Monitoring method and system based on intrinsic multi-scale sample entropy

Abstract

The invention provides a monitoring method and system based on intrinsic multi-scale sample entropy. The monitoring method comprises the following steps: acquiring original data; preprocessing the data to obtain a data time sequence to be analyzed; a step of applying the EEMD to the data time sequence and decomposing the data time sequence into E-IMFs; and an iMSE calculation step, wherein a plurality of iMSEs are obtained. At least a portion of the plurality of iMSEs is used as an output of the monitoring method.

Description

Technical Field

[0001] The present application relates to the field of medical parameter monitoring, and in particular to a monitoring method and system based on intrinsic multi-scale sample entropy. Background Art

[0002] Anesthesia is a unique medical procedure. It has no therapeutic value in itself, but it is an integral part of medical interventions, from life-saving surgical procedures to routine invasive medical examinations. Methods to measure the depth of anesthesia (DOA) are critical in surgical procedures and provide a good aid to the surgical process.

[0003] Existing methods for detecting the depth of anesthesia include continuous electrocardiogram (ECG) monitoring, continuous pulse oximetry (SpO2) monitoring, blood pressure monitoring, anesthetic concentration monitoring, carbon dioxide monitoring, temperature measurement, electroencephalogram (EEG) monitoring, etc. Among them, most devices are designed to monitor important physiological indicators mainly related to cardiovascular or pulmonary function, while electroencephalogram (EEG) is a device specifically used to monitor the depth of anesthesia through consciousness assessment. It is the only device designed to monitor the dynamic function of the brain to determine the patient's level of consciousness; it is the basis for many objective measures of anesthesia depth. EEG monitoring methods currently include compressed spectral array (CSA), bispectral index (BIS), Methods such as index and entropy analysis; however, these existing methods generally have the disadvantage that the dimension of feature parameters is not comprehensive enough. Summary of the invention

[0004] In order to overcome some problems existing in the prior art, the present application provides a monitoring method and system based on intrinsic multi-scale sample entropy.

[0005] The first aspect of the present application provides a monitoring method based on intrinsic multi-scale sample entropy, comprising the following steps:

[0006] Steps to obtain raw data: collect data as raw data;

[0007] Data preprocessing step: preprocess the acquired raw data to obtain the data time series to be analyzed;

[0008] The steps of applying EEMD to the data time series are as follows: the data time series is processed by the EEMD analysis method, and is adaptively decomposed into a plurality of ensemble intrinsic mode functions (E-IMFs);

[0009] The calculation steps of intrinsic multi-scale sample entropy (iMSE calculation steps):

[0010] Formula (1) is used to calculate the partial component sum of N set intrinsic mode functions to obtain N partial component sums;

[0011]

[0012] Among them, c j (t) is the collective intrinsic mode function (E-IMF), t represents the number of data, N represents the total number of E-IMFs, which is a positive integer, n represents the nth E-IMF, and x n (t) represents the sum from the first E-IMF to the nth E-IMF;

[0013] Then, the multi-scale sample entropy of each partial component and is calculated to obtain N intrinsic multi-scale sample entropies (iMSE), which constitute a set of intrinsic multi-scale sample entropies (iMSE set);

[0014] At least a portion of the iMSE set (eg, 2, 3, 5 or 8 of 1MSE-8MSE, etc.) is used as the output of the monitoring method.

[0015] In one embodiment, the monitoring method further includes a step of calculating a set intrinsic multi-scale sample entropy (eiMSE calculation step): setting a plurality of white noise sequences with different amplitude ratios, applying them respectively to the step of applying the EEMD to the data time series, and performing the iMSE calculation step to obtain multiple rounds of iMSE sets; taking the set average of the iMSE sets under multiple rounds as the result, obtaining a plurality of set intrinsic multi-scale sample entropies (eiMSE), constituting a set of set intrinsic multi-scale sample entropies (eiMSE set); at least a part of the eiMSE set (for example, 2, 3, 5 or 8 of e1MSE-e8MSE, etc.) is used as the output of the monitoring method.

[0016] In one embodiment, in the step of acquiring raw data, a sliding window is used to collect data, and the data in each window is used as the raw data; the monitoring method can obtain multiple iMSEs and / or multiple eiMSEs of each window.

[0017] In one embodiment, in the step of acquiring raw data, the sampling frequency of the data is not less than 100 Hz.

[0018] In one embodiment, in the data preprocessing step, the data preprocessing includes filtering, outlier removal and electrical noise removal.

[0019] In one embodiment, the step of applying EEMD to the data time series specifically includes:

[0020] Add white noise: add a white noise sequence to the data time series to be analyzed; the amplitude ratio of the white noise is any value between 0.05 and 0.2;

[0021] Decompose IMFs using EMD: Adaptively decompose the data time series with white noise into multiple initial intrinsic mode functions (IMFs) using EMD;

[0022] Repeat the steps of adding white noise and decomposing IMFs using EMD, but use a newly randomly generated white noise sequence with the same amplitude ratio each time; thereby obtaining multiple batches with multiple IMFs respectively;

[0023] The ensemble average of multiple IMFs of multiple batches is taken as the decomposition result to obtain the multiple E-IMFs.

[0024] In one embodiment, in the step of applying EEMD to the data time series, the step of adding white noise and the step of decomposing IMFs using EMD are repeated 10-20 times to obtain 10-20 batches.

[0025] In one embodiment, in the iMSE calculation step, a maximum of 8 of the multiple E-IMFs are selected for calculating the partial component sum; that is, N is a positive integer less than or equal to 8, for example, 8, 7, 6, etc.

[0026] In one embodiment, in the step of calculating eiMSE, the amplitude ratio of white noise is set to 0.05-0.2, and 6-16 sets of trials are selected as the white noise sequences with multiple different amplitude ratios.

[0027] In one embodiment, the monitoring method further comprises a data output step, wherein at least a portion of the multiple iMSEs or eiMSEs of each window are sorted and output, and the results are displayed on a monitor and archived as a monitoring guide.

[0028] In one embodiment, in the step of outputting the data, at least one of the following methods is selected:

[0029] (1) Outputting all or part of the multiple iMSEs of each window to obtain an iMSE monitoring diagram of each iMSE on the time axis for monitoring guidance;

[0030] (2) A linear regression model is used to perform linear regression fitting on multiple iMSEs of each window to obtain an index related to the iMSE, and the exponential curve of each window is output as an eiMSE monitoring curve for monitoring guidance;

[0031] (3) Outputting all or part of the multiple eiMSEs of each window to obtain an eiMSE monitoring graph of each eiMSE on the time axis for monitoring and guidance;

[0032] (4) A linear regression model is used to perform linear regression fitting on the multiple eiMSEs of each window to obtain an index related to eiMSE, and the exponential curve related to each window is output as an eiMSE monitoring curve for monitoring guidance.

[0033] In one embodiment, a monitoring diagram and a monitoring curve are used simultaneously for monitoring guidance.

[0034] The second aspect of the present application provides a monitoring system based on intrinsic multi-scale sample entropy, which adopts the monitoring method described in any of the above embodiments. The monitoring system includes a data acquisition module and a data analysis module, wherein:

[0035] The data acquisition module is configured to collect data, thereby obtaining raw data;

[0036] The data analysis module includes a windowing module, a data preprocessing module, an EEMD decomposition module, an iMSE calculation module and a result output module; wherein,

[0037] The windowing module is configured to collect data in a windowed manner and obtain the original data of each window;

[0038] A data preprocessing module is configured to filter, remove outliers and remove electrical noise from the collected raw data to obtain a data time series to be analyzed;

[0039] An EEMD decomposition module is configured to adaptively decompose the data time series to be analyzed in the current window into multiple E-IMFs using an EEMD analysis method;

[0040] The iMSE calculation module is configured to select the first N E-IMFs from the multiple E-IMFs, calculate the partial component sums using formula (1), and obtain N partial component sums.

[0041]

[0042] Among them, c j (t) is the E-IMF, t represents the number of data, N represents the total number of E-IMFs, which is a positive integer, n represents the nth E-IMF, x n (t) represents the sum from the first E-IMF to the nth E-IMF;

[0043] Then, the multi-scale sample entropy of each partial component and the iMSE set of the current window is obtained;

[0044] The result output module is configured to sort and output the multiple iMSEs of each window.

[0045] In one embodiment, the monitoring system also includes: an eiMSE calculation module, which is configured to set multiple white noise sequences with different amplitude ratios, and apply them to the EEMD decomposition module and the iMSE calculation module respectively to obtain multiple rounds of iMSE sets; taking the set average of the iMSE sets under multiple rounds as the result, to obtain multiple eiMSEs of the current window; the result output module is configured to sort and output the multiple eiMSEs of each window.

[0046] In one embodiment, the monitoring system further comprises:

[0047] a data archiving module configured to store data for future reference;

[0048] The data display module is configured to display the output result of the result output module on a monitor as a monitoring guide.

[0049] The third aspect of the present application provides an application of the monitoring method based on intrinsic multi-scale sample entropy described in any of the above embodiments, and the monitoring method can be used for medical monitoring, including but not limited to anesthesia depth monitoring. Correspondingly, the original data is EEG data.

[0050] In a fourth aspect, the present application provides a computer device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the computer program, the monitoring method based on intrinsic multi-scale sample entropy as described in any of the foregoing embodiments is implemented.

[0051] In a fifth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements a monitoring method based on intrinsic multi-scale sample entropy as described in any of the foregoing embodiments.

[0052] The monitoring method provided by at least one embodiment of the present application is based on intrinsic multiscale sample entropy (iMSE) or ensemble intrinsic multiscale sample entropy (eiMSE). The advantages of iMSE and eiMSE are: it is a two-dimensional complexity feature, which includes the characteristics of the past multiscale entropy and adds more features that it does not have. When fitting the depth of anesthesia, there are more control parameters to fit the actual state of anesthesia. In addition, in previous studies, we also found that the intrinsic multiscale sample entropy (iMSE) has a higher classification accuracy when classifying anesthesia at different depths.

[0053] The monitoring method provided by at least one embodiment of the present application can be effectively applied to the medical field and other related fields, and contribute to data monitoring. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] Figure 1is a flow chart of a monitoring method according to one embodiment of the present application;

[0055] Figure 2 It is a schematic diagram of collecting EEG data;

[0056] Figure 3a It is the original data of the current window;

[0057] Figure 3b is the preprocessed data time series to be analyzed;

[0058] Figure 4 This is an example of multiple E-IMFs obtained by EEMD decomposition;

[0059] Figure 5 is a curve distribution diagram of iMSE according to an implementation mode;

[0060] Figure 6 is with Figure 5 The corresponding two-dimensional graph of iMSE;

[0061] Figure 7 is a curve distribution diagram of eiMSE according to an implementation mode;

[0062] Figure 8 is with Figure 7 The corresponding two-dimensional graph of eiMSE;

[0063] Fig. 9 It is the eiMSE two-dimensional monitoring diagram obtained by the monitoring method of this embodiment during the entire anesthesia surgery;

[0064] Fig.10 is the eiMSE monitoring curve obtained by the monitoring method of this embodiment during the entire anesthesia surgery;

[0065] Fig.11 It is the BIS monitoring curve and the preprocessed EEG data;

[0066] Fig.12 It is a connection diagram of computer equipment. DETAILED DESCRIPTION

[0067] In order to make the purpose, technical solutions and advantages of the present application clearer, the present application is described and illustrated below in conjunction with the accompanying drawings and implementation methods. It should be understood that the specific implementation methods described herein are only used to explain the present application and are not used to limit the present application. Based on the implementation methods provided in the present application, all other implementation methods obtained by ordinary technicians in the field without making creative work are within the scope of protection of the present application.

[0068] Obviously, the drawings described below are only some examples or implementations of the present application. For ordinary technicians in this field, the present application can also be applied to other similar scenarios based on these drawings without creative work. In addition, it can also be understood that although the efforts made in this development process may be complicated and lengthy, for ordinary technicians in this field related to the content disclosed in this application, some changes in design, manufacturing or production based on the technical content disclosed in this application are just conventional technical means, and should not be understood as insufficient content disclosed in this application.

[0069] Reference to "embodiment" in this application means that a particular feature, structure, or characteristic described in conjunction with the embodiment may be included in at least one embodiment of the present application. The appearance of the phrase in various locations in the specification does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment that is mutually exclusive with other embodiments. It is explicitly and implicitly understood by those of ordinary skill in the art that the embodiments described in this application may be combined with other embodiments without conflict.

[0070] Unless otherwise defined, the technical terms or scientific terms involved in this application should be understood by people with ordinary skills in the technical field to which this application belongs. The words "one", "a", "a", "the" and the like involved in this application do not indicate a quantitative limitation, and may represent the singular or plural. The terms "include", "comprise", "have" and any of their variations involved in this application are intended to cover non-exclusive inclusions; for example, a process, method, system, product or device that includes a series of steps or modules (units) is not limited to the listed steps or units, but may also include steps or units that are not listed, or may also include other steps or units inherent to these processes, methods, products or devices. The words "connect", "connected", "coupled" and the like involved in this application are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The "multiple" involved in this application refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist.

[0071] Multiscale Sample Entropy (CK Peng et al., 2002\2005) introduces the concept of scale on the basis of traditional entropy, and transforms the original entropy from a point into a line that changes with the scale. Traditional multiscale sample entropy is a quantification of the complexity of the data as a whole. But in fact, the data itself contains different components, and under the influence of different components, the multiscale sample entropy will also change.

[0072] This application proposes a new concept of intrinsic multiscale sample entropy (Intrinsic Multiscale SampleEntropy, iMSE) and ensemble intrinsic multiscale sample entropy (Ensemble Intrinsic Multiscale SampleEntropy, eiMSE), and applies it to fields such as medical monitoring, such as monitoring of anesthesia depth, etc. This implementation mainly takes the monitoring of anesthesia depth as an example to illustrate; however, the analysis and processing methods in other fields or directions are the same or similar. The intrinsic multiscale sample entropy and ensemble intrinsic multiscale sample entropy provided by this application, on the basis of multiscale sample entropy, take into account the different components contained in the data itself, increase the dimension of complexity, and provide high-dimensional characteristic parameters for fitting the depth of anesthesia.

[0073] The first embodiment of the present application provides an anesthesia depth monitoring method based on intrinsic multi-scale sample entropy (hereinafter referred to as the monitoring method). Figure 1 It is a flow chart of the monitoring method according to the present embodiment. It should be noted that the steps shown in the flow chart of the monitoring method or the accompanying drawings can be executed in a computer system such as a set of computer executable instructions. The monitoring method comprises the following steps:

[0074] S100 steps to obtain raw data: Figure 2 As shown, a single electrode deployed on the patient's forehead and a reference electrode near the ear (temporal bone mastoid) can be used to collect the patient's EEG data (EEG data, in μV) to obtain raw data. The sampling frequency of the electrode is generally not less than 100 Hz, for example, it can be 100 Hz, 256 Hz, 500 Hz, etc.

[0075] In order to report the results in near real time, the EEG data can be collected in a windowed manner. In this embodiment, a sliding window is used; after setting the window length and the sliding step length, the window sliding stage is entered to obtain the raw data of each window.

[0076] Below, a more detailed explanation is given using specific data as an example for easier understanding; however, the protection of this application is not limited to this. Set the sampling frequency of the electrode to 100 Hz, set the window length to 60 s, and the sliding step length to 1 s; after turning on the system, start collecting data, collect 100 EEG data every 1 second and transmit them to the computer; when the computer receives 60 s of EEG data, it obtains the raw data of the current window, which contains 6000 (60*100) data. During the window sliding stage, multiple raw data can be obtained in sequence. For example, since the window sliding stage is entered, after the EEG data collection starts, 6000 data from the 1st to the 60th second are obtained first, which are the first original data of the first window; after the window moves for 1 second, 6000 data from the 2nd to the 61st second can be obtained, which are the second original data of the second window; after the window moves for another 1 second, 6000 data from the 3rd to the 62nd second can be obtained, which are the third original data of the third window, and so on. In the window sliding stage, multiple original data of each window can be obtained in sequence. According to this method, a large number of original data of each window from the beginning to the end of the operation can be obtained. The first 59 seconds of the latter window are the same as the last 59 seconds of the previous window, so that the results can be reported in near real time.

[0077] S200 Data preprocessing step: preprocessing the acquired raw data of the current window to obtain a time series of data to be analyzed in the current window.

[0078] It is worth noting that when analyzing data subsequently, only the data in the current window is analyzed as an example; the data in other windows are processed in the same way.

[0079] The amplitude of EEG data from a typical adult's scalp is only 10μV to 100μV, which is relatively small, but the acquired raw data is full of signal interference with relatively large amplitudes, such as interference from blinking, clenching teeth, body movement, touching EEG electrodes, and nearby power devices, as well as drift. Therefore, the raw data needs to be preprocessed.

[0080] The data preprocessing includes filtering, outlier removal and electrical noise removal.

[0081] Still taking the specific data as an example, the sampling frequency of the original data is 100Hz, and the current window acquires 6000 original data. The filter is set to high pass 0.5Hz; low pass 48Hz, so that the high pass filter removes the frequency below 0.5Hz, and the low pass filter removes the frequency above 48Hz, and at the same time achieves the removal of electrical noise (50Hz power frequency). In addition, in order to eliminate the interference of other equipment or actions in the operating room, all outliers are deleted, where any value greater than the mean plus 5 times the standard deviation of the absolute value of the original data filtered in the current window is defined as an outlier. Specifically, the mean and standard deviation are calculated for the absolute value of the 6000 filtered data in the current window. When the absolute value of the filtered data is greater than the mean plus 5 times the standard deviation, it is defined as an outlier and removed; when the outlier is removed, the corresponding outlier will be averaged with multiple values ​​around it, so the overall data length will not be reduced, and it is still 6000, that is, the data time series to be analyzed after processing still contains 6000 data. Figure 3a is an example of the raw data of a window. Figure 3b is an example of the preprocessed data to be analyzed in the window; their horizontal coordinates are all time (unit s), and the vertical coordinates are EEG data (unit μV). When the data of a window is in the processing state, it is the current window.

[0082] The steps of applying S300 EEMD to the data time series are as follows: the EEMD analysis method is used to process the data time series to be analyzed in the current window, and it is adaptively decomposed into multiple set intrinsic mode functions E-IMFs (s represents a complex number, multiple).

[0083] In the prior art, Empirical Mode Decomposition (EMD, Huang et al., 1998) is an adaptive method designed to analyze nonlinear and non-stationary data. It can extract intrinsic mode functions (IMFs) from any data and has the following good characteristics: they are all zero mean, symmetric about zero, and binary narrow band; therefore, IMFs provide compact support for the distribution of data from the trend scale to the whole. In order to ensure the stability of the EMD method, an integrated method of adding different noises can also be used to help decomposition (Wu and Huang, 2009), which is called ensemble empirical mode decomposition (EEMD). EEMD can be implemented by the following steps: (1) adding a white noise sequence to the target data sequence to be analyzed; (2) using EMD to decompose the target data sequence with added white noise into IMFs; (3) repeating steps (1) and (2), but using different white noise sequences each time; (4) taking the ensemble average of the IMFs obtained from multiple trial decompositions as the final decomposition result to obtain the ensemble intrinsic mode functions E-IMFs.

[0084] In this implementation, step S300 specifically includes:

[0085] S301 Add white noise: Add a white noise sequence to the data time series to be analyzed in the current window. The white noise sequence is randomly generated and has the same length as the data time series. The amplitude of the white noise is determined according to the amplitude of the data to be analyzed, and is generally any value between 0.05-0.2 of the amplitude of the data time series, that is, the amplitude ratio of the white noise is any value between 0.05-0.2. If it is too small, the effect is not obvious; if it is too large, it interferes with the original data.

[0086] S302 uses EMD to decompose IMFs: the data time series with the white noise sequence added is adaptively decomposed into a plurality of initial intrinsic mode functions IMFs using EMD to form an initial set of intrinsic mode functions (IMFs set).

[0087] S303 continuously repeats S301 and S302, but uses a newly randomly generated white noise sequence with the same amplitude ratio each time, thereby obtaining multiple batches with IMFs sets. Generally, it is repeated 10-20 times, thereby obtaining 10-20 batches, each batch having an IMFs set containing multiple IMFs.

[0088] S304 takes the collective average of the IMFs sets of the multiple batches as the decomposition result, obtains multiple collective intrinsic mode functions E-IMFs, and constitutes a set of collective intrinsic mode functions E-IMFs (E-IMFs set).

[0089] In step S303, the data is processed by using multiple cyclic decompositions but with different white noise each time. This processing method can not only make up for the deficiency of EMD, but also reduce the influence of white noise on the final data.

[0090] The IMFs set includes multiple IMFs, which may be IMF1, IMF2, IMF3, etc. The E-IMFs set includes multiple E-IMFs, which may be E-IMF1, E-IMF2, E-IMF3, etc. It is well known to those skilled in the art that when calculating the set average, the result is obtained by averaging each element in each set; for example, in step S304, the set average of multiple batches of IMFs sets is to average IMF1 in each batch, average IMF2 in each batch, ..., so as to obtain an E-IMFs set containing these average values.

[0091] Both the IMFs set and the E-IMFs set can be used in the subsequent iMSE calculation steps. Since the E-IMFs set is iterated multiple times relative to the IMFs set, the E-IMFs set is more preferred. However, it is worth understanding that when describing the decomposition results of EEMD, in the absence of the E-IMFs set, the result can also be considered as the IMFs set, that is, the IMFs set is used as the multiple set intrinsic mode functions. At this time, EEMD can also be regarded as EMD.

[0092] Let’s continue to take specific data as an example. The data time series to be analyzed has 6000 data. A white noise sequence containing 6000 data is randomly generated, and its amplitude is set to 10% of the amplitude of the data time series (the amplitude ratio is 0.1). Then, the EEMD analysis method is used for decomposition, and 9 set intrinsic mode functions E-IMFs can be obtained, namely E-IMF1, E-IMF2, E-IMF3, E-IMF4, E-IMF5, E-IMF6, E-IMF7, E-IMF8, and E-IMF9, which constitute an E-IMFs set; each set intrinsic mode function E-IMF contains 6000 data. Figure 4 The example of adding 10% white noise sequence and obtaining 9 E-IMFs by EEMD decomposition is shown in Figure 1. E-IMF1 to E-IMF9 are arranged from high frequency to low frequency, which is the result of EEMD adaptive decomposition. Since EMD and EEMD analysis methods are already very mature methods, we will not describe them in detail here.

[0093] S400: Calculation step of intrinsic multi-scale sample entropy (iMSE calculation step): Select up to 8 of the multiple E-IMFs of the current window to calculate the intrinsic multi-scale sample entropy (iMSE) of the current window. Specifically:

[0094] S401 selects the first N E-IMFs of the current window, where N is a positive integer not greater than 8.

[0095] S402 uses formula (1) to calculate the partial component sums of the selected E-IMFs to obtain N partial component sums:

[0096]

[0097] Among them, c j (t) is the collective intrinsic mode function (E-IMF), t represents the number of data, N represents the total number of E-IMFs, n represents the nth E-IMF, and x n (t) represents the sum from the 1st E-IMF to the nth E-IMF.

[0098] S403 calculates the multi-scale sample entropy of each partial component and , thereby obtaining N intrinsic multi-scale sample entropies (iMSE) of the current window, forming a set of intrinsic multi-scale sample entropies (iMSE set) of the current window.

[0099] In step S403, since the intrinsic mode function is combined with the multiscale sample entropy, the result obtained is called intrinsic multiscale sample entropy (iMSE). The calculation method (such as calculation formula, etc.) of the multiscale sample entropy is a conventional method in the art, and can be calculated with reference to existing literature, such as Costa M, etc., Multiscale Entropy Analysis of Biological Signals [J], Physical Review E—Statistical, Nonlinear, and Soft Matter Physics, 2005, 71 (2): 021906; and Richman JS, Lake DE, Moorman JR. Sample entropy [M] / / Methods in enzymology. Academic Press, 2004, 384: 172-184.

[0100] Continuing with the specific data as an example, in step S300, a white noise sequence with an amplitude ratio of 10% is set to be added, and 9 E-IMFs are obtained by EEMD, such as Figure 4As shown. Then select E-IMF1, E-IMF2, E-IMF3, E-IMF4, E-IMF5, E-IMF6, E-IMF7, E-IMF8. It is worth understanding that if only 7 E-IMFs are obtained in some cases, all of them can be selected, not necessarily 8.

[0101] Formula (1) is used to calculate the partial component sums of these 8 E-IMFs, and 8 partial component sums are obtained respectively. In formula (1), c j (t) is the collective intrinsic mode function (E-IMF), t represents 6000 data, N = 8, represents the total number of E-IMFs, n represents the nth E-IMF, which is 1, 2, ..., 8, x n (t) represents the sum from the 1st E-IMF to the nth E-IMF. The sum of each component obtained by formula (1) still contains 6000 data.

[0102] Then, the multi-scale sample entropy of each partial component and is calculated respectively. The 8 partial components and 8 intrinsic multi-scale sample entropies (iMSE) are obtained, namely 1MSE, 2MSE, 3MSE, 4MSE, 5MSE, 6MSE, 7MSE, and 8MSE; they constitute the iMSE set of the current window.

[0103] S500: Calculation step of the set intrinsic multi-scale sample entropy (eiMSE calculation step): Set multiple white noise sequences with different amplitude ratios, apply them to step S300 respectively (wherein, when step S303 is repeated, the same amplitude ratio is used), and perform step S400 to obtain multiple rounds of iMSE sets. The set average of the iMSE sets under multiple rounds is taken as the result to obtain multiple sets of intrinsic multi-scale sample entropies (eiMSE) of the current window, forming a set of the set intrinsic multi-scale sample entropy of the current window (eiMSE set).

[0104] In one embodiment, the amplitude ratio of white noise is set to 0.05-0.2, and 6-16 set trials are taken in 0.05-0.2. Steps S300 and S400 are performed for each set trial respectively, so as to obtain the iMSE sets of each round respectively, and each iMSE set contains multiple iMSEs. For white noise under different amplitudes, there will be some differences in the E-IMFs obtained by EEMD, and there will be some differences in the corresponding iMSE sets; the iMSE sets under multiple trials are averaged to obtain an eiMSE set containing multiple eiMSEs. Its advantage is that this is a set average, which is the most likely feature in terms of probability and a more stable result.

[0105] Both the intrinsic multi-scale sample entropy and the set intrinsic multi-scale sample entropy can be represented by a curve distribution diagram (with the scale of the multi-scale sample entropy as the horizontal axis and the entropy value as the vertical axis) and a two-dimensional diagram.

[0106] Let's continue with the specific data as an example. Take 7 white noise amplitude ratios between 0.05-0.2, so that 7 set trials can be obtained, which can be used for 7 rounds of calculation. In the first round, set the amplitude of the white noise sequence to 0.1, enter step S300, and obtain the E-IMFs set of the first round (including multiple loop iterations of 0.1 white noise) through steps S301-S304, and then enter step S400, and obtain the iMSE set of the first round through steps S401-S403. Assuming that it contains 8 iMSEs (i.e. 1MSE, 2MSE, 3MSE, 4MSE, 5MSE, 6MSE, 7MSE, 8MSE)... and so on... the iMSE sets of all 7 rounds can be obtained.

[0107] The 1MSE in the iMSE sets of rounds 1 to 7 are averaged to obtain the first set of intrinsic multi-scale sample entropy e1 MSE, the 2MSE in the iMSE sets of each round are averaged to obtain the second set of intrinsic multi-scale sample entropy e2MSE,…, the 8MSE in the iMSE sets of each round are averaged to obtain the eighth set of intrinsic multi-scale sample entropy e8MSE, thereby obtaining eight sets of intrinsic multi-scale sample entropy (eiMSE), which constitute the eiMSE set.

[0108] Figure 5 and Figure 6 For the current window, when the amplitude ratio of white noise is 0.1, an example of the iMSE set of the first round is obtained, which contains 8 iMSEs; Figure 5 It is displayed in the form of a curve distribution diagram, where the horizontal axis is the scale and the vertical axis is the entropy value; Figure 6 It is displayed in a two-dimensional graph. Figure 5 Correspondingly, the horizontal axis in the figure is the scale, and the vertical axis corresponds to each iMSE. Different colors represent different entropy values ​​to form a two-dimensional graph.

[0109] Figure 7 and Figure 8 This is an example of the eiMSE set (the ensemble average of the iMSE sets of each round) calculated for the current window when the amplitude ratio of 7 white noises is taken between 0.05 and 0.2, which also contains 8 eiMSEs; Figure 7 It is displayed in the form of a curve distribution diagram, where the horizontal axis is the scale and the vertical axis is the entropy value; Figure 8 It is displayed in a two-dimensional graph. Figure 7Correspondingly, the horizontal axis in the figure is the scale, and the vertical axis corresponds to each eiMSE. Different colors represent different entropy values ​​to form a two-dimensional graph.

[0110] Although the above steps describe how to obtain the intrinsic multiscale sample entropy (iMSE) and its set of the current window, and the set intrinsic multiscale sample entropy (eiMSE) and its set; it is obvious that the above method can be used to obtain the iMSE and its set, and eiMSE and its set of any window respectively; wherein each iMSE set includes multiple iMSEs, and each eiMSE set includes multiple eiMSEs, for example, 8.

[0111] The output step of S600 data is as follows: multiple intrinsic multi-scale sample entropies (iMSE) and / or multiple set intrinsic multi-scale sample entropies (eiMSE) of each window are sorted and output, and the results are displayed on the monitor and archived, which can be used as a depth guide for clinical anesthesia.

[0112] In a specific implementation, each iMSE in the iMSE set of each window is output to obtain a two-dimensional graph of each iMSE on the time axis (i.e., an iMSE monitoring graph). The number of graphs corresponds to the number of iMSEs, and these graphs can be used as a monitoring guide for clinical anesthesia.

[0113] In a specific implementation, each eiMSE in the eiMSE set of each window is output to obtain a two-dimensional graph of each eiMSE on the time axis (i.e., eiMSE monitoring graph). The number of graphs corresponds to the number of eiMSEs, and these graphs can be used as a monitoring guide for clinical anesthesia.

[0114] In one embodiment, a linear regression model is used to perform linear regression fitting on multiple iMSEs of each window to obtain an anesthetic depth index for iMSE as an output. Further, an anesthetic depth index curve (iMSE monitoring curve) for each window is output as a monitoring guide for clinical anesthesia.

[0115] In one embodiment, a linear regression model is used to perform linear regression fitting on multiple eiMSEs of each window to obtain an anesthesia depth index related to eiMSE as an output. Further, an anesthesia depth index curve (eiMSE monitoring curve) for each window is output as a monitoring guide for clinical anesthesia.

[0116] It is worth understanding that since the calculation of the set intrinsic multi-scale sample entropy (eiMSE) is a huge project, it has many parameters and occupies a high amount of computer memory. Therefore, when calculating, if the set intrinsic multi-scale sample entropy cannot be calculated quickly (affecting the output time of the monitoring parameters), then the intrinsic multi-scale sample entropy (iMSE) that consumes less time can be preferentially selected. In addition, the two-dimensional monitoring map can be generated by methods such as pcolor maps and contour maps. When using the iMSE monitoring curve, it can replace the iMSE monitoring map because it is more intuitive; similarly, the eiMSE monitoring curve can replace the eiMSE monitoring map. In a preferred embodiment, the monitoring map and the monitoring curve are output simultaneously for monitoring guidance.

[0117] In the specific operation, in step S100, the first original data obtained from the first window is processed in sequence by steps S200, S300, S400 and S500, and the iMSE set and eiMSE set of the first window can be obtained, which respectively include multiple iMSEs and eiMSEs, assuming that they are 8; the second original data obtained from the second window is processed in sequence by steps S200, S300, S400 and S500, and the iMSE set and eiMSE set of the second window can be obtained, which also include multiple iMSEs and eiMSEs, assuming that they are 8; the third original data obtained from the third window is processed in sequence by steps S200, S300, S400 and S500, and the iMSE set and eiMSE set of the third window can be obtained, which also include multiple iMSEs and eiMSEs, assuming that they are 8. . . . According to this calculation method, during the data collection process, the newly collected window data can be continuously calculated to obtain the respective iMSE sets and eiMSE sets.

[0118] In this example, the eiMSE set is used as the illustration object: each eiMSE (e1 MSE, e2MSE, e3MSE, …, e8MSE) in the eiMSE set of each window is arranged in chronological order according to the data acquisition time, so as to obtain the eiMSE monitoring graph output in sequence. Then, each eiMSE in the eiMSE set of each window is input into the linear regression model, and the anesthesia depth index corresponding to each window can be obtained, thereby obtaining the eiMSE monitoring curve (including all eiMSE parameters). Specifically, the change trend of e1 MSE to e8MSE output in chronological order for all windows (the entire anesthesia surgery process) can be displayed in an eiMSE monitoring graph, refer to Fig. 9 ; You can also display the eiMSE monitoring curves of all windows (the entire anesthesia surgery process) in chronological order in one curve, refer to Fig.10 .

[0119] Combine the following Figure 9-11The eiMSE monitoring diagram and eiMSE monitoring curve of an anesthesia surgery process are described in more detail to more clearly describe and understand the present application.

[0120] Fig. 9 The original data collected during the entire anesthesia surgery process (within the time axis of 0-8700s, as the horizontal axis) are shown, and the eiMSE monitoring diagrams obtained by the monitoring method described in this embodiment respectively correspond to e1 MSE to e8 MSE in the eiMSE set of each window.

[0121] by Fig. 9 Take the e2MSE in the upper middle as an example, where the horizontal axis corresponds to the operation time of 0-8700s, from the beginning to the end of the operation; the vertical axis corresponds to the e2MSE value of each time window within the scale of 1-25 (marked with different colors); each e2MSE value is obtained by using the original data of the current time window through steps S100-S500. It can be seen from the figure that the e2MSE value of the front section of the monitoring chart is relatively low (the color tends to blue and sky blue with smaller entropy values), and fluctuates greatly, indicating that it is in the stage of surgical preparation and induction; then the e2MSE value of the middle section is relatively stable (the color tends to red and yellow with larger entropy values), indicating that it is in the middle stage of the operation; the final section is the late stage of the operation, and the patient's consciousness gradually awakens, so the e2MSE value decreases again, and fluctuates greatly.

[0122] Fig.10 The figure shows the original data collected within the time axis of 0-8700s during the entire anesthesia surgery process, and the eiMSE monitoring curve obtained by the monitoring method described in this embodiment. Among them, the horizontal axis of the monitoring curve is the time axis (surgery time from 0 to 8700s), and the vertical axis is the anesthesia depth index. That is, the anesthesia depth index corresponding to each second in 8700s is the result obtained after calculating 6000 original data through the steps described above. It can be seen from the figure that the anesthesia depth index value in the front section of the monitoring curve is relatively high and fluctuates greatly, indicating that it is in the surgical preparation and induction stage; then the anesthesia depth index value in the middle section is relatively stable, indicating that it is in the middle stage of the operation; the final section is the late stage of the operation, and the patient's consciousness gradually becomes clear, so the anesthesia depth index value increases again, and fluctuates greatly. It can be understood that, Fig.10 The monitoring curve shown is Fig. 9 The monitoring graph in is a completely corresponding relationship. Fig. 9 The result of linear regression of the data; the two can be used together to assist each other.

[0123] Fig.10 and Fig.11The EEG signal from the same source, the eiMSE monitoring curve and the BIS curve obtained by the monitoring method of the present application, the characteristics displayed by the monitoring curve obtained by the present application are the same as those of the corresponding BIS curve ( Fig.11 ) trends are close, indicating that the monitoring method of the present application is reliable.

[0124] Comparative analysis:

[0125] This content compares the anesthesia depth monitoring method based on intrinsic multi-scale sample entropy provided by this application with the BIS anesthesia monitoring method in the prior art.

[0126] Both monitoring methods use the same raw data, Fig.11 The monitoring curve obtained by the BIS monitoring method (hereinafter referred to as the BIS monitoring curve) and the corresponding pre-processed EEG data, Fig.11 In the figure, the horizontal axis is time (unit s), and the vertical axis is EEG data (unit μV), which is used to represent the value of EEG data. Only the trend of BIS monitoring curve over time is shown in the figure; Fig.10 eiMSE monitoring curve obtained in this implementation manner.

[0127] Fig.11 The recorded EEG signals and their BIS values ​​are shown, where the induced sharp changes are obvious. The data were somewhat disturbed during the surgery, which caused a great disturbance in the BIS score. The BIS score also showed that the patient's gradual recovery time was prolonged in the later stage of the surgery.

[0128] Fig.10 It is an index of anesthesia depth based on the linear regression of the ensemble intrinsic multiscale sample entropy (eiMSE). It also has obvious sharp changes in the induction phase, but is more stable during surgery and also has sharp changes in the recovery phase. Therefore, it can provide better monitoring guidance for doctors.

[0129] This implementation provides new tools needed to quantitatively study anesthesia. We believe that with the theoretical breakthrough of intrinsic multiscale sample entropy, we have the new tools needed to quantitatively study anesthesia, which will take the current state-of-the-art technology a big step forward and meet many urgent needs.

[0130] The second embodiment of the present application provides an anesthesia depth monitoring system based on intrinsic multi-scale sample entropy (hereinafter referred to as a monitoring system), which is used to implement the monitoring method described in any of the above embodiments. The monitoring system includes: a data acquisition module and a data analysis module, wherein:

[0131] The data acquisition module is configured to collect EEG data using electrodes to obtain raw data;

[0132] The data analysis module includes a windowing module, a data preprocessing module, an EEMD decomposition module, an iMSE calculation module and a result output module; wherein,

[0133] A windowing module is configured to collect the EEG data in a windowed manner to obtain raw data of each window;

[0134] A data preprocessing module is configured to filter, remove outliers and remove electrical noise from the collected raw data to obtain a data time series to be analyzed;

[0135] An EEMD decomposition module is configured to adaptively decompose the data time series to be analyzed in the current window into multiple E-IMFs using an EEMD analysis method;

[0136] The iMSE calculation module is configured to select the first N E-IMFs from the multiple E-IMFs, calculate the partial component sum using formula (1), obtain N partial component sums, and then calculate the multi-scale sample entropy of each partial component sum, thereby obtaining N iMSEs of the current window to form an iMSE set; the formula (1) refers to the first implementation mode;

[0137] The result output module is configured to sort and output the multiple iMSEs of each window.

[0138] In some embodiments, the data analysis module further includes an eiMSE calculation module, which is configured to set a plurality of white noise sequences with different amplitude ratios, apply them to the EEMD decomposition module and the iMSE calculation module respectively, and obtain multiple rounds of iMSE sets; taking the set average of the iMSE sets under multiple rounds as the result, obtaining multiple sets of intrinsic multi-scale sample entropy (eiMSE) of the current window, and forming an eiMSE set;

[0139] The result output module is further configured to sort and output multiple eiMSEs of each window, which can replace the output of iMSE.

[0140] In some embodiments, the monitoring system may further include:

[0141] a data archiving module configured to store data for future reference;

[0142] The data display module is configured to display the output results of the result output module on a monitor, which can be used as a depth guide for clinical anesthesia.

[0143] It is worth understanding that the technical features described in the monitoring method can also be reasonably applied to the monitoring system of this embodiment, and the descriptions that have been made will not be repeated. It should be noted that the above-mentioned modules can be functional modules or program modules, which can be implemented by software or hardware. For modules implemented by hardware, the above-mentioned modules can be located in the same processor, or the above-mentioned modules can be located in different processors in any combination.

[0144] The third embodiment of the present application provides an application of the monitoring method described in any of the above embodiments, which can be used for medical monitoring. The medical monitoring includes but is not limited to anesthesia depth monitoring.

[0145] A fourth embodiment of the present application provides a computer device, which may include a processor 101 and a memory 102 storing computer program instructions, such as Fig.12 shown.

[0146] Specifically, the processor 101 may include a central processing unit (CPU), or an application specific integrated circuit (ASIC), or may be configured to implement one or more integrated circuits of the embodiments of the present application. The memory 102 may include a large capacity memory for data or instructions. By way of example and not limitation, the memory 102 may include a hard disk drive (HDD), a floppy disk drive, a solid state drive (SSD), a flash memory, an optical disk, a magneto-optical disk, a magnetic tape, or a universal serial bus (USB) drive, or a combination of two or more of these. Where appropriate, the memory 102 may include a removable or non-removable (or fixed) medium. Where appropriate, the memory 102 may be inside or outside a data processing device.

[0147] The memory 102 may be used to store or cache various data files that need to be processed and / or used for communication, as well as possible computer program instructions executed by the processor 101 .

[0148] The processor 101 implements the monitoring method described in any of the above embodiments by reading and executing the computer program instructions stored in the memory 102.

[0149] In some implementations, the computer device may further include a communication interface 103 and a bus 104. Fig.12As shown, the processor 101, the memory 102, and the communication interface 103 are connected through a bus 104 and communicate with each other. The communication interface 103 is used to implement the communication between the modules, devices, units and / or equipment in the implementation mode of the present application. The communication interface 103 can also implement data communication with other components.

[0150] A fifth embodiment of the present application provides a computer-readable storage medium having computer program instructions stored thereon; when the computer program instructions are executed by a processor, the monitoring method described in any of the above embodiments is implemented.

[0151] The technical features of the above-described embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above-described embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0152] The above-described embodiments only express several embodiments of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be construed as limiting the scope of the application. It should be noted that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all fall within the scope of protection of the present application. Therefore, the scope of protection of the patent of the present application shall be subject to the attached claims.

Claims

1. A monitoring method based on intrinsic multi-scale sample entropy, characterized in that: The following steps are involved: Steps to obtain raw data: collect data as raw data; Data preprocessing step: preprocess the acquired raw data to obtain the data time series to be analyzed; The steps of applying EEMD to the data time series are as follows: the data time series is processed by the EEMD analysis method, and is adaptively decomposed into a plurality of ensemble intrinsic mode functions (E-IMFs); The calculation steps of intrinsic multi-scale sample entropy (iMSE calculation steps): Formula (1) is used to calculate the partial component sum of N set intrinsic mode functions to obtain N partial component sums; Among them, c j (t) is the collective intrinsic mode function (E-IMF), t represents the number of data, N represents the total number of E-IMFs, which is a positive integer, n represents the nth E-IMF, and x n (t) represents the sum from the first E-IMF to the nth E-IMF; Then, the multi-scale sample entropy of each partial component and is calculated to obtain N intrinsic multi-scale sample entropies (iMSE), which constitute a set of intrinsic multi-scale sample entropies (iMSE set); Wherein, at least a part of the iMSE set is used as the output of the monitoring method.

2. The monitoring method according to claim 1, characterized in that: The method further comprises a step of calculating the set intrinsic multi-scale sample entropy (the step of calculating eiMSE): setting a plurality of white noise sequences with different amplitude ratios, applying them respectively to the step of applying the EEMD to the data time series, and performing the step of calculating the iMSE to obtain multiple rounds of iMSE sets; taking the set average of the iMSE sets under multiple rounds as the result, obtaining multiple sets of intrinsic multi-scale sample entropy (eiMSE); Wherein, at least a portion of the plurality of eiMSEs is used as output of the monitoring method.

3. The monitoring method according to claim 1 or 2, characterized in that: In the step of obtaining the raw data, a sliding window is used to collect data, and the data of each window is used as the raw data; the output of each window can be obtained by using the monitoring method; The monitoring method also includes a data output step, wherein the output of each window is displayed as a result on a monitor and archived as a monitoring guide.

4. The monitoring method according to claim 3, characterized in that: In the step of outputting the data, at least one of the following methods is selected: (1) Outputting all or part of the multiple iMSEs of each window to obtain an iMSE monitoring diagram of each iMSE on the time axis for monitoring guidance; (2) A linear regression model is used to perform linear regression fitting on multiple iMSEs of each window to obtain an index related to the iMSE, and the exponential curve of each window is output as an eiMSE monitoring curve for monitoring guidance; (3) Outputting all or part of the multiple eiMSEs of each window to obtain an eiMSE monitoring graph of each eiMSE on the time axis for monitoring and guidance; (4) A linear regression model is used to perform linear regression fitting on the multiple eiMSEs of each window to obtain an index related to eiMSE, and the exponential curve related to each window is output as an eiMSE monitoring curve for monitoring guidance.

5. The monitoring method according to claim 1, 2 or 4, characterized in that: In the step of acquiring raw data, the sampling frequency of the data is not less than 100 Hz; in the step of preprocessing the data, the preprocessing includes filtering, outlier removal and electrical noise removal.

6. The monitoring method according to claim 1, 2 or 4, characterized in that: The step of applying EEMD to the data time series specifically includes: Add white noise: add a white noise sequence to the data time series to be analyzed; the amplitude ratio of the white noise is any value between 0.05 and 0.2; Decompose IMFs using EMD: Adaptively decompose the data time series with white noise into multiple initial intrinsic mode functions (IMFs) using EMD; Repeat the steps of adding white noise and decomposing IMFs using EMD, but use a newly randomly generated white noise sequence with the same amplitude ratio each time; thereby obtaining multiple batches with multiple IMFs respectively; The ensemble average of multiple IMFs of multiple batches is taken as the decomposition result to obtain the multiple E-IMFs.

7. The monitoring method according to claim 6, characterized in that: In the step of applying the EEMD to the data time series, the step of adding white noise and the step of decomposing IMFs using EMD are repeated 10-20 times to obtain 10-20 batches; in the step of calculating the iMSE, a maximum of 8 of the multiple E-IMFs are selected for calculating the sum of partial components; that is, N is a positive integer less than or equal to 8; in the step of calculating the eiMSE, the amplitude ratio of the white noise is set to 0.05-0.2, and 6-16 sets are taken for trial as the white noise sequences with multiple different amplitude ratios.

8. A monitoring system based on intrinsic multi-scale sample entropy, characterized in that: It includes data acquisition module and data analysis module, among which, The data acquisition module is configured to collect data, thereby obtaining raw data; The data analysis module includes a windowing module, a data preprocessing module, an EEMD decomposition module, an iMSE calculation module and a result output module; wherein, The windowing module is configured to collect data in a windowed manner to obtain raw data of each window; The data preprocessing module is configured to filter, remove outliers and remove electrical noise on the collected raw data to obtain a data time series to be analyzed; The EEMD decomposition module is configured to adaptively decompose the data time series to be analyzed in the current window into multiple E-IMFs using the EEMD analysis method; The iMSE calculation module is configured to select the first N E-IMFs from the plurality of E-IMFs, calculate the partial component sums using formula (1), and obtain N partial component sums. Among them, c j (t) is the E-IMF, t represents the number of data, N represents the total number of E-IMFs, which is a positive integer, n represents the nth E-IMF, x n (t) represents the sum from the first E-IMF to the nth E-IMF; Then, the multi-scale sample entropy of each partial component and the iMSE set of the current window is obtained; The result output module is configured to sort and output the multiple iMSEs of each window.

9. The monitoring system according to claim 8, characterized in that: It further includes an eiMSE calculation module, which is configured to set multiple white noise sequences with different amplitude ratios, apply them to the EEMD decomposition module and the iMSE calculation module respectively, and obtain multiple rounds of iMSE sets; take the set average of the iMSE sets under multiple rounds as the result, and obtain multiple eiMSEs of the current window; the result output module is configured to sort and output the multiple eiMSEs of each window.

10. The monitoring system according to claim 8 or 9, characterized in that: Also includes: a data archiving module configured to store data for future reference; The data display module is configured to display the output result of the result output module on a monitor as a monitoring guide.

11. An application of the monitoring method based on intrinsic multi-scale sample entropy according to any one of claims 1 to 7, wherein the monitoring method can be used for medical monitoring, including anesthesia depth monitoring.

12. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the monitoring method based on intrinsic multi-scale sample entropy as described in any one of claims 1 to 7 when executing the computer program.

13. A computer-readable storage medium having a computer program stored thereon, wherein when the program is executed by a processor, the monitoring method based on intrinsic multi-scale sample entropy as described in any one of claims 1 to 7 is implemented.

Images