Patient anesthesia state analysis method and system based on physical sign data

Abstract

The invention provides a patient anesthesia state analysis method and system based on physical sign data, and relates to the technical field of anesthesia state monitoring and analysis. The system comprises an anesthesia state monitoring module, an electroencephalogram sign correlation analysis module, an anesthesia state local anomaly detection module and an anesthesia state diffusion detection module. According to the method, the correlation between the electroencephalogram state and the physical sign data is utilized, the anesthesia state of the patient is comprehensively analyzed by combining the electroencephalogram state and the physical sign parameters, static analysis of the local anesthesia stage and cross-stage dynamic diffusion detection are combined, a more refined and intelligent anesthesia state monitoring scheme is provided, and the monitoring accuracy of the anesthesia state is improved. The accuracy of anesthesia abnormity detection is improved, comprehensive monitoring and real-time early warning of the anesthesia state can be achieved, and the safety of a patient in the anesthesia process is ensured.

Description

Technical Field

[0001] The present invention relates to the technical field of anesthesia status monitoring and analysis, and in particular to a method and system for analyzing the anesthesia status of a patient based on vital sign data. Background Art

[0002] Anesthesia state monitoring is a crucial part of anesthesia management, which aims to ensure that the patient is at an appropriate depth of anesthesia during surgery to ensure patient safety and optimize the anesthesia effect. Traditional anesthesia state assessment relies more on EEG signals, such as monitoring the patient's anesthesia depth through EEG indicators such as brain impedance index. This scheme that focuses on the changes in EEG signals for anesthesia state analysis, due to the rapid effect of anesthetic drugs during the anesthesia induction stage, the patient's brain waves will undergo more drastic changes. At this time, the changes in EEG signals are more sensitive and drastic, so anesthesia depth assessment and abnormality detection can be performed more accurately. However, with the stabilization of the concentration of anesthetic drugs, the depth of anesthesia reaches a relatively stable state, the changes in EEG waves tend to be stable, and the amplitude of changes is small, resulting in the difficulty of detecting potential local abnormalities in the anesthesia maintenance stage when relying solely on EEG signals for monitoring. At this time, the depth of anesthesia is no longer as drastic as in the induction stage, and possible anesthesia state abnormalities are difficult to detect in advance through changes in EEG signals.

[0003] During the anesthesia maintenance stage, the patient's anesthesia depth is relatively stable, the effect of drug concentration on neural activity tends to be stable, and the correlation between the patient's physiological state and vital sign data is more significant, which can provide real-time information about the effects of anesthetic drugs on organ systems. Therefore, anesthesia state monitoring based only on EEG signals may not fully reflect the comprehensive effects of anesthetic drugs on various organ systems in the body, and it is easy to ignore potential anesthesia abnormalities or physiological fluctuations, resulting in delayed discovery of abnormal anesthesia states in patients. Summary of the invention

[0004] In response to the above technical problems, the present invention provides a method and system for analyzing the anesthesia status of patients based on vital sign data, which combines multi-dimensional analysis of EEG signals and vital sign data. Through correlation analysis of real-time EEG signals and vital sign data, it can more accurately monitor changes in the anesthesia status of patients during the anesthesia maintenance stage, promptly discover potential anesthesia abnormalities, and improve the accuracy of real-time analysis of the anesthesia status of patients.

[0005] As one aspect of the present invention, a method for analyzing a patient's anesthesia state based on vital sign data is provided, comprising: After obtaining the real-time anesthesia state monitoring data of the patient to be analyzed in the current detection stage, the real-time anesthesia state monitoring data includes the real-time EEG state monitoring data and real-time vital sign monitoring data of the patient to be analyzed, extracting the real-time EEG state parameter sequence and multiple real-time vital sign state parameter sequences of the patient to be analyzed; The conditional entropy between the real-time EEG state parameter sequence and multiple real-time physical sign state parameter sequences is calculated respectively to construct a real-time EEG physical sign association vector of the patient to be analyzed at the current detection stage; Determine the first local anesthesia stage of the patient to be analyzed according to the real-time EEG state monitoring data in the current detection stage, and determine the EEG sign target association vector and the EEG sign association reference range corresponding to the real-time EEG sign association vector; Performing local abnormality detection of the anesthesia state of the patient to be analyzed according to the EEG sign target association vector and the EEG sign association reference range, and obtaining the local abnormality detection result of the anesthesia state; If the result of the local abnormality detection of the anesthesia state is that the patient to be analyzed does not have a local abnormality of the anesthesia state, then performing an anesthesia state diffusion detection on the patient to be analyzed to generate an anesthesia state diffusion abnormality detection result; The anesthetic state diffusion detection of the patient to be analyzed includes obtaining real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage, determining the second local anesthesia stage of the patient to be analyzed according to the real-time anesthetic state monitoring data of the next detection stage, generating a target anesthetic state diffusion path of the patient to be analyzed, determining the EEG sign target diffusion vector of the target anesthetic state diffusion path, and performing anesthetic state diffusion detection on the real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage based on the EEG sign target diffusion vector.

[0006] Preferably, the EEG sign target association vector and the EEG sign association reference range further include: Acquire historical anesthesia status monitoring data of multiple patients in the anesthesia maintenance stage, the historical anesthesia status monitoring data including the historical EEG status monitoring data and historical vital sign monitoring data of the patients in the anesthesia maintenance stage; The anesthesia maintenance stage is divided into multiple local anesthesia stages according to the historical EEG state monitoring data, multiple groups of historical sample data of each local anesthesia stage are extracted from multiple groups of historical anesthesia state monitoring data, and a stage sample data set of each local anesthesia stage is constructed; Extract the historical EEG state parameter sequence and multiple historical sign state parameter sequences of each group of historical sample data in each stage sample data set, calculate the conditional entropy between the historical EEG state parameter sequence and multiple historical sign state parameter sequences in each group of historical sample data, and construct the historical EEG sign association vector of each group of historical sample data; Constructing an EEG sign association matrix of each stage sample data set according to multiple historical EEG sign association vectors in the stage sample data set, performing covariance calculation on each EEG sign association matrix to generate a covariance matrix of each EEG sign association matrix, and performing eigenvalue decomposition on multiple covariance matrices to obtain multiple reference eigenvectors and eigenvalues ​​of each reference eigenvector; Determine multiple target feature vectors from multiple reference feature vectors, calculate the association score of each sign state parameter according to the eigenvalues ​​corresponding to the multiple target feature vectors, generate the EEG sign association reference vector for each local anesthesia stage, and determine the EEG sign target association vector and EEG sign association reference range for each local anesthesia stage according to the EEG sign association reference vector.

[0007] Preferably, determining the EEG sign target association vector and the EEG sign association reference range for each local anesthesia stage according to the EEG sign association reference vector includes: Based on the EEG sign association reference vector, multiple historical EEG sign association vectors in the stage sample data set of each local anesthesia stage are associated and corrected to obtain a historical EEG sign correction association vector corresponding to each historical EEG sign association vector; A target EEG sign association vector for each local anesthesia stage is generated based on multiple historical EEG sign correction association vectors, and the distances between the EEG sign target association vector and multiple historical EEG sign correction association vectors are calculated respectively. The EEG sign association reference range is determined based on the EEG sign target association vector, and a target cumulative analysis is performed on the EEG sign association reference range, including determining the EEG sign association reference range with the EEG sign target association vector as the center, and calculating the target cumulative ratio after each expansion of the EEG sign association reference range based on the target reference distance, completing the expansion of the EEG sign association reference range after reaching the target cumulative threshold, and outputting the EEG sign association reference range for each local anesthesia stage.

[0008] Preferably, for the EEG sign target diffusion vector, it also includes: Determine the anesthesia stage diffusion path of each local anesthesia stage, and construct a diffusion sample data set containing multiple groups of diffusion sample data for each anesthesia stage diffusion path; The first EEG sign correction association vector and the second EEG sign correction association vector of each group of diffusion sample data in each diffusion sample data set are extracted, and the association state diffusion trend vector of each group of diffusion sample data is generated according to the first EEG sign correction association vector and the second EEG sign correction association vector, and the EEG sign target diffusion vector of the diffusion path of each anesthesia stage is generated according to multiple association state diffusion trend vectors.

[0009] Preferably, the local abnormality detection of the anesthetic state of the patient to be analyzed is performed according to the EEG sign target association vector and the EEG sign association reference range to obtain the local abnormality detection result of the anesthetic state, including: According to the EEG sign association reference vector corresponding to the first local anesthesia stage of the patient to be analyzed, the real-time EEG sign association vector of the patient to be analyzed in the current detection stage is corrected to generate a real-time EEG sign association correction vector, and the target distance between the real-time EEG sign association correction vector and the EEG sign target association vector is calculated. If the target distance is within the EEG sign association reference range, it is considered that the patient to be analyzed has no local abnormality in the anesthetic state; otherwise, it is considered that the patient to be analyzed has a local abnormality in the anesthetic state and an abnormal anesthetic state warning is issued.

[0010] Preferably, performing anesthetic state diffusion detection on the real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage based on the EEG sign target diffusion vector includes: The real-time EEG sign association vector corresponding to the real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage is extracted; the real-time EEG sign diffusion vector of the patient to be analyzed is calculated based on the real-time EEG sign association vector of the patient to be analyzed in the current detection stage; the real-time EEG sign diffusion vector is analyzed according to the EEG sign target diffusion vector, the diffusion deviation state of each sign state parameter is determined, and the anesthetic state diffusion abnormality detection result of the patient to be analyzed is generated.

[0011] Preferably, the correlation score of each physical sign state parameter is calculated based on the eigenvalues ​​corresponding to the multiple target eigenvectors, including: The element value of each sign state parameter in the target feature vector is analyzed by the eigenvalue corresponding to the target feature vector to obtain the local score of each sign state parameter in each target feature vector. The local scores of the sign state parameters in multiple target feature vectors are accumulated to obtain the association score of each sign state parameter.

[0012] As another aspect of the present invention, a patient anesthesia state analysis system based on vital sign data is also provided, which is used to implement the above-mentioned patient anesthesia state analysis method based on vital sign data, comprising: The anesthesia state monitoring module is used to extract the real-time EEG state parameter sequence and multiple real-time vital sign state parameter sequences of the patient to be analyzed after acquiring the real-time anesthesia state monitoring data of the patient to be analyzed in the current detection stage, the real-time anesthesia state monitoring data including the real-time EEG state monitoring data and real-time vital sign monitoring data of the patient to be analyzed; The EEG sign correlation analysis module is used to calculate the conditional entropy between the real-time EEG state parameter sequence and multiple real-time sign state parameter sequences, so as to construct a real-time EEG sign correlation vector of the patient to be analyzed at the current detection stage; The module for detecting local abnormality of the anesthesia state is used to determine the first local anesthesia stage of the patient to be analyzed according to the real-time EEG state monitoring data in the current detection stage, determine the EEG sign target association vector and the EEG sign association reference range corresponding to the real-time EEG sign association vector, and perform local abnormality detection of the anesthesia state on the patient to be analyzed according to the EEG sign target association vector and the EEG sign association reference range to obtain the local abnormality detection result of the anesthesia state; an anesthetic state diffusion detection module, for performing an anesthetic state diffusion detection on the patient to be analyzed to generate an anesthetic state diffusion abnormality detection result if the anesthetic state local abnormality detection result indicates that the patient to be analyzed does not have a local abnormality in the anesthetic state; The anesthetic state diffusion detection of the patient to be analyzed includes obtaining real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage, determining the second local anesthesia stage of the patient to be analyzed according to the real-time anesthetic state monitoring data of the next detection stage, generating a target anesthetic state diffusion path of the patient to be analyzed, determining the EEG sign target diffusion vector of the target anesthetic state diffusion path, and performing anesthetic state diffusion detection on the real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage based on the EEG sign target diffusion vector.

[0013] The present invention has the following beneficial effects: The present invention combines EEG state and vital sign parameters to comprehensively analyze the patient's anesthetic state, and is particularly suitable for the anesthesia maintenance stage. By dividing the anesthesia maintenance stage into multiple local anesthesia stages, the patient's real-time anesthesia state monitoring data is analyzed in detail, and the deep-level correlation rules between EEG state and vital sign data are excavated, thereby achieving high-precision detection of the anesthetic state and timely discovering potential abnormalities. Moreover, on the basis of the analysis of the local anesthesia stage, anesthesia state diffusion detection is further introduced, and cross-stage data comparison and dynamic analysis are performed in adjacent local anesthesia stages. By monitoring the diffusion path of the anesthetic state, potential and gradually accumulated anesthesia state abnormalities can be effectively captured. By utilizing the correlation between EEG state and vital sign data, the static analysis of the local anesthesia stage and the dynamic diffusion detection across stages are combined, providing a more refined and intelligent anesthesia state monitoring solution, improving the accuracy of anesthesia abnormality detection, and being able to achieve comprehensive monitoring and real-time early warning of the anesthetic state, thereby ensuring the safety of patients during anesthesia. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] Figure 1 The figure is a flow chart of a method for analyzing the anesthesia status of a patient based on vital sign data in one of the implementation processes of the present invention.

[0015] Figure 2 The present invention is a schematic structural diagram of a patient anesthesia status analysis system based on vital sign data in one of the implementation processes of the present invention. DETAILED DESCRIPTION

[0016] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0017] As one of the implementation processes of the present invention, a method for analyzing the anesthesia state of a patient based on vital sign data is provided, such as Figure 1 As shown, the method specifically comprises the following steps: Step S1: after obtaining the real-time anesthesia status monitoring data of the patient to be analyzed in the current detection stage, extracting the real-time EEG state parameter sequence and multiple real-time vital sign state parameter sequences of the patient to be analyzed.

[0018] It should be noted that the real-time anesthesia status monitoring data of the patient to be analyzed in the current detection stage are usually obtained from various anesthesia monitoring devices, including but not limited to EEG monitors, electrocardiographs, blood pressure monitoring equipment, respiratory monitors, etc.

[0019] Real-time anesthesia state monitoring data includes two important parts: real-time EEG state monitoring data and real-time vital sign monitoring data. Real-time EEG state monitoring data refers to the EEG signal of the patient obtained through electroencephalogram (EEG) or other EEG monitoring equipment, mainly including parameters such as the frequency and amplitude of brain waves, reflecting the patient's anesthesia depth, consciousness state and other information. Multiple EEG state parameters can be extracted by in-depth analysis of the EEG signal, such as the BIS value obtained by analyzing the EEG signal as a quantitative parameter to characterize the EEG state, thereby constructing a real-time EEG state parameter sequence containing multiple EEG state parameters of the patient to be analyzed. Real-time vital sign state monitoring data refers to the physiological sign data of the patient obtained through a variety of vital sign monitoring devices, such as heart rate, blood pressure, respiratory rate, body temperature and other vital sign monitoring equipment. Multiple vital sign parameters corresponding to the data collected by each type of vital sign monitoring equipment can be extracted to construct multiple real-time vital sign state parameter sequences of the patient to be analyzed. These parameter sequences reflect the impact of anesthetic drugs on the patient's physiological system.

[0020] It is worth noting that simply analyzing the abnormality of the anesthesia state based on a variety of thresholds set based on empirical knowledge may not be able to detect potential abnormalities in the data in a timely manner. Therefore, the real-time collected data can be comprehensively analyzed in stages to obtain specific anesthesia state analysis results at each monitoring stage. For example, when the patient is in the anesthesia maintenance stage, the patient's brain electrical activity changes will be relatively small because the anesthesia state tends to be stable. In this case, a test can be performed every 5 minutes or every 10 minutes to ensure that abnormalities in the anesthesia process can be captured in time and handled accordingly.

[0021] Step S2: respectively calculating the conditional entropy between the real-time EEG state parameter sequence and multiple real-time vital sign state parameter sequences to construct a real-time EEG vital sign association vector of the patient to be analyzed at the current detection stage.

[0022] It should be noted that after obtaining the real-time EEG state parameter sequence and the sign state parameter sequence, the correlation between the EEG signal and the sign data is quantified by analyzing multiple sequences. In this embodiment, the quantification of the correlation is achieved by calculating the conditional entropy between the real-time EEG state parameter sequence and multiple real-time sign state parameter sequences. Conditional entropy describes the uncertainty of other information under the condition that certain information is known. In this embodiment, the EEG signal is used as known information, and the correlation between the EEG and multiple sign parameters is analyzed respectively to measure the degree of change of the sign state under the condition that the EEG state is known. A lower conditional entropy indicates that there is a strong correlation or dependency between the two. Through the calculated conditional entropy, the EEG sign association vector of the patient to be analyzed in the current detection stage is constructed, which includes the quantitative correlation between the EEG state and each sign parameter, which can provide data reference for subsequent abnormal anesthesia state detection.

[0023] Step S3, determining the first local anesthesia stage of the patient to be analyzed according to the real-time EEG state monitoring data in the current detection stage, and determining the EEG sign target association vector and the EEG sign association reference range corresponding to the real-time EEG sign association vector.

[0024] It should be noted that for the first local anesthesia stage, it is mainly determined based on the EEG state corresponding to the real-time EEG state monitoring data in the current detection stage. For example, the reference range of the EEG state in the anesthesia maintenance stage is determined based on the BIS value. If the BIS value range is between 20-60, it is regarded as the anesthesia maintenance stage, and the EEG state of this stage can be divided into multiple local anesthesia stages. For example, the BIS difference of 2 is used as a reference for stage division. Then, according to the local anesthesia stage corresponding to the real-time EEG state monitoring data in the current detection stage, the first local anesthesia stage of the patient to be analyzed is finally obtained. The reference data corresponding to the first local anesthesia stage is determined by the EEG sign target association vector and the EEG sign association reference range corresponding to each anesthesia stage obtained in advance based on historical data analysis. A benchmark can be provided based on the EEG sign target association vector and the EEG sign association reference range to perform abnormal analysis on the current anesthesia state of the patient to be analyzed.

[0025] Step S4: performing local abnormality detection of the anesthetic state on the patient to be analyzed according to the EEG sign target association vector and the EEG sign association reference range to obtain a local abnormality detection result of the anesthetic state.

[0026] It should be noted that the purpose of local abnormality detection of the anesthetic state of the patient to be analyzed is to determine whether there is an abnormal change in the anesthetic state of the patient to be analyzed in the current detection stage. In this process, the real-time EEG sign association vector of the patient to be analyzed in the current detection stage is corrected according to the EEG sign association reference vector corresponding to the first local anesthesia stage of the patient to be analyzed to generate a real-time EEG sign association correction vector. Among them, the EEG sign association reference vector corresponding to the first local anesthesia stage describes the relationship between the EEG state and the sign parameters of the patient to be analyzed in the current stage, including the degree of influence and correlation of different sign parameters on the EEG state. The correction process can specifically adopt a weighted method to adjust the current state of the patient to be analyzed, strengthen some key feature correlations and weaken some sign parameters with relatively lower reference value.

[0027] For example, at different stages of local anesthesia, the correlation between changes in EEG state and changes in patient blood pressure may be different. If a unified standard is used for analysis, it is likely that a certain strength of correlation is normal at a certain stage, but as the anesthesia state changes, the same correlation strength means that there may be abnormalities, resulting in some potential abnormalities not being discovered in time. In addition, there may be noise in the actual data collection process, and correction can provide a more stable state analysis.

[0028] Then, by calculating the target distance between the real-time EEG sign associated correction vector and the EEG sign target associated vector, abnormality identification is performed based on the target distance and the corresponding EEG sign associated reference range. The calculation of the target distance can be determined using a distance measurement method such as Euclidean distance. Compare the calculated target distance with the EEG sign associated reference range. If the target distance is within the normal range, it means that the current anesthesia state is within a reasonable fluctuation range and there is a local abnormality in the anesthesia state; if the target distance exceeds the normal reference range, it means that the patient is likely to have a local abnormality in the anesthesia state. In this case, an anesthesia state warning can be performed by triggering an early warning, which is convenient for alerting relevant medical staff to conduct a more detailed analysis and prepare in advance for possible abnormal anesthesia states.

[0029] Step S5: If the result of the local abnormality detection of the anesthesia state is that the patient to be analyzed does not have a local abnormality of the anesthesia state, then an anesthesia state diffusion detection is performed on the patient to be analyzed to generate an anesthesia state diffusion abnormality detection result.

[0030] It should be noted that the above-mentioned anesthesia state recognition is specifically carried out through staged analysis and detection to explore the correlation differences between the real-time EEG state and the physical sign state of the patient under different anesthesia states, thereby improving the accuracy of the analysis of the patient's anesthesia state. Considering that there may be deficiencies based on stage analysis alone, after further detecting that the patient is deemed to have no local abnormality in the anesthesia state during the stage analysis, the patient's anesthesia state diffusion is detected, specifically through the data of the next detection stage, the states of the two adjacent stages are comprehensively analyzed, and further abnormality analysis is performed from the perspective of the evolution of the anesthesia state.

[0031] In this process, the anesthetic state diffusion detection is performed on the patient to be analyzed. Specifically, after obtaining the real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage, the second local anesthesia stage of the patient to be analyzed is determined according to the real-time anesthetic state monitoring data of the next detection stage, that is, the local anesthesia stage corresponding to the patient is judged according to the corresponding BIS value in the data. According to the stage change relationship, the target anesthetic state diffusion path of the patient to be analyzed is determined. For example, if the local anesthesia stage corresponding to the patient to be analyzed changes from the BIS value of 50-52 to 52-54, the corresponding diffusion path is determined. The historical data is analyzed in advance to determine the associated changes in the state of different local anesthesia stages during the diffusion process, so as to determine the EEG sign target diffusion vector of each anesthetic state diffusion path. After querying the target anesthetic state diffusion path corresponding to the patient to be analyzed in two adjacent detection stages, the anesthetic state diffusion detection is performed on the patient to be analyzed according to the EEG sign target diffusion vector of the target anesthetic state diffusion path.

[0032] Specifically, the real-time EEG sign association vector corresponding to the real-time anesthesia status monitoring data of the patient to be analyzed in the next detection stage is extracted, and then the real-time EEG sign diffusion vector of the patient to be analyzed is calculated based on the real-time EEG sign association vector of the patient to be analyzed in the current detection stage.

[0033] Among them, the real-time EEG sign association vectors in the current detection stage and the next detection stage respectively characterize the association strength between the EEG state and the sign state. By performing difference analysis on the real-time EEG sign association vector in the current detection stage through the real-time EEG sign association vector in the next detection stage, the quantitative value of the synergistic change trend of the association strength corresponding to each sign state parameter in the diffusion process can be extracted to construct the real-time EEG sign diffusion vector of the patient to be analyzed. The real-time EEG sign diffusion vector is analyzed according to the EEG sign target diffusion vector, and the diffusion deviation state of each sign state parameter can be determined. Specifically, the EEG sign target diffusion vector reflects the quantitative value of the synergistic change trend of each sign state parameter under ideal conditions. If the change amplitude of a sign state parameter is greater than the synergistic change trend under ideal conditions, it is considered that the diffusion deviation state is abnormal, otherwise it is considered to be a normal diffusion state. In this way, the diffusion detection of the anesthetic state of the patient to be analyzed is realized, and the abnormal diffusion detection result of the anesthetic state of the patient to be analyzed is generated.

[0034] The significance of anesthetic status diffusion detection lies in that even if the patient appears normal in the current detection stage, that is, the anesthetic status is temporarily correct at a certain static angle, but the accumulation of minor abnormalities may lead to deeper abnormalities during the dynamic change process. In this way, effective early warning of potential abnormalities in the anesthesia process is achieved, and the monitoring accuracy of possible anesthetic status abnormalities of patients during the anesthesia maintenance stage is improved.

[0035] It is worth adding that the evaluation of the patient's anesthetic state based on EEG signals, especially EEG indicators such as the EEG impedance index that are widely used to monitor the depth of anesthesia, will cause drastic changes in the patient's brain waves due to the rapid effect of anesthetic drugs during the anesthesia induction stage. At this time, the EEG signal is more sensitive and accurate in evaluating the depth of anesthesia. During the induction stage, the EEG signal can accurately evaluate the depth of anesthesia and detect abnormalities. However, as the concentration of anesthetic drugs stabilizes, when the patient is in the anesthesia maintenance stage, the changes in brain waves tend to be stable and the amplitude of changes is small, resulting in difficulty in timely detection of local abnormalities that may occur during the anesthesia maintenance stage when only relying on EEG signals for anesthesia state analysis, and failure to fully capture potential fluctuations and abnormalities in the patient's anesthesia state, which may lead to delays in the identification and intervention of abnormal anesthesia states.

[0036] The embodiment of the present invention provides a method for analyzing the anesthesia state of a patient based on vital sign data, which comprehensively analyzes the anesthesia state of a patient in combination with EEG state and vital sign parameters, and is particularly suitable for the anesthesia maintenance stage. The anesthesia maintenance stage is divided into multiple smaller local stages, and the real-time data of each local stage is analyzed in detail, and the deep-level correlation between EEG state and vital sign data is mined, thereby achieving high-precision detection of the anesthesia state and timely discovering potential abnormalities. And on the basis of the analysis of the local anesthesia stage, anesthesia state diffusion detection is further introduced. Considering the dynamic evolution of the anesthesia state, that is, considering that although no abnormality is found in a certain local stage, the anesthesia state may change slightly over time, it is necessary to compare and dynamically analyze data across stages in adjacent stages. By monitoring the diffusion path of the anesthesia state, potential and gradually accumulated anesthesia state abnormalities can be effectively captured. The present invention utilizes the correlation between EEG state and vital sign data, combines the static analysis of the local anesthesia stage with the dynamic diffusion detection across stages, and provides a more refined and intelligent anesthesia state monitoring solution. It overcomes the limitation of anesthesia state analysis based only on EEG signals, improves the accuracy of anesthesia abnormality detection, and can achieve comprehensive monitoring and real-time early warning of the anesthesia state, ensuring the safety of patients during anesthesia.

[0037] As an optional implementation process, in order to achieve higher precision of anesthesia state monitoring, the anesthesia state of the patient in the anesthesia maintenance stage is modeled by analyzing historical data, so as to provide a more reliable reference for abnormal detection of anesthesia state. For the EEG sign target association vector and EEG sign association reference range in step S3, the specific generation process includes: The historical anesthesia status monitoring data of multiple patients in the anesthesia maintenance stage are obtained, and the anesthesia maintenance stage is divided into multiple local anesthesia stages according to the historical EEG state monitoring data. Multiple groups of historical sample data of each local anesthesia stage are extracted from the multiple groups of historical anesthesia status monitoring data to construct a stage sample data set for each local anesthesia stage.

[0038] It should be noted that the historical anesthesia status monitoring data obtained for multiple patients during the anesthesia maintenance stage mainly include the historical EEG state monitoring data and historical vital sign monitoring data of the patients during the anesthesia maintenance stage, so as to reflect the actual specific EEG state and vital sign state changes of the patients used as sample data during the anesthesia maintenance stage. For the division of local anesthesia stages, the method mentioned in the above content can be specifically adopted. Through the multiple local anesthesia stages obtained by division, the historical anesthesia status monitoring data of each patient can be divided according to the corresponding stage to obtain multiple groups of historical sample data corresponding to each local anesthesia stage.

[0039] In this process, it is taken into account that different patients have different durations of anesthesia maintenance stage due to differences in physiological data and surgical types during actual operations, resulting in different durations of each local anesthesia stage for different patients. In this case, the data of each patient in the same local anesthesia stage can be further segmented based on the same duration. For example, some patients' anesthesia maintenance stage may be as long as seven or eight hours, or as short as two or three hours. In this case, the data in each local anesthesia stage can be divided into multiple groups of data at intervals of, for example, 10 minutes to achieve a unified data scale, thereby constructing a stage sample data set containing multiple groups of historical sample data for each local anesthesia stage.

[0040] The historical EEG state parameter sequence and multiple historical sign state parameter sequences of each group of historical sample data in the sample data set of each stage are extracted, and the conditional entropy between the historical EEG state parameter sequence and multiple historical sign state parameter sequences in each group of historical sample data is calculated respectively, and the historical EEG sign association vector of each group of historical sample data is constructed.

[0041] It should be noted that after quantifying the EEG state and physical sign state in each group of historical sample data respectively and extracting the historical EEG state parameter sequence and multiple historical physical sign state parameter sequences of each group of historical sample data, in order to quantify the correlation between the EEG state and the physical sign state, the conditional entropy between the historical EEG state parameter sequence and the historical physical sign state parameter sequence in each group of historical sample data is calculated respectively, so as to measure the degree of change and correlation strength of the physical sign data under the condition of known EEG state, so as to quantify the dependence between the EEG and physical sign states, and finally construct a historical EEG sign association vector that characterizes the dependent change relationship between the EEG state and different physical sign state parameters in each group of historical sample data.

[0042] It is worth noting that although the historical EEG sign association vectors extracted between each group of historical sample data may come from the same patient, and there are differences in the corresponding patient's physiological parameters and in vivo metabolic conditions, but in this local anesthesia stage, there is still a commonality in the change relationship of the data in the same stage. Individual differences are more likely to cause differences in the duration of the patient's anesthesia state. For example, some patients have a faster metabolism, which leads to faster changes in the EEG state, that is, there are differences in the rate of change of the EEG state, but in this change process, the law of changes in the EEG state causing the sign parameters to change is common. For example, as the BIS value gradually increases or decreases, the patient's physical signs such as blood pressure and heart rate will also change accordingly. This commonality makes it possible to model and analyze the change rules in the process of analyzing the anesthesia status monitoring data of different patients, so as to remove the interference of individual differences in the data and identify the common changes therein.

[0043] The EEG sign association matrix of each stage sample data set is constructed according to multiple historical EEG sign association vectors in the stage sample data set, and the covariance calculation is performed on each EEG sign association matrix to generate the covariance matrix of each EEG sign association matrix. The multiple covariance matrices are subjected to eigenvalue decomposition to obtain multiple reference eigenvectors and the eigenvalues ​​of each reference eigenvector.

[0044] It should be noted that the historical EEG sign association vectors of multiple groups of historical sample data in the collection are spliced ​​to form a high-dimensional EEG sign association matrix, and the eigendecomposition method is used to extract multiple main characteristic directions in the high-dimensional matrix. In this process, the covariance of the EEG sign association matrix is ​​first calculated to obtain the corresponding covariance matrix, so as to better capture the change law and internal structure of the matrix data. Then, the covariance matrix corresponding to each EEG sign association matrix is ​​decomposed by eigenvalue to obtain multiple reference eigenvectors of each covariance matrix and the eigenvalues ​​of each reference eigenvector. The reference eigenvector represents the main characteristic direction in the matrix, and the corresponding eigenvalue reflects the importance of these features. The reference eigenvector with a larger eigenvalue indicates that it has a greater impact on the analysis of the anesthetic state, while the eigenvector with a smaller eigenvalue may have less impact on the change of the anesthetic state.

[0045] Determine multiple target feature vectors from multiple reference feature vectors, calculate the association score of each sign state parameter according to the eigenvalues ​​corresponding to the multiple target feature vectors, generate the EEG sign association reference vector for each local anesthesia stage, and determine the EEG sign target association vector and EEG sign association reference range for each local anesthesia stage according to the EEG sign association reference vector.

[0046] It should be noted that, for the selection of multiple target feature vectors, the selection can be specifically made according to the eigenvalues ​​corresponding to the reference feature vectors. For example, by accumulating the eigenvalues, multiple reference feature vectors are sorted from large to small in order of eigenvalues, and they are selected one by one and the sum of the eigenvalues ​​corresponding to the selected reference feature vectors is calculated. After reaching a preset threshold, the screening is completed and the multiple selected reference feature vectors are recorded as target feature vectors, or the selection is made directly based on the preset feature threshold, and the reference feature vectors whose eigenvalues ​​are greater than the preset feature threshold are recorded as target feature vectors.

[0047] For the selected multiple target feature vectors, the numerical value corresponding to each element in the vector represents the weight of one of the physical sign state parameters, and the eigenvalue corresponding to the target feature vector represents the importance or contribution of the target feature vector. After normalizing the multiple eigenvalues, the normalized importance parameter of each target feature vector can be obtained. The weight of each physical sign state parameter in each target feature vector is then further strengthened by the importance parameter, for example, the product of the two is calculated to obtain the local score of each physical sign state parameter in each target feature vector, and the local scores of the physical sign state parameters in multiple target feature vectors are accumulated to obtain the association score of each physical sign state parameter. Finally, the EEG sign association reference vector of the local anesthesia stage is constructed according to the association score of the sign state parameters. The EEG sign association reference vector represents the typical association change pattern between the EEG state and the sign state in the set, and can provide the common association rules between the EEG state and the sign state in different local anesthesia stages during the anesthesia maintenance stage. Then, through the common association rules between the EEG state and the sign state, a comprehensive analysis is performed on multiple historical EEG sign association vectors corresponding to each local anesthesia stage to generate the EEG sign target association vector and EEG sign association reference range of the local anesthesia stage.

[0048] As an optional implementation process, the process of determining the EEG sign target association vector and the EEG sign association reference range for each local anesthesia stage according to the EEG sign association reference vector specifically includes the following contents: Based on the EEG sign association reference vector, association correction is performed on multiple historical EEG sign association vectors in the stage sample data set of each local anesthesia stage to obtain a historical EEG sign correction association vector corresponding to each historical EEG sign association vector, and an EEG sign target association vector for each local anesthesia stage is generated based on the multiple historical EEG sign correction association vectors.

[0049] It should be noted that the significance of correcting the historical EEG sign association vector in each local anesthesia stage is that the EEG state and sign state involved in each historical EEG sign association vector may have different association relationships due to fluctuations caused by individual differences in patients. The EEG sign association reference vector is based on the common laws of a large amount of data. The association correction can make each historical EEG sign association vector more representative to adapt to individual differences. The correction process adopts a weighted approach, and the specific value of each sign state parameter of the EEG sign association reference vector is used as the weight of the correction, so as to obtain the historical EEG sign correction association vector corresponding to each historical EEG sign association vector. Then, multiple corrected vectors are aggregated, for example, the mean of multiple historical EEG sign correction association vectors is calculated to obtain an EEG sign target association vector that comprehensively represents the group association pattern of the EEG state and sign state in the local anesthesia stage.

[0050] It is worth adding that the above-mentioned EEG sign target association vector is specifically generated based on the EEG sign association reference vector by considering the individual personalized characteristics, that is, considering the physiological differences of different patients, and the EEG sign association reference vector is generated by eigenvalue decomposition. The generation process of the EEG sign association reference vector extracts the main characteristic direction and common laws of the data, but removes the influence of individual differences. In the actual anesthesia state abnormality detection process, due to the more significant personalized physiological characteristics between different patients, only considering the general law may not fully reflect the specific situation of individual patients, resulting in false positives for some individuals during the detection process. Taking the EEG sign target association vector as a reference can take into account the particularity of individual patients, and has strong applicability and flexibility for the monitoring and abnormality detection of individual patients. It can analyze the anesthesia state from the perspective of more in line with the physiological characteristics of individual patients and adapting to the anesthesia change laws of different individuals, making the anesthesia state analysis process more personalized.

[0051] For the EEG sign association reference range, after generating the EEG sign target association vector, the distances between the EEG sign target association vector and multiple historical EEG sign corrected association vectors are calculated respectively, the EEG sign association reference range is determined according to the EEG sign target association vector, and the EEG sign association reference range is subjected to target cumulative analysis to obtain the EEG sign association reference range for each local anesthesia stage.

[0052] It should be noted that the EEG sign target association vector is specifically used as a benchmark to measure the changing correlation characteristics between the EEG state and the sign state of each patient during the local anesthesia stage. In this process, the EEG sign association reference range is further determined based on multiple reference individuals, that is, to measure the difference between the patient's general state and the baseline state during the actual anesthesia state analysis process. The reference range is an interval that includes normal anesthesia state fluctuations and is used to define the boundary between normal and abnormal.

[0053] In this process, the EEG sign association reference range is first determined with the EEG sign target association vector as the center. The initial EEG sign association reference range only has the origin, and then the EEG sign association reference range is gradually expanded and the target cumulative analysis is performed. Each time the range is expanded based on the target reference distance. After each range expansion operation, the ratio of the cumulative number of targets within the range to the total number of patients is counted to obtain the target cumulative ratio. If the target cumulative ratio reaches the target cumulative threshold after a certain expansion, such as 85%, that is, 85% of the historical data is included in the expanded reference range, in this case, it is considered that the reference range has fully covered the data fluctuation of the normal anesthesia state, and the expansion of the EEG sign association reference range is completed, and the EEG sign association reference range of the local anesthesia stage is output. Otherwise, the expansion operation continues until the threshold is reached and output. The final determined EEG sign association reference range will be used as the normal anesthesia state fluctuation range of the local anesthesia stage. Any real-time EEG sign association vector that exceeds this range can be regarded as a potential anesthesia abnormality and triggers the abnormality detection mechanism. This method helps medical staff detect potential abnormalities in the patient's anesthesia status and improves the accuracy and safety of anesthesia status analysis.

[0054] As an optional implementation process, for step S5, the process of performing anesthesia state diffusion detection on the patient to be analyzed is specifically to first determine the target anesthesia state diffusion path of the patient to be analyzed, and implement anesthesia state diffusion detection according to the EEG sign target diffusion vector corresponding to the target anesthesia state diffusion path. Among them, the target anesthesia state diffusion path is specifically generated by data mining the historical anesthesia state monitoring data of multiple patients in the anesthesia maintenance stage, which specifically includes the following generation process: Determine the anesthesia stage diffusion path of each local anesthesia stage, construct a diffusion sample data set containing multiple groups of diffusion sample data for each anesthesia stage diffusion path, and extract the first EEG sign corrected association vector and the second EEG sign corrected association vector of each group of diffusion sample data in each diffusion sample data set.

[0055] It should be noted that for each local anesthesia stage, except for the two stages at the boundary of the anesthesia maintenance stage, the remaining stages correspond to two anesthesia stage diffusion paths, that is, diffusion from itself to the direction of BIS value increase and decrease. For at least one anesthesia stage diffusion path of each local anesthesia stage, multiple sets of diffusion sample data of each anesthesia stage diffusion path are extracted based on the historical anesthesia status monitoring data of each patient, so as to construct a diffusion sample data set of each anesthesia stage diffusion path.

[0056] Among them, in the process of constructing the stage sample data set of each local anesthesia stage, the data was first preliminarily divided based on the local anesthesia stage according to the patient's actual BIS value. However, this process did not take into account the patient's anesthetic state change process. For example, the patient's local anesthesia stage changes from stage A to stage B and stage C in the direction of decreasing BIS value, and then regresses from stage C to stage B and stage A in sequence. In this process, the historical sample data in stage A are used to construct the stage sample data set of local anesthesia stage A, that is, the patient's BIS value has a process of decreasing and increasing, so that a local anesthesia stage will contain data of the decreasing process and the increasing process. The data in each local anesthesia stage was finally divided into multiple groups of historical sample data, but in the process of constructing the stage sample data set, this BIS value change, that is, the diffusion change of the anesthetic state, was not considered. As for the diffusion sample data set of the diffusion path of the anesthesia stage, the change process of the anesthesia state is specifically considered. For example, in the process of transforming from stage A to stage B, stage A contains three groups of historical sample data A1, A2, and A3 that change in time series, and stage B contains three groups of historical sample data B1, B2, and B3 that change in time series. Then, in the anesthesia stage diffusion path from stage A to stage B, a specific group of diffusion sample data is the historical sample data A3 and B1, which are used as data references in the diffusion process of the local anesthesia stage to describe the change information of the anesthesia state in the scenario across the anesthesia stage. According to the above steps, the first EEG sign correction association vector and the second EEG sign correction association vector corresponding to each group of diffusion sample data can be obtained, which respectively characterize the historical EEG sign correction association vectors corresponding to the current local anesthesia stage and the next diffused local anesthesia stage, that is, the above-mentioned association vector that takes into account the personalized physiological characteristics of the patient.

[0057] The association state diffusion trend vector of each group of diffusion sample data can be generated by the first EEG sign correction association vector and the second EEG sign correction association vector. The association state diffusion trend vector can be specifically composed of the difference between the characteristic values ​​of each sign state parameter between the second EEG sign correction association vector and the first EEG sign correction association vector, which represents the synergistic change trend of the EEG state change and the sign state synergistic change process in the diffusion process of the anesthesia state. Finally, on the basis of considering the particularity of individual patients, the association state diffusion trend vectors representing the synergistic change trend of multiple groups of diffusion sample data are aggregated to obtain the EEG sign target diffusion vector that comprehensively represents the change trend of the group association pattern of the EEG state and the sign state in the diffusion path of each anesthesia stage.

[0058] The significance of selecting the first EEG sign corrected association vector and the second EEG sign corrected association vector to generate the association state diffusion trend vector and finally constructing the EEG sign target diffusion vector is that, since the construction of the diffusion sample data is based on two adjacent groups of historical sample data across stages, in this short period of time, the degree of change in the patient's own EEG state is actually relatively small, which is different from the changes in the entire local anesthesia stage. As a result, if the association state diffusion trend vector is generated only based on the historical EEG sign association vector corresponding to the diffusion sample data, the diffusion trend finally represented will be relatively weak, resulting in the problem of insufficient detection accuracy in actual detection. By generating the association state diffusion trend vector based on the corrected first EEG sign corrected association vector and the second EEG sign corrected association vector, the actual changes in the data in a short period of time can be amplified from a global perspective, so that the anesthesia state diffusion detection can be better realized based on the EEG sign target diffusion vector.

[0059] In this case, for the actual anesthesia state diffusion detection process, since the generation of the EEG sign target diffusion vector takes into account the individual differences of patients and then characterizes the common characteristics of the group, it can include the diffusion trend of some special individuals. If this diffusion trend is not exceeded during the real-time monitoring process, it is considered to be a normal diffusion state, otherwise it is considered that the diffusion state is abnormal. For patients who are judged to be in a normal diffusion state, when the patient is in a new local anesthesia stage, the above scheme can be used to perform local static detection of the new local anesthesia stage again. If it does not meet the relevant local state changes, it will still be marked as abnormal, avoiding missed detection of abnormal states and ensuring the accuracy of the analysis of the patient's anesthesia state.

[0060] As one of the implementation processes of the present invention, a patient anesthesia state analysis system based on vital sign data is also provided, which is used to implement the above-mentioned patient anesthesia state analysis method based on vital sign data, such as Figure 2 As shown, the system includes: The anesthesia state monitoring module is used to extract the real-time EEG state parameter sequence and multiple real-time vital sign state parameter sequences of the patient to be analyzed after acquiring the real-time anesthesia state monitoring data of the patient to be analyzed in the current detection stage, the real-time anesthesia state monitoring data including the real-time EEG state monitoring data and real-time vital sign monitoring data of the patient to be analyzed; The EEG sign correlation analysis module is used to calculate the conditional entropy between the real-time EEG state parameter sequence and multiple real-time sign state parameter sequences, so as to construct a real-time EEG sign correlation vector of the patient to be analyzed at the current detection stage; The module for detecting local abnormality of the anesthesia state is used to determine the first local anesthesia stage of the patient to be analyzed according to the real-time EEG state monitoring data in the current detection stage, determine the EEG sign target association vector and the EEG sign association reference range corresponding to the real-time EEG sign association vector, and perform local abnormality detection of the anesthesia state on the patient to be analyzed according to the EEG sign target association vector and the EEG sign association reference range to obtain the local abnormality detection result of the anesthesia state; an anesthetic state diffusion detection module, for performing an anesthetic state diffusion detection on the patient to be analyzed to generate an anesthetic state diffusion abnormality detection result if the anesthetic state local abnormality detection result indicates that the patient to be analyzed does not have a local abnormality in the anesthetic state; The anesthetic state diffusion detection of the patient to be analyzed includes obtaining real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage, determining the second local anesthesia stage of the patient to be analyzed according to the real-time anesthetic state monitoring data of the next detection stage, generating a target anesthetic state diffusion path of the patient to be analyzed, determining the EEG sign target diffusion vector of the target anesthetic state diffusion path, and performing anesthetic state diffusion detection on the real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage based on the EEG sign target diffusion vector.

[0061] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be obvious to those skilled in the art. Parts not described in detail in this specification belong to the prior art known to those skilled in the art.

Claims

1. A method for analyzing the anesthesia state of a patient based on vital sign data, characterized in that: include: After obtaining the real-time anesthesia state monitoring data of the patient to be analyzed in the current detection stage, the real-time anesthesia state monitoring data includes the real-time EEG state monitoring data and real-time vital sign monitoring data of the patient to be analyzed, extracting the real-time EEG state parameter sequence and multiple real-time vital sign state parameter sequences of the patient to be analyzed; The conditional entropy between the real-time EEG state parameter sequence and multiple real-time physical sign state parameter sequences is calculated respectively to construct a real-time EEG physical sign association vector of the patient to be analyzed at the current detection stage; Determine the first local anesthesia stage of the patient to be analyzed according to the real-time EEG state monitoring data in the current detection stage, and determine the EEG sign target association vector and the EEG sign association reference range corresponding to the real-time EEG sign association vector; Performing local abnormality detection of the anesthesia state of the patient to be analyzed according to the EEG sign target association vector and the EEG sign association reference range, and obtaining the local abnormality detection result of the anesthesia state; If the result of the local abnormality detection of the anesthesia state is that the patient to be analyzed does not have a local abnormality of the anesthesia state, then performing an anesthesia state diffusion detection on the patient to be analyzed to generate an anesthesia state diffusion abnormality detection result; The anesthetic state diffusion detection of the patient to be analyzed includes obtaining real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage, determining the second local anesthesia stage of the patient to be analyzed according to the real-time anesthetic state monitoring data of the next detection stage, generating a target anesthetic state diffusion path of the patient to be analyzed, determining the EEG sign target diffusion vector of the target anesthetic state diffusion path, and performing anesthetic state diffusion detection on the real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage based on the EEG sign target diffusion vector.

2. A method for analyzing the anesthesia state of a patient based on vital sign data according to claim 1, characterized in that: For the EEG sign target association vector and EEG sign association reference range, it also includes: Acquire historical anesthesia status monitoring data of multiple patients in the anesthesia maintenance stage, the historical anesthesia status monitoring data including the historical EEG status monitoring data and historical vital sign monitoring data of the patients in the anesthesia maintenance stage; The anesthesia maintenance stage is divided into multiple local anesthesia stages according to the historical EEG state monitoring data, multiple groups of historical sample data of each local anesthesia stage are extracted from multiple groups of historical anesthesia state monitoring data, and a stage sample data set of each local anesthesia stage is constructed; Extract the historical EEG state parameter sequence and multiple historical sign state parameter sequences of each group of historical sample data in each stage sample data set, calculate the conditional entropy between the historical EEG state parameter sequence and multiple historical sign state parameter sequences in each group of historical sample data, and construct the historical EEG sign association vector of each group of historical sample data; Constructing an EEG sign association matrix of each stage sample data set according to multiple historical EEG sign association vectors in the stage sample data set, performing covariance calculation on each EEG sign association matrix to generate a covariance matrix of each EEG sign association matrix, and performing eigenvalue decomposition on multiple covariance matrices to obtain multiple reference eigenvectors and eigenvalues ​​of each reference eigenvector; Determine multiple target feature vectors from multiple reference feature vectors, calculate the association score of each sign state parameter according to the eigenvalues ​​corresponding to the multiple target feature vectors, generate the EEG sign association reference vector for each local anesthesia stage, and determine the EEG sign target association vector and EEG sign association reference range for each local anesthesia stage according to the EEG sign association reference vector.

3. A method for analyzing the anesthesia state of a patient based on vital sign data according to claim 2, characterized in that: The EEG sign target association vector and the EEG sign association reference range of each local anesthesia stage are determined according to the EEG sign association reference vector, including: Based on the EEG sign association reference vector, multiple historical EEG sign association vectors in the stage sample data set of each local anesthesia stage are associated and corrected to obtain a historical EEG sign correction association vector corresponding to each historical EEG sign association vector; A target EEG sign association vector for each local anesthesia stage is generated based on multiple historical EEG sign correction association vectors, and the distances between the EEG sign target association vector and multiple historical EEG sign correction association vectors are calculated respectively. The EEG sign association reference range is determined based on the EEG sign target association vector, and a target cumulative analysis is performed on the EEG sign association reference range, including determining the EEG sign association reference range with the EEG sign target association vector as the center, and calculating the target cumulative ratio after each expansion of the EEG sign association reference range based on the target reference distance, completing the expansion of the EEG sign association reference range after reaching the target cumulative threshold, and outputting the EEG sign association reference range for each local anesthesia stage.

4. A method for analyzing the anesthesia state of a patient based on vital sign data according to claim 3, characterized in that: For the EEG sign target diffusion vector, it also includes: Determine the anesthesia stage diffusion path of each local anesthesia stage, and construct a diffusion sample data set containing multiple groups of diffusion sample data for each anesthesia stage diffusion path; The first EEG sign correction association vector and the second EEG sign correction association vector of each group of diffusion sample data in each diffusion sample data set are extracted, and the association state diffusion trend vector of each group of diffusion sample data is generated according to the first EEG sign correction association vector and the second EEG sign correction association vector, and the EEG sign target diffusion vector of the diffusion path of each anesthesia stage is generated according to multiple association state diffusion trend vectors.

5. The method for analyzing the anesthesia state of a patient based on vital sign data according to claim 3, characterized in that: According to the EEG sign target association vector and the EEG sign association reference range, the local abnormality detection of the anesthesia state of the patient to be analyzed is performed to obtain the local abnormality detection result of the anesthesia state, including: According to the EEG sign association reference vector corresponding to the first local anesthesia stage of the patient to be analyzed, the real-time EEG sign association vector of the patient to be analyzed in the current detection stage is corrected to generate a real-time EEG sign association correction vector, and the target distance between the real-time EEG sign association correction vector and the EEG sign target association vector is calculated. If the target distance is within the EEG sign association reference range, it is considered that the patient to be analyzed has no local abnormality in the anesthetic state; otherwise, it is considered that the patient to be analyzed has a local abnormality in the anesthetic state and an abnormal anesthetic state warning is issued.

6. A method for analyzing the anesthesia state of a patient based on vital sign data according to claim 4, characterized in that: Based on the EEG sign target diffusion vector, the real-time anesthesia state monitoring data of the patient to be analyzed in the next detection stage is subjected to anesthesia state diffusion detection, including: The real-time EEG sign association vector corresponding to the real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage is extracted; the real-time EEG sign diffusion vector of the patient to be analyzed is calculated based on the real-time EEG sign association vector of the patient to be analyzed in the current detection stage; the real-time EEG sign diffusion vector is analyzed according to the EEG sign target diffusion vector, the diffusion deviation state of each sign state parameter is determined, and the anesthetic state diffusion abnormality detection result of the patient to be analyzed is generated.

7. The method for analyzing the anesthesia state of a patient based on vital sign data according to claim 2, characterized in that: The correlation score of each physical sign state parameter is calculated based on the eigenvalues ​​corresponding to multiple target eigenvectors, including: The element value of each sign state parameter in the target feature vector is analyzed by the eigenvalue corresponding to the target feature vector to obtain the local score of each sign state parameter in each target feature vector. The local scores of the sign state parameters in multiple target feature vectors are accumulated to obtain the association score of each sign state parameter.

8. A patient anesthesia status analysis system based on vital sign data, characterized in that: The system is used to implement a method for analyzing a patient's anesthesia state based on vital sign data as described in any one of claims 1 to 7, comprising: The anesthesia state monitoring module is used to extract the real-time EEG state parameter sequence and multiple real-time vital sign state parameter sequences of the patient to be analyzed after acquiring the real-time anesthesia state monitoring data of the patient to be analyzed in the current detection stage, the real-time anesthesia state monitoring data including the real-time EEG state monitoring data and real-time vital sign monitoring data of the patient to be analyzed; The EEG sign correlation analysis module is used to calculate the conditional entropy between the real-time EEG state parameter sequence and multiple real-time sign state parameter sequences, so as to construct a real-time EEG sign correlation vector of the patient to be analyzed at the current detection stage; The module for detecting local abnormality of the anesthesia state is used to determine the first local anesthesia stage of the patient to be analyzed according to the real-time EEG state monitoring data in the current detection stage, determine the EEG sign target association vector and the EEG sign association reference range corresponding to the real-time EEG sign association vector, and perform local abnormality detection of the anesthesia state on the patient to be analyzed according to the EEG sign target association vector and the EEG sign association reference range to obtain the local abnormality detection result of the anesthesia state; an anesthetic state diffusion detection module, for performing an anesthetic state diffusion detection on the patient to be analyzed to generate an anesthetic state diffusion abnormality detection result if the anesthetic state local abnormality detection result indicates that the patient to be analyzed does not have a local abnormality in the anesthetic state; The anesthetic state diffusion detection of the patient to be analyzed includes obtaining real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage, determining the second local anesthesia stage of the patient to be analyzed according to the real-time anesthetic state monitoring data of the next detection stage, generating a target anesthetic state diffusion path of the patient to be analyzed, determining the EEG sign target diffusion vector of the target anesthetic state diffusion path, and performing anesthetic state diffusion detection on the real-time anesthetic state monitoring data of the patient to be analyzed in the next detection stage based on the EEG sign target diffusion vector.

Images