Patient cardiovascular and cerebrovascular disease data processing method and system

Abstract

The invention provides a patient cardiovascular and cerebrovascular disease data processing method and system, and the method comprises the steps: extracting an electroencephalogram signal data set of a corresponding frequency band of a patient according to the frequency bands of real electroencephalogram signals EEG corresponding to a plurality of diseases, the electroencephalogram signal data set comprising electroencephalogram signals EEG in a plurality of time periods; performing denoising processing on the EEG in the EEG data set to obtain a denoised EEG; performing dispersion evaluation processing on the de-noised EEG to obtain dispersion evaluation of each segment of de-noised EEG; extracting the de-noised EEG of which the dispersion evaluation is smaller than a preset threshold as an EEG for training; according to the method, the EEG for training is subjected to feature extraction and compared with the real EEG of the corresponding disease so as to verify the EEG in the EEG data set, the authenticity of the data is verified on the aspect of an algorithm, and time waste caused by the fact that a large amount of time is adopted to verify the reliability of the data is avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a method and system for processing data of cardiovascular and cerebrovascular diseases of patients. Background Art

[0002] Cardiovascular and cerebrovascular diseases refer to diseases caused by cardiac ischemia or bleeding due to hardening of the heart and arteries, including coronary heart disease, angina pectoris, hypertension, arrhythmia, myocardial infarction, cerebral infarction, etc. Cardiovascular and cerebrovascular diseases have always been known as the number one killer threatening human health, and hypertension, hyperlipidemia, coronary heart disease, and diabetes have become the "plague of the times."

[0003] With the improvement of social living standards and the accelerated pace of work, cardiovascular and cerebrovascular diseases have become the number one killer threatening the health and longevity of middle-aged and elderly people.

[0004] In the existing technology, the EEG signals of clinical responses to similar types of cardiovascular and cerebrovascular diseases (such as cerebral infarction, cerebral embolism, stroke, etc.) are similar. It is impossible to accurately judge such diseases clinically. More reliable analysis at the data level is needed to approach the actual type of disease. However, at the data level, since human EEG signals are occasionally irregular, it is not guaranteed to be correct in judging the actual type of disease. It often takes a lot of time to verify the reliability of the data, resulting in a waste of time. Summary of the invention

[0005] In view of the deficiencies in the prior art, the purpose of the present invention is to provide a method for processing patient cardiovascular and cerebrovascular disease data, aiming to solve the technical problems mentioned in the background technology.

[0006] In order to achieve the above object, the present invention is implemented by the following technical solutions:

[0007] A method for processing patient cardiovascular and cerebrovascular disease data, comprising the following steps:

[0008] S10, extracting an EEG signal data set of a frequency band corresponding to the patient according to the frequency bands of the real EEG signals corresponding to the various diseases, wherein the EEG signal data set includes the EEG signals in multiple time periods;

[0009] S20, performing denoising on the EEG signal in the EEG signal data set to obtain a denoised EEG signal;

[0010] S30, performing a discreteness evaluation process on the denoised EEG signal to obtain a discreteness evaluation of each segment of the denoised EEG signal;

[0011] S40, extracting the denoised EEG signal with a discreteness evaluation less than a preset threshold as a training EEG signal;

[0012] S50, extracting features from the training EEG signal, and comparing it with the real EEG signal of the corresponding disease, so as to verify the EEG signal in the EEG signal data set.

[0013] According to one aspect of the above technical solution, the specific steps of step S20 include:

[0014] Using the Daubechies5 wavelength function to perform a three-layer decomposition on the EEG signal in the EEG signal data set to remove power frequency interference and electromyographic noise in the EEG signal;

[0015] Using a 600ms and 200ms sliding average filter to remove the T wave, P wave and QRS complex wave of the EEG signal to obtain a processed EEG signal;

[0016] The processed EEG signal is used as a baseline, and the processed EEG signal is removed from the EEG signal to obtain a denoised EEG signal.

[0017] According to one aspect of the above technical solution, the specific steps of step S30 include:

[0018] Performing a first permutation entropy transformation on each segment of the denoised EEG signal to obtain a first permutation entropy of the denoised EEG signal;

[0019] Performing a second permutation entropy transformation on each segment of the denoised EEG signal to obtain a second permutation entropy of the denoised EEG signal;

[0020] The first permutation entropy and the second permutation entropy are selected according to the frequency of the EEG signal after denoising described in the paragraph to obtain a dispersion evaluation.

[0021] According to one aspect of the above technical solution, the specific steps of performing a first permutation entropy conversion on each segment of the denoised EEG signal to obtain the first permutation entropy of the denoised EEG signal include:

[0022] The denoised EEG signals of each segment are reconstructed in time and space to obtain:

[0023] X(q)={x(q),x(q+γ),....,x(q+(n-1)γ)}

[0024] The sampling time of each segment of the denoised EEG signal is consistent, γ is the delay time, n is the embedding dimension, and X(q) represents the first reconstructed component of the qth segment of the denoised EEG signal;

[0025] Arrange the first reconstructed components of each segment of the denoised EEG signal in ascending order to obtain:

[0026] x(n+(p1-1)γ)≤x(q+(p2-1)γ)≤…≤x(q+(p k -1)γ)

[0027] If the values ​​of the first reconstructed components do not have the same value, then p k represents x(n+(p1-1)γ), if the values ​​of the first reconstruction components are the same, then p ki The values ​​are arranged in ascending order, that is, when p k1 <p k2 When x(q+(p k1 -1)γ)≤x(q+(p k2 -1)γ);

[0028] Symbolize each segment of the first reconstructed component to obtain a first symbolized sequence:

[0029] E(l)=(p1,p1,…,p m )

[0030] Where, l = 1, 2, ..., k, and k is less than or equal to m! ;

[0031] The first symbolized sequence is defined as the first permutation entropy in the form of Shannon information entropy:

[0032]

[0033] Among them, when r j =1 / m!, Ur(m) reaches the maximum value In(m!);

[0034] The first permutation entropy is normalized to reflect the discreteness of the denoised EEG signal in this segment.

[0035] According to one aspect of the above technical solution, the specific steps of performing a second permutation entropy conversion on each segment of the denoised EEG signal to obtain the second permutation entropy of the denoised EEG signal include:

[0036] The denoised EEG signals of each segment are reconstructed in time and space to obtain:

[0037] X′(q)={x′(q), x′(q+γ), ..., x′(q+(n-1)γ)}, wherein the sampling time of each segment of the denoised EEG signal is consistent, γ is the delay time, n is the embedding dimension, and X′(q) represents the second reconstructed component of the qth segment of the denoised EEG signal;

[0038] Arrange the second reconstructed components of each segment of the denoised EEG signal in ascending order to obtain:

[0039] x′(n+(p1-1)γ)≤x′(q+(p2-1)γ)≤…≤x′(q+(p k -1)γ)

[0040] If the values ​​of the second reconstruction components do not have the same value, then p k represents x′(n+(p1-1)γ). If there is a second reconstruction component with the same value, then p ki The value represents the consistent second reconstruction component, that is, x′(n+(p k1 -1)γ)=x′(q+(p k2 -1)γ);

[0041] Each segment of the second reconstructed component is symbolized to obtain a second symbolized sequence:

[0042] E′(l)=(p1,p2,…,p m )

[0043] Where, l = 1, 2, ..., k, and k is less than or equal to m! ;

[0044] The second symbolized sequence is defined as the second permutation entropy in the form of Shannon information entropy:

[0045]

[0046] Among them, when r′ j =1 / m!, U′ r (m) reaches the maximum value In(m!);

[0047] The second permutation entropy is normalized to reflect the discreteness of the denoised EEG signal in this segment.

[0048] According to one aspect of the above technical solution, the specific steps of selecting the first permutation entropy and the second permutation entropy according to the frequency of the EEG signal after denoising described in the paragraph to obtain the discreteness evaluation include:

[0049] Decomposing the denoised EEG signal in X layers, and reconstructing the scale frequency band of each layer after decomposition to obtain reconstructed signals of X+1 frequency bands;

[0050] The dispersion analysis is performed by using the first permutation entropy and the second permutation entropy, as follows:

[0051] Selecting several continuous points of denoised EEG data as basic data, and selecting movement data according to frequency bands to arrange and group the basic data;

[0052] Obtain the data in each permutation group, and calculate the average data of the data in each permutation group as the permutation entropy value;

[0053] Analyze the permutation entropy value to select the frequency band with the greatest discrimination, use the frequency band with the greatest discrimination as the discrimination frequency band, and compare and analyze the discrimination frequency band by using the first permutation entropy and the second permutation entropy to obtain the influence of different basic data lengths, different embedding dimensions and different delay times on the discrimination effect;

[0054] According to the influence of the distinction effect, the first permutation entropy or the second permutation entropy is selected to process the denoised EEG signal of the segment to obtain a discreteness evaluation.

[0055] According to one aspect of the above technical solution, the specific steps of step S50 include:

[0056] The training EEG signal is defined as And perform discrete Fourier transform to obtain multiple frequency spectra of the training EEG signal:

[0057]

[0058] in, is the frequency domain representation of the training EEG signal, parameter w=0,1,2,……,W-1; W represents the signal length of the training EEG signal, k=0,1,2,……,W / 2-1, k is the sampling point within the sampling frequency range, and the sampling frequency range is F s is the sampling frequency;

[0059] The multiple spectra are divided into L consecutive segments denoted as R l , and denote the boundary of each of the paragraphs as Ω l , that is, R l =[Ω l-1 ,Ω l ];

[0060] Where, l = 0, 1, 2, ..., L;

[0061] Applying a scaling function and a wavelet function to filter and decompose the signal on each of the frequency spectrums to obtain an EEG signal for comparison, so as to extract features of the EEG signal for training;

[0062] The comparison EEG signal is compared with the real EEG signal of the corresponding disease in the database to verify the EEG signal in the EEG signal data set.

[0063] According to one aspect of the above technical solution, the specific steps of applying a scaling function and a wavelet function to filter the decomposed signal on each of the paragraphs to obtain an EEG signal for comparison, and extracting features from the EEG signal for training include:

[0064] Applying the scaling function φ(Ω) on each of the frequency spectra creates a filter and decomposes the signal to obtain a first verification signal:

[0065]

[0066] Wherein, P represents the amplitude of the current spectrum;

[0067] Apply the wavelet function to each of the spectra Create the filter and decompose the signal to get the second verification signal:

[0068]

[0069] The first verification signal and the second verification signal are integrated to obtain an EEG electroencephalogram (EEG) signal for comparison, and then feature extraction is performed on the EEG electroencephalogram (EEG) signal for training.

[0070] The present invention also provides a patient cardiovascular and cerebrovascular disease data processing system, comprising:

[0071] Extraction module: used to extract EEG data sets of corresponding frequency bands of patients according to the frequency bands of real EEG signals corresponding to various diseases, wherein the EEG data sets include the EEG signals in multiple time periods;

[0072] Denoising module: used for denoising the EEG signal in the EEG signal data set to obtain the denoised EEG signal;

[0073] The denoising module is specifically used to: perform three-layer decomposition of the EEG signal in the EEG signal data set using the Daubechies5 wavelength function to remove power frequency interference and electromyographic noise in the EEG signal;

[0074] Using a 600ms and 200ms sliding average filter to remove the T wave, P wave and QRS complex wave of the EEG signal to obtain a processed EEG signal;

[0075] Taking the processed EEG as a baseline, removing the processed EEG from the EEG to obtain a denoised EEG;

[0076] Evaluation module: used for performing discreteness evaluation processing on the denoised EEG signal to obtain discreteness evaluation of each segment of the denoised EEG signal;

[0077] The evaluation module is specifically used to: perform a first permutation entropy conversion on each segment of the denoised EEG signal to obtain a first permutation entropy of the denoised EEG signal;

[0078] Performing a second permutation entropy transformation on each segment of the denoised EEG signal to obtain a second permutation entropy of the denoised EEG signal;

[0079] The first permutation entropy and the second permutation entropy are selected according to the frequency of the EEG signal after denoising described in the paragraph to obtain a dispersion evaluation;

[0080] The specific steps of performing the first permutation entropy conversion on each segment of the denoised EEG signal to obtain the first permutation entropy of the denoised EEG signal include:

[0081] The denoised EEG signals of each segment are reconstructed in time and space to obtain:

[0082] X(q)={x(q),x(q+γ),…,r(q+(n-1)γ)}

[0083] The sampling time of each segment of the denoised EEG signal is consistent, γ is the delay time, n is the embedding dimension, and X(q) represents the first reconstructed component of the qth segment of the denoised EEG signal;

[0084] Arrange the first reconstructed components of each segment of the denoised EEG signal in ascending order to obtain:

[0085] x(n+(p1-1)γ)≤x(q+(p2-1)γ)≤…≤x(q+(p k -1)γ)

[0086] If the values ​​of the first reconstructed components do not have the same value, then p k represents x(n+(p1-1)γ), if the values ​​of the first reconstruction components are the same, then p kiThe values ​​are arranged in ascending order, that is, when p k1 <p k2 When x(q+(p k1 -1)γ)≤x(q+(p k2 -1)γ);

[0087] Symbolize each segment of the first reconstructed component to obtain a first symbolized sequence:

[0088] E(l)=(p1,p2,…,p m )

[0089] Where, l = 1, 2, ..., k, and k is less than or equal to m! ;

[0090] The first symbolized sequence is defined as the first permutation entropy in the form of Shannon information entropy:

[0091]

[0092] Among them, when r j =1 / m!, Ur(m) reaches the maximum value In(m!);

[0093] Normalizing the first permutation entropy to reflect the discreteness of the EEG signal after denoising;

[0094] The specific steps of performing a second permutation entropy conversion on each segment of the denoised EEG signal to obtain the second permutation entropy of the denoised EEG signal include:

[0095] The denoised EEG signals of each segment are reconstructed in time and space to obtain:

[0096] X′(q)={x′(q),x′(q+γ),...,x′(q+(n-1)γ)}

[0097] Wherein, the sampling time of each segment of the denoised EEG signal is consistent, γ is the delay time, n is the embedding dimension, and X'(q) represents the second reconstructed component of the qth segment of the denoised EEG signal;

[0098] Arrange the second reconstructed components of each segment of the denoised EEG signal in ascending order to obtain:

[0099] x′(n+(p1-1)γ)<x′(q+(p2-1)γ)<…<x′(q+(p k -1)γ)

[0100] If the values ​​of the second reconstruction components do not have the same value, then p krepresents x′(n+(p1-1)γ). If there is a second reconstruction component with the same value, then p ki The value represents the consistent second reconstruction component, that is, x′(n+(p k1 -1)γ)=x′(q+p k2 -1)γ);

[0101] Each segment of the second reconstructed component is symbolized to obtain a second symbolized sequence:

[0102] E′(l)=(p1,p2,…,p m )

[0103] Where, l = 1, 2, ..., k, and k is less than or equal to m! ;

[0104] The second symbolized sequence is defined as the second permutation entropy in the form of Shannon information entropy:

[0105]

[0106] Among them, when r′ j =1 / m!, U′r(m) reaches the maximum value In(m!);

[0107] The second permutation entropy is normalized to reflect the discreteness of the EEG signal after denoising in the segment;

[0108] The specific steps of selecting the first permutation entropy and the second permutation entropy according to the frequency of the EEG signal after denoising described in the paragraph to obtain the discreteness evaluation include:

[0109] Decomposing the denoised EEG signal in X layers, and reconstructing the scale frequency band of each layer after decomposition to obtain reconstructed signals of X+1 frequency bands;

[0110] The dispersion analysis is performed by using the first permutation entropy and the second permutation entropy, as follows:

[0111] Selecting several continuous points of denoised EEG data as basic data, and selecting movement data according to frequency bands to arrange and group the basic data;

[0112] Obtain the data in each permutation group, and calculate the average data of the data in each permutation group as the permutation entropy value;

[0113] Analyze the permutation entropy value to select the frequency band with the greatest discrimination, use the frequency band with the greatest discrimination as the discrimination frequency band, and compare and analyze the discrimination frequency band by using the first permutation entropy and the second permutation entropy to obtain the influence of different basic data lengths, different embedding dimensions and different delay times on the discrimination effect;

[0114] According to the influence of the distinction effect, the first permutation entropy or the second permutation entropy is selected to process the denoised EEG signal of the segment to obtain a discreteness evaluation;

[0115] Training module: used to extract the denoised EEG signal with a discreteness evaluation less than a preset threshold as the training EEG signal;

[0116] Verification module: used for extracting features from the training EEG signal and comparing it with the real EEG signal of the corresponding disease to verify the EEG signal in the EEG signal data set;

[0117] The verification module is specifically used for:

[0118] The training EEG signal is defined as And perform discrete Fourier transform to obtain multiple frequency spectra of the training EEG signal:

[0119]

[0120] in, is the frequency domain representation of the training EEG signal, parameter w=0,1,2,……,W-1; W represents the signal length of the training EEG signal, k=0,1,2,……,W / 2-1, k is the sampling point within the sampling frequency range, and the sampling frequency range is F s is the sampling frequency;

[0121] The multiple spectra are divided into L consecutive segments denoted as R l , and denote the boundary of each of the paragraphs as Ω l , that is, R l =[Ω l-1 ,Ω l ];

[0122] Where, l = 0, 1, 2, ..., L;

[0123] Applying a scaling function and a wavelet function to filter and decompose the signal on each of the frequency spectrums to obtain an EEG signal for comparison, so as to extract features of the EEG signal for training;

[0124] Comparing the comparison EEG signal with the real EEG signal of the corresponding disease in the database to verify the EEG signal in the EEG signal data set;

[0125] The specific steps of applying the scaling function and the wavelet function to filter the decomposed signal on each of the paragraphs to obtain the EEG signal for comparison, and extracting features from the EEG signal for training include:

[0126] Applying the scaling function φ(Ω) on each of the frequency spectra creates a filter and decomposes the signal to obtain a first verification signal:

[0127]

[0128] Wherein, P represents the amplitude of the current spectrum;

[0129] Apply the wavelet function to each of the spectra Create the filter and decompose the signal to get the second verification signal:

[0130]

[0131] The first verification signal and the second verification signal are integrated to obtain an EEG electroencephalogram (EEG) signal for comparison, and then feature extraction is performed on the EEG electroencephalogram (EEG) signal for training.

[0132] Compared with the prior art, the present invention has the following beneficial effects:

[0133] The EEG data set of the patient's corresponding frequency band is extracted according to the frequency band of the real EEG signal corresponding to a variety of diseases, wherein the frequency bands of the real EEG signals of the various diseases are close, and the EEG data set includes the EEG signals of the EEG in a plurality of time periods, so as to preliminarily obtain the processing data to be used in the present invention, and then the EEG signal is denoised to obtain a clean denoised EEG signal, so as to prevent the EEG signal from being disturbed by noise during the acquisition process, making it difficult to directly extract useful information from the original EEG signal, and to prevent the EEG signal from being disturbed by noise, resulting in a high frequency band, and the real disease is misjudged in the later stage, and then the denoised EEG signal is subjected to a discreteness evaluation process to eliminate data with too high discreteness, which means that this period The EEG signal after denoising is very random. It may be misselected because a part of this EEG signal after denoising is close to the frequency band of the real EEG signal, and cannot be used as useful data. Then the data with low discreteness is used as the usable training EEG signal. A large part of these data is in the frequency band of the real EEG signal, and has stronger verification. Finally, the training EEG signal is feature extracted to extract the deeper signal of this signal, which is compared with the deeper signal of the real EEG signal in the database to verify whether the EEG signal in the original EEG signal data set is the same as the prediction, and then the authenticity of the data is verified at the algorithm level, avoiding taking a lot of time to verify the reliability of the data, resulting in a waste of time. BRIEF DESCRIPTION OF THE DRAWINGS

[0134] Figure 1 This is a flow chart of a method for processing patient cardiovascular and cerebrovascular disease data in a first embodiment of the present invention;

[0135] Figure 2 It is a structural block diagram of a patient's cardiovascular and cerebrovascular disease data processing system in a second embodiment of the present invention;

[0136] The following specific implementation manner will further illustrate the present invention in conjunction with the above-mentioned drawings. DETAILED DESCRIPTION

[0137] In order to facilitate the understanding of the present invention, the present invention will be described more fully below with reference to the relevant drawings. Several embodiments of the present invention are given in the drawings. However, the present invention can be implemented in many different forms and is not limited to the embodiments described herein. On the contrary, the purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive.

[0138] It should be noted that when an element is referred to as being "fixed to" another element, it may be directly on the other element or there may be a central element. When an element is considered to be "connected to" another element, it may be directly connected to the other element or there may be a central element at the same time. The terms "vertical", "horizontal", "left", "right" and similar expressions used herein are for illustrative purposes only.

[0139] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used herein in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items.

[0140] See also Figure 1 , shown is a method for processing patient cardiovascular and cerebrovascular disease data according to a first embodiment of the present invention, characterized in that it comprises the following steps:

[0141] S10, extracting an EEG signal data set of a patient's corresponding frequency band according to the frequency bands of the real EEG signals corresponding to the various diseases, wherein the EEG signal data set includes EEG signals in multiple time periods;

[0142] S20, performing denoising on the EEG signal in the EEG signal data set to obtain a denoised EEG signal;

[0143] S30, performing a discreteness evaluation process on the denoised EEG signal to obtain a discreteness evaluation of each segment of the denoised EEG signal;

[0144] S40, extracting the denoised EEG signal with a discreteness evaluation less than a preset threshold as a training EEG signal;

[0145] S50, extracting features from the training EEG signal, and comparing it with the real EEG signal of the corresponding disease, so as to verify the EEG signal in the EEG signal data set.

[0146] It can be understood that the present invention extracts the EEG data set of the patient's corresponding frequency band according to the frequency band of the real EEG signal corresponding to a variety of diseases, wherein the frequency bands of the real EEG signals of the various diseases are close, and the EEG data set includes the EEG signals of the EEG in a plurality of time periods, so as to preliminarily obtain the processing data to be used in the present invention, and then perform denoising on the EEG signal to obtain a clean denoised EEG signal, so as to prevent the EEG signal from being disturbed by noise during the acquisition process, making it difficult to directly extract useful information from the original EEG signal, and at the same time prevent the EEG signal from being disturbed by noise, resulting in a high frequency band, and later misjudging the real disease, and then perform discreteness evaluation on the denoised EEG signal to eliminate data with too high discreteness. This means that this section of denoised EEG is very random, and it may be misselected just because a part of this section of denoised EEG is close to the frequency band of the real EEG, and cannot be used as useful data. Then the data with low discreteness is used as the usable training EEG. A large part of these data is in the frequency band of the real EEG, and has stronger verification. Finally, the training EEG is feature extracted to extract the deeper signals of this section of the signal, and compared with the deeper signals of the real EEG in the database to verify whether the EEG in the original EEG data set is the same as predicted, thereby verifying the authenticity of the data at the algorithm level, avoiding a lot of time to verify the reliability of the data, resulting in a waste of time.

[0147] Specifically, in this embodiment, the specific steps of step S20 include:

[0148] Using the Daubechies5 wavelength function to perform a three-layer decomposition on the EEG signal in the EEG signal data set to remove power frequency interference and electromyographic noise in the EEG signal;

[0149] Using a 600ms and 200ms sliding average filter to remove the T wave, P wave and QRS complex wave of the EEG signal to obtain a processed EEG signal;

[0150] The processed EEG signal is used as a baseline, and the processed EEG signal is removed from the EEG signal to obtain a denoised EEG signal.

[0151] It is understandable that the EEG signal is easily interfered by noise during the acquisition process, and it is difficult to directly extract useful information from the original EEG signal. At the same time, in order to prevent the EEG signal from being interfered by noise, resulting in a high frequency band, the real disease is misjudged in the later stage. Therefore, filtering is required before further processing. The present invention decomposes the signal into three layers through the Daubechies5 (db5) wavelength function to remove the power frequency interference and electromyographic noise, and then uses a 600ms and 200ms sliding average filter to remove the T wave, P wave and QRS complex wave, and then uses the processed signal as the baseline and subtracts it from the original signal to obtain a baseline-corrected denoised EEG signal.

[0152] Furthermore, the specific steps of step S30 include:

[0153] Performing a first permutation entropy transformation on each segment of the denoised EEG signal to obtain a first permutation entropy of the denoised EEG signal;

[0154] Performing a second permutation entropy transformation on each segment of the denoised EEG signal to obtain a second permutation entropy of the denoised EEG signal;

[0155] The first permutation entropy and the second permutation entropy are selected according to the frequency of the EEG signal after denoising described in the paragraph to obtain a dispersion evaluation.

[0156] Furthermore, the specific steps of performing the first permutation entropy conversion on each segment of the denoised EEG signal to obtain the first permutation entropy of the denoised EEG signal include:

[0157] The denoised EEG signals of each segment are reconstructed in time and space to obtain:

[0158] X(q)={x(q), r(q+γ),…, r(q+(n-1)γ)}

[0159] The sampling time of each segment of the denoised EEG signal is consistent, γ is the delay time, n is the embedding dimension, and X(q) represents the first reconstructed component of the qth segment of the denoised EEG signal;

[0160] Arrange the first reconstructed components of each segment of the denoised EEG signal in ascending order to obtain:

[0161] x(n+(p1-1)γ)≤x(q+(p2-1)γ)≤…≤x(q+(p k -1)γ)

[0162] If the values ​​of the first reconstructed components do not have the same value, then p krepresents x(n+(p1-1)γ), if the values ​​of the first reconstruction components are the same, then p ki The values ​​are arranged in ascending order, that is, when p k1 <p k2 When x(q+(p k1 -1)γ)≤x(q+(p k2 -1)γ);

[0163] Symbolize each segment of the first reconstructed component to obtain a first symbolized sequence:

[0164] E(l)=(p1,p2,…,p m )

[0165] Where, l = 1, 2, ..., k, and k is less than or equal to m! ;

[0166] The first symbolized sequence is defined as the first permutation entropy in the form of Shannon information entropy:

[0167]

[0168] Among them, when r j =1 / m!, Ur(m) reaches the maximum value In(m!);

[0169] The first permutation entropy is normalized to reflect the discreteness of the denoised EEG signal in this segment.

[0170] It can be understood that the Ur value mainly reflects two types of problems. The first type of problem is that it can test the degree of randomness of the first reconstructed component. If the Ur value is larger, it means that the first reconstructed component is more random and has no utilization value. On the contrary, it means that it is more stable and can be used as data for subsequent use. The second type of problem is that it can characterize the subtle changes in the first reconstructed component. If the Ur value is larger, it means that the change is more subtle, which means that this segment of data is more complex and less stable, and also has no utilization value. From the perspective of algorithm implementation, it only needs to compare the size relationship of the first reconstructed components of different segments, without knowing the specific size of the first reconstructed component. It only needs to set a threshold that meets actual needs to select the denoised EEG signal to enter the next step. In addition, during the acquisition process, because the interference of DC and extremely low frequency basically cannot change the size relationship between the values ​​in the sequence, we also save the trouble of DC isolation processing of the signal.

[0171] Furthermore, the specific steps of performing a second permutation entropy conversion on each segment of the denoised EEG signal to obtain the second permutation entropy of the denoised EEG signal include:

[0172] The denoised EEG signals of each segment are reconstructed in time and space to obtain:

[0173] X′(q)={x′(q), x′(q+γ), ..., x′(q+(n-1)γ)}, wherein the sampling time of each segment of the denoised EEG signal is consistent, γ is the delay time, n is the embedding dimension, and X′(q) represents the second reconstructed component of the qth segment of the denoised EEG signal;

[0174] Arrange the second reconstructed components of each segment of the denoised EEG signal in ascending order to obtain:

[0175] x′(n+(p1-1)γ)≤x′(q+(p2-1)γ)≤…≤x′(q+(p k -1)γ)

[0176] If the values ​​of the second reconstruction components do not have the same value, then p k represents x′(n+(p1-1)γ). If there is a second reconstruction component with the same value, then p ki The value represents the consistent second reconstruction component, that is, x′(n+(p k1 -1)γ)=x′(q+p k2 -1)γ);

[0177] Each segment of the second reconstructed component is symbolized to obtain a second symbolized sequence:

[0178] E′(l)=(p1,p2,…,p m )

[0179] Where, l = 1, 2, ..., k, and k is less than or equal to m! ;

[0180] The second symbolized sequence is defined as the second permutation entropy in the form of Shannon information entropy:

[0181]

[0182] Among them, when r′ j =1 / m!, U′ r (m) reaches the maximum value In(m!);

[0183] The second permutation entropy is normalized to reflect the discreteness of the denoised EEG signal in this segment.

[0184] It can be understood that when the first permutation entropy algorithm processes the case of equal values ​​in a sequence, its processing method is to mark them according to the order in which the values ​​appear. In order to avoid the same values, the logic of the first permutation entropy algorithm is to artificially add noise, but this method is effective when the resolution is very high or the probability of equal values ​​is very low. For EEG signals, the signal resolution collected is low and the probability of equal values ​​is high, resulting in artificially adding noise, which will greatly affect the authenticity of the data and make it impossible to accurately describe the data. Therefore, we use the same symbol to represent equal values. If the values ​​of the second reconstructed components are consistent, p is used. ki The value represents the consistent second reconstruction component, that is, x′(n+(p k1 -1)γ)=x′(q+(p k2 -1)γ);

[0185] Taking two different second reconstruction components as an example, when

[0186] X'(1)=(0.5, 0.3, 0.8, 0.1, 0.3), X'(1)=(0.5, 0.3, 0.8, 0.1, 0.4)

[0187] After sorting in ascending order

[0188] X'(1)=(0.1, 0.3, 0.3, 0.5, 0.8), X'(2)=(0.1, 0.3, 0.3, 0.5, 0.8)

[0189] If the first permutation entropy algorithm is used to symbolize, then

[0190] E(1)=E(2)=(4,2,5,1,3)

[0191] If the second permutation entropy algorithm is used to symbolize, then

[0192] E'(1)=(4,2,2,1,3), E'(2)=(4,2,5,1,3)

[0193] When the probability of equal value situations is low, the algorithm of first permutation entropy is better, and when the probability of equal value situations is high, the algorithm of second permutation entropy is better;

[0194] The meaning of the second permutation entropy U′r(m) is the same as that of the first permutation entropy Ur(m), which is only used for the purpose of distinction and will not be described in detail.

[0195] Furthermore, the specific steps of selecting the first permutation entropy and the second permutation entropy according to the frequency of the EEG signal after denoising described in this paragraph to obtain the discreteness evaluation include:

[0196] Decomposing the denoised EEG signal in X layers, and reconstructing the scale frequency band of each layer after decomposition to obtain reconstructed signals of X+1 frequency bands;

[0197] The dispersion analysis is performed by using the first permutation entropy and the second permutation entropy, as follows:

[0198] Selecting several continuous points of denoised EEG data as basic data, and selecting movement data according to frequency bands to arrange and group the basic data;

[0199] Obtain the data in each permutation group, and calculate the average data of the data in each permutation group as the permutation entropy value;

[0200] Analyze the permutation entropy value to select the frequency band with the greatest discrimination, use the frequency band with the greatest discrimination as the discrimination frequency band, and compare and analyze the discrimination frequency band by using the first permutation entropy and the second permutation entropy to obtain the influence of different basic data lengths, different embedding dimensions and different delay times on the discrimination effect;

[0201] According to the influence of the distinction effect, the first permutation entropy or the second permutation entropy is selected to process the denoised EEG signal of the segment to obtain a discreteness evaluation.

[0202] It can be understood that the denoised EEG signal of this segment can be analyzed and processed by a single reconstruction, thereby eliminating the influence of the denoised EEG signal of other segments on this segment. In this embodiment, the EEG signal is decomposed into 7 layers using the bior4.4 wavelet, and finally a reconstructed signal of 8 frequency bands is obtained. The reconstructed signals of the 8 frequency bands are then analyzed by the first permutation entropy and the second permutation entropy respectively. Specifically, 2000 basic data lengths are selected, and then N=500 is selected as the length of the moving data. 50 data points are moved each time, that is, 650-1150, 700-1200. In this way, 30 groups can be divided. The permutation entropy values ​​of the 30 groups of data are calculated, and the average of the permutation entropy values ​​of the 30 groups of data is used as the final permutation entropy value to reduce the impact of the fluctuation of the permutation entropy value, and then the largest frequency band is distinguished according to the final permutation entropy value to observe whether the permutation entropy value calculated by the first permutation entropy algorithm or the second permutation entropy algorithm has the best effect. For this distinction, the effects of different basic data lengths, different embedding dimensions and different delay times on the distinction effect are considered to ensure the representativeness of the data. The representative data are arranged in sequence according to the permutation entropy value as the discreteness evaluation of each segment of the EEG after denoising.

[0203] Then, the denoised EEG signal with a discreteness evaluation higher than or lower than a preset threshold is used as the training EEG signal.

[0204] Furthermore, the specific steps of step S50 include:

[0205] The training EEG signal is defined as And perform discrete Fourier transform to obtain multiple frequency spectra of the training EEG signal:

[0206]

[0207] in, is the frequency domain representation of the training EEG signal, parameter w=0,1,2,……,W-1; W represents the signal length of the training EEG signal, k=0,1,2,……,W / 2-1, k is the sampling point within the sampling frequency range, and the sampling frequency range is F s is the sampling frequency;

[0208] The multiple spectra are divided into L consecutive segments denoted as R l , and denote the boundary of each of the paragraphs as Ω l , that is, R l =[Ω l-1 ,Ω l ];

[0209] Where, l = 0, 1, 2, ..., L;

[0210] Applying a scaling function and a wavelet function to filter and decompose the signal on each of the frequency spectrums to obtain an EEG signal for comparison, so as to extract features of the EEG signal for training;

[0211] The comparison EEG signal is compared with the real EEG signal of the corresponding disease in the database to verify the EEG signal in the EEG signal data set.

[0212] According to one aspect of the above technical solution, the specific steps of applying a scaling function and a wavelet function to filter the decomposed signal on each of the paragraphs to obtain an EEG signal for comparison, and extracting features from the EEG signal for training include:

[0213] Applying the scaling function φ(Ω) on each of the frequency spectra creates a filter and decomposes the signal to obtain a first verification signal:

[0214]

[0215] Wherein, P represents the amplitude of the current spectrum;

[0216] Apply the wavelet function to each of the spectra Create the filter and decompose the signal to get the second verification signal:

[0217]

[0218] The first verification signal and the second verification signal are integrated to obtain an EEG electroencephalogram (EEG) signal for comparison, and then feature extraction is performed on the EEG electroencephalogram (EEG) signal for training.

[0219] It can be understood that after selecting the training EEG signal, the feature extraction of the training EEG signal can be performed. Specifically, by analyzing the Fourier spectrum of the training EEG signal and creating an adaptive frequency of the Fourier spectrum, the spectrum of the training EEG signal can be segmented by methods such as local maximum and local minimum between local maximums. Then, the empirical wavelet function and the scaling function are applied to the segmented spectrum to decompose the signal and extract deep useful information to obtain the first verification signal and the second verification signal respectively. The first verification signal and the second verification signal are integrated as the comparison EEG signal, and the scaling function φ(Ω) and the wavelet function of the comparison EEG signal are compared with the real EEG signal in the database. The corresponding signal comparison obtained is to use the EEG signal as the final verification result when the error is less than a certain threshold. This can confirm the EEG signal frequency band at the data level, greatly improving the accuracy.

[0220] In summary, the method for processing patient cardiovascular and cerebrovascular disease data in the above-mentioned embodiment of the present invention verifies the authenticity of the data at the algorithm level, avoiding taking a lot of time to verify the reliability of the data, which causes a waste of time.

[0221] Please refer to Figure 2 The second embodiment of the present invention further provides a patient cardiovascular and cerebrovascular disease data processing system, comprising:

[0222] Extraction module 11: used to extract EEG data sets of corresponding frequency bands of patients according to frequency bands of real EEG signals corresponding to multiple diseases, wherein the EEG data sets include the EEG signals in multiple time periods;

[0223] Denoising module 12: used for performing denoising processing on the EEG signal in the EEG signal data set to obtain a denoised EEG signal;

[0224] The denoising module 12 is specifically used to: perform three-layer decomposition of the EEG signal in the EEG signal data set using the Daubechies5 wavelength function to remove power frequency interference and electromyographic noise in the EEG signal;

[0225] Using a 600ms and 200ms sliding average filter to remove the T wave, P wave and QRS complex wave of the EEG signal to obtain a processed EEG signal;

[0226] Taking the processed EEG as a baseline, removing the processed EEG from the EEG to obtain a denoised EEG;

[0227] Evaluation module 13: used for performing discreteness evaluation processing on the denoised EEG signal to obtain discreteness evaluation of each segment of the denoised EEG signal;

[0228] The evaluation module 13 is specifically used to: perform a first permutation entropy conversion on each segment of the denoised EEG signal to obtain a first permutation entropy of the denoised EEG signal;

[0229] Performing a second permutation entropy transformation on each segment of the denoised EEG signal to obtain a second permutation entropy of the denoised EEG signal;

[0230] The first permutation entropy and the second permutation entropy are selected according to the frequency of the EEG signal after denoising described in the paragraph to obtain a dispersion evaluation;

[0231] The specific steps of performing the first permutation entropy conversion on each segment of the denoised EEG signal to obtain the first permutation entropy of the denoised EEG signal include:

[0232] The denoised EEG signals of each segment are reconstructed in time and space to obtain:

[0233] X(q)={x(q),x(q+γ),…,x(q+(n-1)γ)}

[0234] The sampling time of each segment of the denoised EEG signal is consistent, γ is the delay time, n is the embedding dimension, and X(q) represents the first reconstructed component of the qth segment of the denoised EEG signal;

[0235] Arrange the first reconstructed components of each segment of the denoised EEG signal in ascending order to obtain:

[0236] x(n+(p1-1)γ)≤x(q+(p2-1)γ≤…≤x(q+(p k -1)γ)

[0237] If the values ​​of the first reconstructed components do not have the same value, then p k represents x(n+(p1-1)γ), if the values ​​of the first reconstruction components are the same, then p kiThe values ​​are arranged in ascending order, that is, when p k1 <p k2 When x(q(p k1 -1)γ)≤x(q+(p k2 -1)γ);

[0238] Symbolize each segment of the first reconstructed component to obtain a first symbolized sequence:

[0239] E(l)=(p1,p2,…,p m )

[0240] Where, l = 1, 2, ..., k, and k is less than or equal to m! ;

[0241] The first symbolized sequence is defined as the first permutation entropy in the form of Shannon information entropy:

[0242]

[0243] Among them, when r j =1 / m!, Ur(m) reaches the maximum value In(m!);

[0244] Normalizing the first permutation entropy to reflect the discreteness of the EEG signal after denoising;

[0245] The specific steps of performing a second permutation entropy conversion on each segment of the denoised EEG signal to obtain the second permutation entropy of the denoised EEG signal include:

[0246] The denoised EEG signals of each segment are reconstructed in time and space to obtain:

[0247] X′(q)=(x′(q),x′(q+γ),...,x′(q+(n-1)γ)}

[0248] Wherein, the sampling time of each segment of the denoised EEG signal is consistent, γ is the delay time, n is the embedding dimension, and X'(q) represents the second reconstructed component of the qth segment of the denoised EEG signal;

[0249] Arrange the second reconstructed components of each segment of the denoised EEG signal in ascending order to obtain:

[0250] x′(n+(p1-1)γ)≤x′(q+(p2-1)γ)≤…≤x′(q+(p k -1)γ)

[0251] If the values ​​of the second reconstruction components do not have the same value, then p krepresents x′(n+(p1-1)γ). If there is a second reconstruction component with the same value, then p ki The value represents the consistent second reconstruction component, that is, x′(n+p k1 -1)γ)=x′(q+(p k2 -1)γ);

[0252] Each segment of the second reconstructed component is symbolized to obtain a second symbolized sequence:

[0253] E′(l)=(p1,p2,…,p m )

[0254] Where, l = 1, 2, ..., k, and k is less than or equal to m! ;

[0255] The second symbolized sequence is defined as the second permutation entropy in the form of Shannon information entropy:

[0256]

[0257] Among them, when r′ j =1 / m!, U′ r (m) reaches the maximum value In(m!);

[0258] The second permutation entropy is normalized to reflect the discreteness of the EEG signal after denoising in the segment;

[0259] The specific steps of selecting the first permutation entropy and the second permutation entropy according to the frequency of the EEG signal after denoising described in the paragraph to obtain the discreteness evaluation include:

[0260] Decomposing the denoised EEG signal in X layers, and reconstructing the scale frequency band of each layer after decomposition to obtain reconstructed signals of X+1 frequency bands;

[0261] The dispersion analysis is performed by using the first permutation entropy and the second permutation entropy, as follows:

[0262] Selecting several continuous points of denoised EEG data as basic data, and selecting movement data according to frequency bands to arrange and group the basic data;

[0263] Obtain the data in each permutation group, and calculate the average data of the data in each permutation group as the permutation entropy value;

[0264] Analyze the permutation entropy value to select the frequency band with the greatest discrimination, use the frequency band with the greatest discrimination as the discrimination frequency band, and compare and analyze the discrimination frequency band by using the first permutation entropy and the second permutation entropy to obtain the influence of different basic data lengths, different embedding dimensions and different delay times on the discrimination effect;

[0265] According to the influence of the distinction effect, the first permutation entropy or the second permutation entropy is selected to process the denoised EEG signal of the segment to obtain a discreteness evaluation;

[0266] Training module 14: used to extract the denoised EEG signal with a discreteness evaluation less than a preset threshold as a training EEG signal;

[0267] Verification module 15: used to extract features from the training EEG signal and compare it with the real EEG signal of the corresponding disease to verify the EEG signal in the EEG signal data set;

[0268] The verification module 15 is specifically used for:

[0269] The training EEG signal is defined as And perform discrete Fourier transform to obtain multiple frequency spectra of the training EEG signal:

[0270]

[0271] in, is the frequency domain representation of the training EEG signal, parameter w=0,1,2,……,W-1; W represents the signal length of the training EEG signal, k=0,1,2,……,W / 2-1, k is the sampling point within the sampling frequency range, and the sampling frequency range is F s is the sampling frequency;

[0272] The multiple spectra are divided into L consecutive segments denoted as R l , and denote the boundary of each of the paragraphs as Ω l , that is, R l =[Ω l-1 ,Ω l ];

[0273] Where, l = 0, 1, 2, ..., L;

[0274] Applying a scaling function and a wavelet function to each of the frequency spectrums to filter and decompose the signal, thereby obtaining an EEG signal for comparison, so as to extract features from the EEG signal for training;

[0275] Comparing the comparison EEG signal with the real EEG signal of the corresponding disease in the database to verify the EEG signal in the EEG signal data set;

[0276] The specific steps of applying the scaling function and the wavelet function to filter the decomposed signal on each of the paragraphs to obtain the EEG signal for comparison, and extracting features from the EEG signal for training include:

[0277] Applying the scaling function φ(Ω) on each of the frequency spectra creates a filter and decomposes the signal to obtain a first verification signal:

[0278]

[0279] Wherein, P represents the amplitude of the current spectrum;

[0280] Apply the wavelet function to each of the spectra Create the filter and decompose the signal to get the second verification signal:

[0281]

[0282] The first verification signal and the second verification signal are integrated to obtain an EEG electroencephalogram (EEG) signal for comparison, and then feature extraction is performed on the EEG electroencephalogram (EEG) signal for training.

[0283] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0284] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the patent of the present invention. It should be pointed out that, for ordinary technicians in this field, several variations and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the attached claims.

Claims

1. A method for processing patient cardiovascular and cerebrovascular disease data, characterized in that: The steps include: S10, extracting an EEG signal data set of a patient's corresponding frequency band according to the frequency bands of the real EEG signals corresponding to the various diseases, wherein the EEG signal data set includes EEG signals in multiple time periods; S20, performing denoising on the EEG signal in the EEG signal data set to obtain a denoised EEG signal; S30, performing a discreteness evaluation process on the denoised EEG signal to obtain a discreteness evaluation of each segment of the denoised EEG signal; S40, extracting the denoised EEG signal with a discreteness evaluation less than a preset threshold as a training EEG signal; S50, extracting features from the training EEG signal, and comparing it with the real EEG signal of the corresponding disease, so as to verify the EEG signal in the EEG signal data set.

2. The method for processing patient cardiovascular and cerebrovascular disease data according to claim 1, characterized in that: The specific steps of step S20 include: Using the Daubechies5 wavelength function to perform a three-layer decomposition on the EEG signal in the EEG signal data set to remove power frequency interference and electromyographic noise in the EEG signal; Using a 600ms and 200ms sliding average filter to remove the T wave, P wave and QRS complex wave of the EEG signal to obtain a processed EEG signal; The processed EEG signal is used as a baseline, and the processed EEG signal is removed from the EEG signal to obtain a denoised EEG signal.

3. The method for processing patient cardiovascular and cerebrovascular disease data according to claim 1, characterized in that: The specific steps of step S30 include: Performing a first permutation entropy transformation on each segment of the denoised EEG signal to obtain a first permutation entropy of the denoised EEG signal; Performing a second permutation entropy transformation on each segment of the denoised EEG signal to obtain a second permutation entropy of the denoised EEG signal; The first permutation entropy and the second permutation entropy are selected according to the frequency of the EEG signal after denoising described in the paragraph to obtain a dispersion evaluation.

4. The method for processing patient cardiovascular and cerebrovascular disease data according to claim 3, characterized in that: The specific steps of performing the first permutation entropy conversion on each segment of the denoised EEG signal to obtain the first permutation entropy of the denoised EEG signal include: The denoised EEG signals of each segment are reconstructed in time and space to obtain: X(q)={x(q),x(q+γ),...,x(q+(n-1)γ)} The sampling time of each segment of the denoised EEG signal is consistent, γ is the delay time, n is the embedding dimension, and X(q) represents the first reconstructed component of the qth segment of the denoised EEG signal; Arrange the first reconstructed components of each segment of the denoised EEG signal in ascending order to obtain: x(n+(p1-1)γ)≤x(q+(p2-1)γ)≤…≤r(q+(p k -1)c) If the values ​​of the first reconstructed components do not have the same value, then p k represents x(n+(p1-1)γ), if the values ​​of the first reconstruction components are the same, then p ki The values ​​are arranged in ascending order, that is, when p k1 <p k2 When x(q+(p k1 -1)γ)≤x(q+(p k2 -1)γ); Symbolize each segment of the first reconstructed component to obtain a first symbolized sequence: E(l)=(p1,p2,…,p m ) Where, l = 1, 2, ..., k, and k is less than or equal to m! ; The first symbolized sequence is defined as the first permutation entropy in the form of Shannon information entropy: Among them, when r j =1 / m!, Ur(m) reaches the maximum value In(m!); The first permutation entropy is normalized to reflect the discreteness of the denoised EEG signal in this segment.

5. The method for processing patient cardiovascular and cerebrovascular disease data according to claim 4, characterized in that: The specific steps of performing a second permutation entropy conversion on each segment of the denoised EEG signal to obtain the second permutation entropy of the denoised EEG signal include: The denoised EEG signals of each segment are reconstructed in time and space to obtain: X′(q)={x′(q), x′(q+γ), ..., x′(q+(n-1)γ)}, wherein the sampling time of each segment of the denoised EEG signal is consistent, γ is the delay time, n is the embedding dimension, and X′(q) represents the second reconstructed component of the qth segment of the denoised EEG signal; Arrange the second reconstructed components of each segment of the denoised EEG signal in ascending order to obtain: x′(n+(p1-1)γ)≤x′(q+(p2-1)γ)≤…≤x′(q+(p k -1)c) If the values ​​of the second reconstruction components do not have the same value, then p k represents x′(n+(p1-1)γ). If there is a second reconstruction component with the same value, then p ki The value represents the consistent second reconstruction component, that is, x′(n+(p k1 -1)γ)=x′(q+p k2 -1)γ); Each segment of the second reconstructed component is symbolized to obtain a second symbolized sequence: And′(l)=(p1,p2,…,p m ) Where, l = 1, 2, ..., k, and k is less than or equal to m! ; The second symbolized sequence is defined as the second permutation entropy in the form of Shannon information entropy: Among them, when r′ j =1 / m!, U′r(m) reaches the maximum value In(m!); The second permutation entropy is normalized to reflect the discreteness of the denoised EEG signal in this segment.

6. The method for processing patient cardiovascular and cerebrovascular disease data according to claim 5, characterized in that: The specific steps of selecting the first permutation entropy and the second permutation entropy according to the frequency of the EEG signal after denoising described in the paragraph to obtain the discreteness evaluation include: Decomposing the denoised EEG signal in X layers, and reconstructing the scale frequency band of each layer after decomposition to obtain reconstructed signals of X+1 frequency bands; The dispersion analysis is performed by using the first permutation entropy and the second permutation entropy, as follows: Selecting several continuous points of denoised EEG data as basic data, and selecting movement data according to frequency bands to arrange and group the basic data; Obtain the data in each permutation group, and calculate the average data of the data in each permutation group as the permutation entropy value; Analyze the permutation entropy value to select the frequency band with the greatest discrimination, use the frequency band with the greatest discrimination as the discrimination frequency band, and compare and analyze the discrimination frequency band by using the first permutation entropy and the second permutation entropy to obtain the influence of different basic data lengths, different embedding dimensions and different delay times on the discrimination effect; According to the influence of the distinction effect, the first permutation entropy or the second permutation entropy is selected to process the denoised EEG signal of the segment to obtain a discreteness evaluation.

7. The method for processing patient cardiovascular and cerebrovascular disease data according to claim 6, characterized in that: The specific steps of step S50 include: The training EEG signal is defined as And perform discrete Fourier transform to obtain multiple frequency spectra of the training EEG signal: in, is the frequency domain representation of the training EEG signal, parameter w=0,1,2,……,W-1; W represents the signal length of the training EEG signal, k=0,1,2,……,W / 2-1, k is the sampling point within the sampling frequency range, and the sampling frequency range is F s is the sampling frequency; The multiple spectra are divided into L consecutive segments denoted as R l , and denote the boundary of each of the paragraphs as Ω l , that is, R l =[Ω l-1 ,Ω l ); Where, l = 0, 1, 2, ..., L; Applying a scaling function and a wavelet function to each of the frequency spectrums to filter and decompose the signal, thereby obtaining an EEG signal for comparison, so as to extract features from the EEG signal for training; The comparison EEG signal is compared with the real EEG signal of the corresponding disease in the database to verify the EEG signal in the EEG signal data set.

8. The method for processing patient cardiovascular and cerebrovascular disease data according to claim 7, characterized in that: The specific steps of applying the scaling function and the wavelet function to filter the decomposed signal on each of the paragraphs to obtain the EEG signal for comparison, and extracting features from the EEG signal for training include: Applying the scaling function φ(Ω) on each of the frequency spectra creates a filter and decomposes the signal to obtain a first verification signal: Wherein, P represents the amplitude of the current spectrum; Apply the wavelet function to each of the spectra Create the filter and decompose the signal to get the second verification signal: The first verification signal and the second verification signal are integrated to obtain an EEG electroencephalogram (EEG) signal for comparison, and then feature extraction is performed on the EEG electroencephalogram (EEG) signal for training.

9. A patient cardiovascular and cerebrovascular disease data processing system, characterized in that: include: Extraction module: used to extract EEG data sets of corresponding frequency bands of patients according to the frequency bands of real EEG signals corresponding to various diseases, wherein the EEG data sets include EEG signals in multiple time periods; Denoising module: used for denoising the EEG signal in the EEG signal data set to obtain the denoised EEG signal; Evaluation module: used for performing discreteness evaluation processing on the denoised EEG signal to obtain discreteness evaluation of each segment of the denoised EEG signal; Training module: used to extract the denoised EEG signal with a discreteness evaluation less than a preset threshold as the training EEG signal; Verification module: used to extract features from the training EEG signal and compare it with the real EEG signal of the corresponding disease to verify the EEG signal in the EEG signal data set.

Images