesia depth monitoring in existing technologies have been resolved. This enables precise monitoring of anesthesia status and automated decision support, thereby improving surgical safety and patient satisfaction.

CN118975778APending Publication Date: 2024-11-19XIAN HONGHUI HOSPITAL
View PDF 0 Cites 8 Cited by

Patent Information

Application Number
CN202411066510.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-05
Publication Date
2024-11-19

AI Technical Summary

Technical Problem

Existing anesthesia depth monitoring technologies suffer from insufficient data integration, limited monitoring accuracy, complex and expensive equipment, and a lack of automated analgesia monitoring, resulting in high complexity and cost of anesthesia management and insufficient accuracy and real-time monitoring.

Method used

An anesthesia depth monitoring system based on electroencephalogram (EEG) signals is adopted, including modules for data acquisition, signal processing, dynamic network analysis, and anesthesia depth prediction. It combines high-precision sensors, bandpass filters, dynamic network analysis, and machine learning algorithms to achieve real-time monitoring and prediction of anesthesia depth.

Benefits of technology

It enables precise and real-time monitoring of anesthesia depth, improves surgical safety and medical efficiency, reduces waste of medical resources, optimizes clinical operation response speed, and enhances postoperative patient satisfaction.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118975778A_ABST
    Figure CN118975778A_ABST
Patent Text Reader

Abstract

The invention relates to the field of biomedical signal processing, and discloses an electroencephalogram signal-based anesthesia depth monitoring system, which comprises a data acquisition module used for acquiring an electroencephalogram signal, a heartbeat signal and a respiration signal; the signal processing module is used for preprocessing the collected signals and extracting signal features; the dynamic network analysis module is used for constructing a dynamic brain function network and calculating network topology features and node features; the anesthesia depth prediction module is used for predicting the anesthesia depth in real time; and the user interface module is used for displaying the real-time monitoring information and the anesthesia depth prediction result. According to the anesthesia depth monitoring system based on the electroencephalogram signals, accurate anesthesia depth control is provided by monitoring and analyzing the anesthesia state in real time, the operation safety and medical efficiency are remarkably improved, waste of medical resources is reduced, the clinical operation response speed is optimized through automatic and data-driven decision support, and the operation efficiency is improved. Meanwhile, the operation experience and satisfaction degree of the patient are improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of biomedical signal processing, and in particular to an anesthesia depth monitoring system based on electroencephalogram signals. Background Art

[0002] In surgical operations under general anesthesia, anesthesiologists need to monitor the patient's anesthetic state in real time to ensure that the patient does not feel pain or regain consciousness during the operation. At the same time, it is crucial to avoid too deep or too shallow anesthesia, as this may lead to serious postoperative complications or psychological trauma. Currently, clinical technologies commonly used to monitor the depth of anesthesia include EEG bispectral index analysis, auditory evoked potentials, anesthesia entropy, etc. These technologies mainly monitor the depth of anesthesia by processing EEG signals.

[0003] Although existing technologies are effective to some extent, they still have some significant limitations. For example, bispectral index (BIS) monitoring has inconsistent responses to different anesthetic drugs and is sensitive to differences between individuals. In addition, the algorithms of these technologies are often proprietary and lack transparency, making it difficult for medical staff to understand how they work and verify the results.

[0004] Existing anesthesia depth monitoring equipment, such as BIS monitors and ECG monitors, are often configured independently, with poor data compatibility and complex connections, making it difficult to achieve multi-dimensional, multi-parameter automatic control. This not only increases the complexity of equipment management in the operating room, but also increases the equipment procurement and maintenance costs of the hospital. At the same time, due to the expensive equipment, the hospital is unable to equip all surgeries with high-quality anesthesia depth monitoring, which increases the treatment costs of patients.

[0005] In addition, existing anesthesia monitoring methods still lack objective indicators for analgesia monitoring. Anesthesiologists usually rely on conventional physiological parameters such as heart rate and mean arterial blood pressure combined with their own clinical experience to make judgments, lacking systematic and automated monitoring and analysis methods.

[0006] In summary, the existing technologies in anesthesia depth monitoring have problems such as insufficient data integration, limited monitoring accuracy, complex and expensive equipment, and lack of automated analgesia monitoring. Therefore, there is an urgent need for an integrated and intelligent anesthesia depth monitoring system that can provide more comprehensive, accurate and real-time monitoring to improve the quality and efficiency of clinical anesthesia management. Summary of the invention

[0007] In view of the deficiencies in the prior art, the present invention provides an anesthesia depth monitoring system based on EEG signals to achieve comprehensive, accurate and real-time monitoring of anesthesia depth.

[0008] To achieve the above objectives, the present invention is implemented through the following technical solutions: an anesthesia depth monitoring system based on EEG signals, comprising:

[0009] Data acquisition module, used to collect EEG signals, heartbeat signals and breathing signals;

[0010] The signal processing module is used to pre-process the collected signals, remove noise, filter, and extract signal features;

[0011] Dynamic network analysis module, used to construct dynamic brain functional network based on preprocessed signal features and calculate network topology features and node features;

[0012] Anesthesia depth prediction module, used for real-time anesthesia depth prediction based on dynamic network features and machine learning algorithms;

[0013] The user interface module is used to display real-time monitoring information and anesthesia depth prediction results.

[0014] Preferably, the data acquisition module includes:

[0015] High-precision EEG sensor, used to collect EEG signals;

[0016] ECG sensor, used to collect ECG signals;

[0017] Chest impedance sensor, used to collect breathing signals.

[0018] Preferably, the signal processing module includes:

[0019] A denoising processing unit, used for denoising the EEG signal, the ECG signal and the chest impedance signal using a bandpass filter;

[0020] The feature extraction unit is used to calculate the power spectrum density, heart rate variability characteristics and pulse conduction time characteristics of the EEG signal.

[0021] Preferably, the power spectrum density calculation formula of the EEG signal is:

[0022] P(f)=10log 10 (∫|X(f,t)| 2 df)

[0023] Among them, X(f,t) represents the short-time Fourier transform result of the EEG signal, f represents frequency, t represents time, and the integral in the formula is the integral of all frequencies, which is used to calculate the power spectral density of the entire frequency band.

[0024] Preferably, the dynamic network analysis module includes:

[0025] A time window division unit is used to divide the preprocessed signal features into 30-second sliding windows;

[0026] A node definition unit, used to take the EEG signal, heart rate variability and pulse conduction time characteristics in each time window as network nodes;

[0027] The edge weight calculation unit is used to calculate the correlation between nodes and construct the edge weight matrix. The formula is:

[0028] w ij (t) = αw ij (t-1)+(1-α)ρ ij (t)

[0029] Among them, w ij (t) is the weight between nodes i and j at time t, ρ ij (t) is the correlation coefficient between nodes i and j at time t, and α is a smoothing factor used to balance historical and current weight information.

[0030] Preferably, the dynamic network analysis module further includes:

[0031] Spectral clustering unit, used for spectral clustering and community structure analysis, the formula is:

[0032]

[0033] Among them, Q represents modularity, which is used to measure the quality of the community structure in the network, and A ij is the actual connection weight between nodes i and j, k i and k j are the degrees of nodes i and j respectively, m is the sum of all connection weights in the network, δ(c i ,c j ) is an indicator function, which is 1 when nodes i and j are in the same community and 0 otherwise;

[0034] The topology stability analysis unit is used to analyze the stability of the network topology structure. The formula is:

[0035] Stability(t)=λ 2 (t)-λ 2 (t-1)

[0036] Among them, λ 2 (t) is the second smallest eigenvalue of the graph Laplacian matrix at time t, representing the algebraic connectivity of the graph, and the difference λ 2 (t)-λ 2 (t-1) represents the change in the stability of the network topology.

[0037] Preferably, the anesthesia depth prediction module comprises:

[0038] A feature integration unit for combining dynamic network features and historical anesthesia status;

[0039] The training and prediction unit is used to perform model training and real-time update using the random forest algorithm and output an anesthesia depth index.

[0040] Preferably, the calculation formula of the anesthesia depth index is:

[0041]

[0042] Among them, w i is the feature weight, f i (t) is the feature value at time t, and N is the number of features. The anesthesia depth index is calculated by weightedly combining the influence of each feature to provide a quantitative index to evaluate the depth of anesthesia.

[0043] Preferably, the user interface module includes:

[0044] Real-time data display unit, used to dynamically display EEG signals, heart rate variability, pulse conduction time characteristics and their trend graphs;

[0045] Anesthesia depth index display unit, used to display the current anesthesia depth and its changing trend;

[0046] Alarm system to trigger alarm and provide advice when the depth of anaesthesia is too shallow or too deep.

[0047] The present invention also provides a method for monitoring the depth of anesthesia based on EEG signals, comprising the following steps:

[0048] Collect EEG signals, heartbeat signals and breathing signals through EEG sensors, ECG sensors and chest impedance sensors;

[0049] De-noising and feature extraction are performed on the original signal to obtain the power spectrum characteristics, heart rate variability characteristics and pulse conduction time characteristics of the EEG signal;

[0050] Based on the preprocessed signal features, a dynamic brain function network is constructed, and node features and edge weights are calculated;

[0051] The dynamic network features are input into the machine learning model for training and real-time update, and the anesthesia depth index is output;

[0052] The signal characteristics, predicted anesthesia depth and its changing trend are dynamically displayed through the user interface, and an alarm is triggered when the anesthesia depth is too shallow or too deep.

[0053] The present invention provides an anesthesia depth monitoring system based on EEG signals. It has the following beneficial effects:

[0054] 1. The EEG-based anesthesia depth monitoring system of the present invention provides precise anesthesia depth control by real-time monitoring and analysis of anesthesia status, significantly improves surgical safety and medical efficiency, reduces waste of medical resources, and optimizes clinical operation response speed through automated and data-driven decision support, while improving patients' surgical experience and satisfaction.

[0055] 2. The present invention can provide a more accurate assessment of the anesthetic state by using dynamic network analysis and real-time anesthesia depth prediction. The use of dynamic networks allows real-time monitoring and analysis of subtle changes in EEG signals and other physiological signals, so that the depth of anesthesia can be accurately determined. This method is more adaptable to individual differences and immediate changes during surgery than traditional anesthesia monitoring methods, effectively avoiding the risk of too shallow or too deep anesthesia and increasing surgical safety.

[0056] 3. The anesthesia depth monitoring system of the present invention can provide real-time feedback on the anesthesia status, support rapid decision-making and adjust the administration of anesthetic drugs. The system's real-time data display and alarm function can immediately alert medical personnel to any dangerous changes in anesthesia levels, allowing doctors to respond quickly and adjust the anesthesia dosage, which is particularly critical for rapid response in emergency situations.

[0057] 4. The present invention uses advanced data processing and machine learning technologies to learn and predict from a large amount of clinical data and provide data-driven decision support. This method not only improves the scientificity and accuracy of anesthesia monitoring, but also makes medical decisions more objective and reliable. In addition, by continuously collecting and analyzing data, the system can continuously optimize its prediction model and further improve the quality and effectiveness of anesthesia management.

[0058] 5. The precise anesthesia depth control of the present invention can greatly improve the patient's surgical experience and reduce postoperative recovery time and discomfort. By continuously monitoring and accurately regulating the anesthesia state, it can ensure that the patient is in the best anesthesia state throughout the operation, which not only reduces the risks associated with intraoperative anesthesia, but also improves the patient's overall satisfaction with the operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0059] Figure 1 It is a schematic diagram of the system architecture of the present invention;

[0060] Figure 2 It is a schematic diagram of the method flow of the present invention. DETAILED DESCRIPTION

[0061] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0062] Please see attached Figure 1 , an embodiment of the present invention provides an anesthesia depth monitoring system based on EEG signals, comprising:

[0063] Data acquisition module

[0064] The data acquisition module is the basic part of the present invention, and its main task is to accurately and effectively obtain physiological signals from the patient, including EEG signals, ECG signals and respiratory signals. The accurate acquisition of these signals is the prerequisite for efficient signal processing, dynamic network analysis and anesthesia depth prediction. To ensure the high quality and synchronization of the signal, the present invention adopts high-precision sensors, and through the design of optimized signal acquisition paths and synchronization mechanisms, the accuracy and real-time performance of data acquisition are guaranteed.

[0065] In this embodiment, the data acquisition module includes three sensors: an electroencephalogram sensor, an electrocardiogram sensor, and a chest impedance sensor. Each sensor is specially designed and configured for its specific physiological signal to optimize the signal acquisition effect.

[0066] In this embodiment, the EEG sensor is used to collect EEG signals, which is a key physiological parameter for evaluating the depth of anesthesia in patients. EEG sensors usually use non-invasive electrodes, which are placed on the patient's scalp and arranged according to the standard positions of the international 10-20 system. To improve the signal quality, the contact impedance between the electrode and the skin is strictly controlled below the recommended technical standards. In addition, the EEG sensor in this system is equipped with a highly sensitive amplifier that can amplify weak EEG signals while filtering out external electromagnetic interference to ensure the purity and accuracy of the collected EEG signals.

[0067] For the acquisition of ECG signals, the ECG sensor used in this embodiment can accurately record the electrical signals generated by cardiac activity. ECG sensors usually include multiple electrode patches that are placed on the patient's chest and limbs to capture the complete spectrum of ECG activity. The design of the ECG sensor ensures high signal-to-noise ratio and low drift performance, making the acquisition of ECG signals stable and reliable.

[0068] The chest impedance sensor is used to collect respiratory signals and monitor respiratory activity by measuring the impedance change of the chest. In this embodiment, the chest impedance sensor includes a pair of electrodes, which are placed on the opposite sides of the patient's chest. As breathing proceeds, the impedance change caused by the volume change of the chest is captured by the electrodes and converted into an electrical signal. The chest impedance sensor is specially designed to minimize false signal interference caused by patient movement and improve the accuracy and stability of respiratory monitoring.

[0069] Through the coordinated work of these three sensors, the data acquisition module can provide comprehensive and accurate physiological parameter input, providing a solid data foundation for the entire anesthesia monitoring system. The high quality of these data ensures that the system can effectively evaluate and predict the depth of anesthesia in the subsequent signal processing and analysis stages, ensuring the timeliness and accuracy of medical intervention, thereby greatly improving patient safety and the success rate of surgery.

[0070] Signal processing module

[0071] The signal processing module is the core part of the present invention. Its function is to denoise and extract features from the original EEG signals, ECG signals and chest impedance signals obtained from the data acquisition module, so as to facilitate the subsequent dynamic network analysis and anesthesia depth prediction. The design and implementation of this module is the key to ensure the accuracy and efficiency of signal analysis, which directly affects the performance and reliability of the monitoring system.

[0072] In this embodiment, the signal processing module includes two main units: a denoising processing unit and a feature extraction unit. The processing task of each unit should not only ensure the accuracy of the data, but also improve the efficiency of data processing.

[0073] In this embodiment, the denoising processing unit uses a bandpass filter to denoise the EEG signal, the ECG signal and the chest impedance signal. The bandpass filter is specially designed to retain the key frequency band of the physiological signal while effectively weakening or eliminating the noise outside the band. For the EEG signal, the bandpass filter frequency range is usually set to 0.5Hz to 50Hz to ensure that the key EEG rhythm is retained while filtering out low-frequency drift and high-frequency electrical noise. The filter settings of the ECG signal and the chest impedance signal are also adjusted accordingly according to their characteristics to ensure the clarity and availability of the signal.

[0074] The feature extraction unit is responsible for calculating the power spectrum density, heart rate variability characteristics and pulse conduction time characteristics of the EEG signal. Power spectrum density is an important indicator for evaluating the state of EEG activity. The calculation formula is as follows:

[0075] P(f)=10log 10 (∫|X(f,t)| 2 df)

[0076] Among them, X(f,t) represents the short-time Fourier transform result of the EEG signal, f represents frequency, t represents time, and the integral in the formula is the integral of all frequencies, which is used to calculate the power spectrum density of the entire frequency band. The implementation of this formula can describe the frequency domain characteristics of EEG activity in detail and help determine the patient's anesthesia depth.

[0077] The heart rate variability feature reflects the fluctuation of heart rate and is very useful for monitoring the patient's physiological stress state. The pulse transit time feature can provide important information about the cardiovascular state and is an important parameter for evaluating cardiovascular function.

[0078] Through these signal processing technologies, the signal processing module can provide accurate and effective physiological parameters, providing the necessary input for dynamic network analysis and anesthesia depth prediction. This not only enhances the practicality of the monitoring system, but also improves the safety and accuracy of anesthesia monitoring.

[0079] Dynamic Network Analysis Module

[0080] The dynamic network analysis module is one of the core components of the present invention, which aims to provide real-time anesthesia depth monitoring and assessment by dynamically constructing and analyzing brain functional networks. This module combines steps such as time window division, node definition, edge weight calculation, spectral clustering, and topological stability analysis to ensure comprehensive and accurate monitoring of changes in EEG activity.

[0081] In this embodiment, the dynamic network analysis module includes several key units: a time window division unit, a node definition unit, an edge weight calculation unit, a spectral clustering unit, and a topological stability analysis unit. These units work together to form a complete analysis process to ensure the accuracy and practicality of dynamic network analysis.

[0082] Time window division unit

[0083] The time window division unit is the first step of dynamic network analysis. In this embodiment, the unit is responsible for dividing the preprocessed signal features according to a 30-second sliding window. Each time window contains data on EEG signals, heart rate variability, and pulse conduction time features, which can capture physiological changes in a short period of time and improve the time resolution and accuracy of the analysis.

[0084] In this embodiment, the time window division method is as follows:

[0085] 1. Get the preprocessed signal data stream.

[0086] 2. The sliding window is divided into 30-second units, with 50% overlap between windows to ensure data continuity and transition smoothness.

[0087] 3. Package the data in each window for subsequent node definition and edge weight calculation.

[0088] The node definition unit is responsible for defining the signal features in each time window as network nodes. Specifically, in this embodiment, the features of the EEG signal (such as power spectrum density), heart rate variability features, and pulse conduction time features are respectively regarded as nodes in the network. In this way, the node set in each time window represents the state of each physiological signal feature in the time period.

[0089] For example:

[0090] The node features of EEG signals include: power spectral density, expressed as P(f).

[0091] The node features of heart rate variability include: standard deviation of heart beat intervals, frequency domain analysis results, etc.

[0092] The node features of pulse conduction time include: changes in pulse arrival time, etc.

[0093] Edge weight calculation unit

[0094] The edge weight calculation unit is used to calculate the correlation between nodes, thereby constructing an edge weight matrix. This matrix is ​​the basis for subsequent network analysis and reflects the degree of mutual influence between different nodes. The specific calculation method is as follows:

[0095] w ij (t) = αw ij (t-1)+(1-α)ρ ij (t)

[0096] Among them, w ij (t) is the weight between nodes i and j at time t, ρ ij (t) is the correlation coefficient between nodes i and j at time t, and α is a smoothing factor used to balance historical and current weight information.

[0097] In this embodiment, the specific steps are as follows:

[0098] 1. Calculate the correlation coefficient ρ between each pair of nodes in the current time window ij (t), using the Pearson correlation coefficient:

[0099]

[0100] Among them, Cov(X i ,X j ) is the signal covariance between nodes i and j, σ i and σ j are the signal standard deviations of nodes i and j respectively.

[0101] 2. Update the edge weight matrix w ij (t).

[0102] 3. Input the edge weight matrix into the subsequent spectral clustering and topological analysis unit.

[0103] Spectral Clustering Unit

[0104] The spectral clustering unit is used to perform spectral clustering and community structure analysis and identify the community structure in the network. The specific formula is as follows:

[0105]

[0106] Among them, Q represents modularity, which is used to measure the quality of the community structure in the network, and A ij is the actual connection weight between nodes i and j, k i and k j are the degrees of nodes i and j respectively, m is the sum of all connection weights in the network, δ(c i ,c j ) is an indicator function that is 1 when nodes i and j are in the same community and 0 otherwise.

[0107] In this embodiment, the specific steps are as follows:

[0108] 1. Construct the edge weight matrix A.

[0109] 2. Calculate the degree k of each node i .

[0110] 3. Calculate the modularity Q of the network and perform spectral clustering analysis.

[0111] 4. Determine the community structure of the network and analyze its changes.

[0112] Topological stability analysis unit

[0113] The topology stability analysis unit is used to analyze the stability of the network topology structure, focusing on the changes in the network in different time windows. The specific formula is as follows:

[0114] Stabilty(t)=λ 2 (t)-λ 2 (t-1)

[0115] Among them, λ 2 (t) is the second smallest eigenvalue of the graph Laplacian matrix at time t, representing the algebraic connectivity of the graph, and the difference λ 2 (t)-λ 2 (t-1) represents the change in the stability of the network topology.

[0116] In this embodiment, the specific steps are as follows:

[0117] 1. Calculate the graph Laplacian matrix L within the current time window.

[0118] 2. Solve the eigenvalue of L and determine the second smallest eigenvalue λ 2 (t).

[0119] 3. Compare the λ of the current time window with that of the previous time window 2 Difference, evaluates the topological stability of the network.

[0120] 4. Use the stability analysis results to assess changes in anesthetic depth.

[0121] Through the above, the dynamic network analysis module can comprehensively and accurately reflect the dynamic changes of EEG activities during anesthesia, thus providing a scientific basis and support for real-time anesthesia depth monitoring.

[0122] Anesthesia Depth Prediction Module

[0123] The anesthesia depth prediction module is a key component of the present invention for achieving real-time anesthesia depth monitoring and evaluation. This module provides accurate anesthesia depth information to anesthesiologists by integrating the data generated by the dynamic network analysis module and historical anesthesia data, using advanced machine learning algorithms. The design of this module allows for continuous updating and optimization of the prediction model to adapt to real-time changes in the patient's physiological parameters and improve the accuracy and reliability of the prediction.

[0124] In this embodiment, the anesthesia depth prediction module includes a feature integration unit and a training and prediction unit.

[0125] Feature Integration Unit

[0126] The feature integration unit is responsible for integrating the network features and historical anesthesia status data from the dynamic network analysis module. This unit combines the node and edge features (such as the power spectrum density of the node, the weight of the edge, etc.) collected from each time window with the patient's previous anesthesia depth records to form a comprehensive feature set, providing a data basis for subsequent anesthesia depth prediction.

[0127] In this embodiment, the specific operation of the feature integration unit is as follows:

[0128] 1. Receive network features output by the dynamic network analysis module: including node features (such as power spectrum density) and edge correlation metrics (such as correlation coefficients between nodes).

[0129] 2. Extract the patient's historical anesthesia data: extract data including past anesthesia depth index, corresponding physiological signal characteristics (such as heart rate variability, pulse conduction time characteristics) from the database.

[0130] 3. Form a comprehensive feature matrix: Combine real-time dynamic network features and historical data to form a feature matrix. Each row represents a complete feature set at a time point, which will be input into the training and prediction unit for anesthesia depth prediction.

[0131] In this way, the feature integration unit ensures that the prediction model can utilize the most comprehensive and up-to-date data, thereby improving the accuracy of the prediction.

[0132] Training and prediction unit

[0133] The training and prediction unit uses the random forest algorithm to train and update the model in real time. Random forest is an integrated learning method that improves the accuracy and robustness of prediction by building multiple decision trees and voting. The algorithm is particularly suitable for processing problems with high-dimensional feature spaces, such as the anesthesia depth prediction in the present invention.

[0134] In this embodiment, the specific operations of the training and prediction unit include:

[0135] 1. Model training:

[0136] Data preprocessing: Preprocess historical data, including standardization, missing value processing, etc.

[0137] Model construction: Use historical data and the corresponding anesthesia depth index to train a random forest model. The model training process includes selecting the optimal number of trees, branching conditions, and other hyperparameters.

[0138] Cross-validation: Use cross-validation to evaluate the performance of the model and adjust the model parameters to optimize the prediction effect.

[0139] 2. Real-time updates:

[0140] Data stream processing: As new data (i.e., real-time dynamic network features and anesthesia depth records) arrive, the model will be appropriately adjusted to reflect the latest data trends.

[0141] Model Update: Incrementally train or retrain the model based on new data to keep the model current and accurate.

[0142] 3. Anesthesia depth prediction:

[0143] Feature input: Use the current feature set to predict the depth of anesthesia. Feature input includes real-time network features and historical data.

[0144] Output prediction results: The model outputs the anesthesia depth index and outputs the prediction results to the anesthesia management system in real time.

[0145] Anesthesia depth index calculation

[0146] The anesthesia depth index is the core indicator output by the anesthesia depth prediction module. Its calculation formula is as follows:

[0147]

[0148] Among them, w i is the feature weight, f i (t) is the feature value at time t, and N is the number of features. The anesthesia depth index is calculated by weightedly combining the influence of each feature to provide a quantitative index to evaluate the depth of anesthesia.

[0149] Specifically:

[0150] Feature weight w i : Determined based on statistical analysis of the relationship between features and depth of anesthesia in historical data. Weights can be dynamically adjusted based on the importance of the features to optimize the accuracy of the prediction.

[0151] Eigenvalue f i : Including power spectral density, heart rate variability characteristics, pulse conduction time characteristics, etc.

[0152] By calculating the anesthesia depth index, the system can provide a comprehensive, quantitative indicator to help anesthesiologists evaluate the patient's anesthesia depth in real time, thereby more accurately adjusting the dosage of anesthetic drugs and improving patient safety and treatment outcomes.

[0153] Through the collaborative work of the feature integration unit and the training and prediction unit, the patient's anesthesia depth can be monitored and predicted in real time. This system can not only improve the accuracy and real-time performance of the prediction, but also provide anesthesiologists with comprehensive evaluation indicators, significantly improving the effect of anesthesia management.

[0154] User interface module

[0155] The user interface module plays a vital role in the present invention. It not only provides an intuitive and easy-to-operate interface for anesthesiologists to monitor the anesthesia process, but also enhances medical safety through visual data presentation and real-time alarm system. This module is the main interface for interacting with users in the entire anesthesia depth monitoring system. It helps medical personnel make quick and accurate decisions by efficiently displaying key monitoring data and prediction results.

[0156] In this embodiment, the user interface module includes a real-time data display unit, an anesthesia depth index display unit and an alarm system.

[0157] Real-time data display unit

[0158] The unit is responsible for dynamically displaying key physiological parameters, including EEG signals, heart rate variability, pulse conduction time characteristics and their historical and real-time trend graphs.

[0159] Each signal is displayed in the form of a graph, such as a time domain waveform graph, a frequency domain analysis graph, etc., so that medical personnel can observe the patient's physiological state and its changes during the anesthesia process at a glance.

[0160] Anesthesia Depth Index Display Unit

[0161] This unit is specially designed to display the current anesthesia depth index and its trend graph over time.

[0162] The anesthesia depth index is expressed in numbers and graphs. The numbers intuitively display the current anesthesia depth, while the trend graph depicts the changes in the index over time during the operation, helping doctors understand the overall fluctuations in anesthesia depth.

[0163] Alarm system

[0164] The alarm system is automatically triggered when it detects that the depth of anesthesia is too shallow or too deep, issuing visual and audible warnings to immediately attract the attention of the anesthesiologist.

[0165] In addition, the system also provides relevant suggestions or preset operation steps, such as recommending adjustments to drug dosages or other clinical interventions to ensure patient safety.

[0166] Through real-time data display, anesthesiologists can instantly obtain the patient's latest physiological status and quickly respond to any physiological changes that may occur.

[0167] The embodiment of the present invention provides an anesthesia depth monitoring system that comprehensively utilizes EEG signals, heartbeat signals, and respiratory signals, and realizes real-time and accurate prediction of anesthesia depth by constructing and analyzing dynamic brain function networks. The system helps to improve the safety and efficiency of anesthesia, reduce the risk of too shallow or too deep anesthesia, and enhance the patient's surgical experience and postoperative recovery quality.

[0168] Please refer to the attached Figure 2 The present invention also provides a method for monitoring the depth of anesthesia based on EEG signals, comprising the following steps:

[0169] S1, collect EEG signals, heartbeat signals and breathing signals through EEG sensors, ECG sensors and chest impedance sensors;

[0170] S2, denoising and feature extraction of the original signal to obtain the power spectrum characteristics, heart rate variability characteristics and pulse conduction time characteristics of the EEG signal;

[0171] S3, construct a dynamic brain function network based on the preprocessed signal features, and calculate the node features and edge weights;

[0172] S4, input the dynamic network features into the machine learning model for training and real-time update, and output the anesthesia depth index;

[0173] S5. Dynamically display signal characteristics, anesthesia depth prediction values ​​and their changing trends through the user interface, and trigger an alarm when the anesthesia depth is too shallow or too deep.

[0174] The method of this embodiment is executed based on the above-mentioned system embodiment, and its principles and technical effects are similar, so they will not be repeated here.

[0175] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. An anesthesia depth monitoring system based on EEG signals, characterized in that: include: Data acquisition module, used to collect EEG signals, heartbeat signals and breathing signals; The signal processing module is used to pre-process the collected signals, remove noise, filter, and extract signal features; Dynamic network analysis module, used to construct dynamic brain functional network based on preprocessed signal features and calculate network topology features and node features; Anesthesia depth prediction module, used for real-time anesthesia depth prediction based on dynamic network features and machine learning algorithms; The user interface module is used to display real-time monitoring information and anesthesia depth prediction results.

2. The anesthesia depth monitoring system based on EEG signals according to claim 1, characterized in that: The data acquisition module comprises: EEG sensor, used to collect EEG signals; ECG sensor, used to collect ECG signals; Chest impedance sensor, used to collect breathing signals.

3. The anesthesia depth monitoring system based on EEG signals according to claim 1, characterized in that: The signal processing module comprises: A denoising processing unit, used for denoising the EEG signal, the ECG signal and the chest impedance signal using a bandpass filter; The feature extraction unit is used to calculate the power spectrum density, heart rate variability characteristics and pulse conduction time characteristics of the EEG signal.

4. The anesthesia depth monitoring system based on EEG signals according to claim 3, characterized in that: The power spectrum density calculation formula of the EEG signal is: P(f)=10log 10 (∫|X(f,t)| 2 df) Among them, X(f,t) represents the short-time Fourier transform result of the EEG signal, f represents frequency, t represents time, and the integral in the formula is the integral of all frequencies, which is used to calculate the power spectral density of the entire frequency band.

5. The anesthesia depth monitoring system based on EEG signals according to claim 1, characterized in that: The dynamic network analysis module includes: A time window division unit is used to divide the preprocessed signal features into 30-second sliding windows; A node definition unit, used to take the EEG signal, heart rate variability and pulse conduction time characteristics in each time window as network nodes; The edge weight calculation unit is used to calculate the correlation between nodes and construct the edge weight matrix. The formula is: w ij (t)=αw ij (t-1)+(1-a)p ij (t) Among them, w ij (t) is the weight between nodes i and j at time t, ρ ij (t) is the correlation coefficient between nodes i and j at time t, and α is a smoothing factor used to balance historical and current weight information.

6. The anesthesia depth monitoring system based on EEG signals according to claim 5, characterized in that: The dynamic network analysis module also includes: Spectral clustering unit, used for spectral clustering and community structure analysis, the formula is: Among them, Q represents modularity, which is used to measure the quality of the community structure in the network, and A ij is the actual connection weight between nodes i and j, k i and k j are the degrees of nodes i and j respectively, m is the sum of all connection weights in the network, δ(c i ,c j ) is an indicator function, which is 1 when nodes i and j are in the same community and 0 otherwise; The topology stability analysis unit is used to analyze the stability of the network topology structure. The formula is: Stability(t)=λ2(t)-λ2(t-1) Among them, λ2(t) is the second smallest eigenvalue of the graph Laplacian matrix at time t, representing the algebraic connectivity of the graph, and the difference λ2(t)-λ2(t-1) represents the change in the stability of the network topology.

7. The anesthesia depth monitoring system based on EEG signals according to claim 1, characterized in that: The anesthesia depth prediction module comprises: A feature integration unit for combining dynamic network features and historical anesthesia status; The training and prediction unit is used to perform model training and real-time update using the random forest algorithm and output an anesthesia depth index.

8. The anesthesia depth monitoring system based on EEG signals according to claim 7, characterized in that: The calculation formula of the anesthesia depth index is: Among them, w i is the feature weight, f i (t) is the feature value at time t, and N is the number of features. The anesthesia depth index is calculated by weightedly combining the influence of each feature to provide a quantitative index to evaluate the depth of anesthesia.

9. The anesthesia depth monitoring system based on EEG signals according to claim 1, characterized in that: The user interface module comprises: Real-time data display unit, used to dynamically display EEG signals, heart rate variability, pulse conduction time characteristics and their trend graphs; Anesthesia depth index display unit, used to display the current anesthesia depth and its changing trend; Alarm system to trigger alarm and provide advice when the depth of anaesthesia is too shallow or too deep.

10. A method for monitoring the depth of anesthesia based on electroencephalogram signals, based on the system according to any one of claims 1 to 9, characterized in that: The following steps are involved: Collect EEG signals, heartbeat signals and breathing signals through EEG sensors, ECG sensors and chest impedance sensors; De-noising and feature extraction are performed on the original signal to obtain the power spectrum characteristics, heart rate variability characteristics and pulse conduction time characteristics of the EEG signal; Based on the preprocessed signal features, a dynamic brain function network is constructed, and node features and edge weights are calculated; The dynamic network features are input into the machine learning model for training and real-time update, and the anesthesia depth index is output; The signal characteristics, predicted anesthesia depth and its changing trend are dynamically displayed through the user interface, and an alarm is triggered when the anesthesia depth is too shallow or too deep.

Citation Information

Cited By

  • Clinical data management method and system for anesthesiology department

    CN119170175A

  • Acupuncture treatment effect real-time evaluation system based on brain-computer interface technology

    CN119339892A

  • Intelligent anesthesia depth monitoring method and system

    CN119924777A

  • Intelligent anesthesia depth monitoring method and system

    CN119924777B

  • Wearable anesthesia depth monitoring equipment and monitoring method thereof

    CN119969959A