A youth depressive patient curative effect evaluation system based on electroencephalogram microstate

Abstract

The application discloses a kind of adolescent depressive patient curative effect evaluation system based on electroencephalogram microstate, specifically relates to electroencephalogram microstate analysis technical field, including data acquisition module, data preprocessing module, data processing module, microstate analysis module, prognosis model construction module, emotion recognition module, curative effect evaluation module and man-machine interaction module;Through prognosis model construction module, emotion recognition module, based on correlation prognosis model, microstate is mapped to emotional state;Through curative effect evaluation module, calculate curative effect evaluation index, obtain the curative effect evaluation result of the adolescent depressive disorder of the to-be-measured adolescent with attempted suicide behavior;Through man-machine interaction module, user can understand the real-time emotional state of adolescent, curative effect evaluation index and curative effect evaluation effect, to comprehensively, accurately and conveniently evaluate the brain stability and adaptability of treatment maintenance period adolescent depressive disorder patient.

Description

Technical Field

[0001] This invention relates to the field of brainwave microstate analysis technology, and more specifically, to a treatment efficacy evaluation system for adolescent patients with depression based on brainwave microstates. Background Technology

[0002] Adolescent depression is a common mental disorder that severely impacts patients' quality of life, academic performance, social functioning, and future development. Currently, treatment for adolescent depression primarily includes medication, psychotherapy, and a combination of these approaches. However, even after effective treatment, adolescents with depression still experience a high relapse rate, approximately 40%–60%. Therefore, effectively monitoring the brain state of adolescents with depression during the maintenance phase of treatment to assess treatment effectiveness and predict relapse risk is a pressing issue that needs to be addressed.

[0003] Currently, the main methods for monitoring the brain state of adolescents with depression include the following:

[0004] Clinical assessment: Standardized scales or questionnaires, such as the Hamilton Depression Rating Scale or the Childhood Depression Rating Scale, are used to assess the patient's depressive symptoms, functional impairment, and suicidal ideation. The advantage of this method is its simplicity and ease of use, but its disadvantages include high subjectivity and susceptibility to patient cooperation and reporting bias.

[0005] Neuroimaging: This method uses techniques such as functional magnetic resonance imaging (fMRI) and positron emission tomography (PET) to observe changes in the structure and function of the patient's brain, such as gray matter volume, cerebral blood flow, metabolic rate, and neuronal activity. The advantages of this method are its high objectivity and ability to reflect changes in deep brain structure and function. However, its disadvantages include high cost, complex operation, and limitations imposed by motion artifacts and head fixation.

[0006] Electroencephalography (EEG): This method records the patient's brain electrical signals by attaching electrodes to the scalp and performs various analyses, such as power spectrum analysis, coherence analysis, and microstate analysis, to reflect the patient's brain electrophysiological activity and network characteristics. The advantages of this method are low cost, simple operation, high temporal resolution, and the ability to reflect rapidly changing brain states. However, its disadvantages include a low signal-to-noise ratio and susceptibility to environmental interference and individual differences.

[0007] In summary, existing methods for monitoring the brain state of adolescents with depression each have their advantages and disadvantages. No single method can comprehensively, accurately, and conveniently assess the brain stability and adaptability of adolescents with depression during the maintenance phase of treatment, thereby evaluating treatment effectiveness and predicting relapse risk. Therefore, the purpose of this invention is to provide a treatment efficacy assessment system for adolescents with depression based on EEG microstates, specifically for assessing the treatment efficacy of depression in adolescents with suicidal attempts. Summary of the Invention

[0008] To overcome the aforementioned deficiencies of existing technologies, this invention provides a system for evaluating the efficacy of treatment for adolescent depression based on EEG microstates. The system collects EEG signals by having adolescents wear a non-invasive EEG device. A data preprocessing module preprocesses the signals, and a microstate analysis module analyzes the microstates. A prognostic model construction module and an emotion recognition module map the microstates to emotional states based on a correlational prognostic model. An efficacy evaluation module analyzes and predicts relapse risk based on evaluation parameters and emotional factors, calculates an efficacy evaluation index, and obtains the efficacy evaluation results for adolescent depression with suicidal attempts. A human-computer interaction module presents the mapped emotional states to the user through a user interface. Users can use this information to understand the adolescent's real-time emotional state, efficacy evaluation index, and efficacy evaluation effect, and can also view historical emotional state trends. This allows for a comprehensive, accurate, and convenient assessment of brain stability and adaptability in adolescents with depression during the maintenance phase of treatment, thus addressing the problems mentioned in the background section.

[0009] To achieve the above objectives, the present invention provides the following technical solution: a system for evaluating the efficacy of treatment in adolescent patients with depression based on electroencephalogram (EEG) microstates, comprising:

[0010] Data acquisition module: Used to collect brain signals from adolescents using non-invasive EEG signal acquisition equipment and transmit the data to the data preprocessing module;

[0011] Data preprocessing module: used to filter and denoise the acquired EEG signals to obtain clean signals, and then transmit them to the data processing module;

[0012] Data processing module: used to detect the clean signal after data preprocessing, evaluate the denoising effect based on the signal-to-noise ratio and mean square error, thereby obtaining the expected clean EEG signal and transmitting it to the microstate analysis module;

[0013] Microstate analysis module: Used to analyze and process pure EEG signals using an improved k-means clustering algorithm. Based on the clustering analysis results, it extracts the average duration, coverage, frequency of occurrence, amplitude of different EEG microstates, and the transition probability from one microstate to another as feature vectors of the clustering results, and transmits them to the prognostic model construction module.

[0014] Prognostic model building module: This module receives data from the microstate analysis module, uses one-way ANOVA and Spearman correlation to compare the resting state microstate parameters among groups, and builds a prognostic model.

[0015] The emotion recognition module is used to convert microstates into emotional states based on the correlation prognostic model, associate specific microstates or microstate sequences with a certain emotional state to obtain emotional factors, and transmit them to the efficacy evaluation module.

[0016] The efficacy assessment module is used to analyze and predict the risk of relapse based on assessment parameters and emotional factors, calculate the efficacy assessment index, and obtain the efficacy assessment results of adolescent depression with suicidal attempts in the test adolescents.

[0017] The human-computer interaction module is used to intuitively present the emotional state, efficacy assessment index, and efficacy assessment results to the user through the terminal's user interface.

[0018] In a preferred embodiment, the data acquisition module acquires data in the following manner: within the acquisition period, in a quiet, dark room with minimal electromagnetic interference, the adolescent being tested is placed in a resting state, and a non-invasive EEG signal acquisition device is worn on the head for 5 minutes according to the international 10-20 system electrode placement standard, to acquire EEG signals from the central and occipital regions; the acquisition frequency is a fixed time point every day, and the acquisition period is one week; the resting state specifically refers to the adolescent being awake, quiet, and with eyes closed.

[0019] In a preferred embodiment, the specific processing procedure of the data preprocessing module is as follows: Bandpass filtering of 1-30Hz is applied to the signals acquired from 64 channels to obtain four frequency bands for each subject's EEG signal: δ (0.5-4Hz), θ (4-8Hz), α (8-13Hz), and β (13-30Hz); the data is then denoised using a wavelet denoising method. Specifically, the wavelet denoising method is based on the characteristic that noise and signal have different intensity distributions in wavelet decomposition coefficients at different frequency bands. The wavelet coefficients corresponding to noise in each frequency band are removed, retaining the wavelet decomposition coefficients of the original signal. Then, wavelet reconstruction is performed on the processed coefficients to obtain a clean signal. The specific formula for wavelet denoising is: τ represents the time shift, a represents the scaling factor, and f(t) represents the input time-domain signal. This represents the wavelet basis function.

[0020] In a preferred embodiment, the pure signal is specifically taken as 80 seconds forward and backward from the center of the EEG signal time series, removing the first and last segments of the data and retaining only the middle effective part, ultimately obtaining 160 seconds of data.

[0021] In a preferred embodiment, the data processing module processes data as follows: it calculates the signal-to-noise ratio (SNR) according to the SNR calculation formula and calculates the mean square error (MSE) according to the MSE calculation formula; the SNR calculation formula is: The formula for calculating the mean square error is: Where Sn represents the signal-to-noise ratio, Sm represents the mean square error, and fi represents the clean signal. This represents the estimated signal after denoising; the larger the signal-to-noise ratio Sn and the smaller the mean square error Sm, the better the signal denoising effect.

[0022] Based on the signal-to-noise ratio Sn and the mean square error Sm, the signal-to-noise ratio threshold Sn corresponding to the pre-set denoising effect is determined. 阈 and mean squared error threshold Sm 阈 Perform a comparison and judgment; if Sn is satisfied... <Sn 阈 And Sm <Sm 阈 This indicates that the denoising effect meets expectations, and the clean EEG signal is transmitted to the microstate analysis module; if Sn is not satisfied... <Sn 阈 And Sm <Sm 阈 This indicates that the noise reduction effect is not as expected and noise reduction processing needs to be performed again until the noise reduction effect meets expectations.

[0023] In a preferred embodiment, the specific analysis process of the micro-state analysis module is as follows:

[0024] A1. Determine the sample set; the processing procedure is as follows:

[0025] A11. For each adolescent's EEG signal, the global field power (GFP) of 160 seconds of data is extracted to obtain a continuous topographic map of GFP. K-means clustering initialization analysis is performed to obtain the initial cluster centers. Here, GFP represents the standard deviation of all electrode voltage values ​​of the topographic map at a certain time point, which is used to describe the intensity of the electric field of a certain topographic map. The initialization analysis is random, so 20 different initializations were performed, and the optimal result was taken. After obtaining the optimal cluster, each time window can be assigned to a cluster, and each cluster is a microstate category.

[0026] A12. After obtaining the initial cluster centers of each adolescent, put all the initial cluster centers of adolescents together, and each initial cluster center is a sample set L, L = {x1, x2, ..., xn};

[0027] A2. Spatial cluster analysis was performed on the GFP peak topographic maps of the entire sample set; the processing procedure is as follows:

[0028] A21. Calculate the sample information entropy, and select the cluster that minimizes the information entropy to divide the sample into two clusters; the formula for calculating the sample information entropy is: Where D kl (f m (x)g q(x)) can measure the difference in the point distribution between cluster m and cluster q. When the information entropy is used in EEG data classification, the similarity between two clusters can be judged by the magnitude of the information entropy. When the information entropy is large, the distribution difference between the two clusters is large and the classification effect is good;

[0029] A22. Determine the global clustering center and select the point with a larger joint probability as the global clustering center point;

[0030] A23. Calculate the quality S of the clustering effect for the existing clusters; the calculation formula for the quality of the clustering effect is: where X is the longitude, Y is the latitude, p(x, y) is the longitude and latitude of the peak point, p(x) is the probability of the longitude occurrence, p(y) is the probability of the latitude occurrence, and p(x, y), p(x), and p(y) can be obtained from the original EEG data, D kl (f m (x)g q (x)) represents the sample information entropy, and w represents the proportionality coefficient;

[0031] A24. Set the judgment value H and make a judgment comparison with S;

[0032] A25. Repeat steps A23, A24, and A25. If S < H, it means that the clustering effect this time is better than the previous clustering effect, and continue the iteration; otherwise, stop the iteration and return K;

[0033] A3. Determine the clustering result according to the optimal clustering number K value; among them, the clustering result includes microstate A, microstate B, microstate C, microstate D, microstate λ, λ = 1, 2... n;

[0034] A4. After assigning a microstate category to each time window, obtain a one-dimensional discrete sequence of the change of the microstate category over time; extract the average duration, coverage range, occurrence frequency, amplitude of the microstate, and the transition probability from one microstate to other microstates as the feature vector of the clustering result and transmit it to the prognosis model construction module.

[0035] In a preferred embodiment, the specific processing process of the prognosis model construction module is as follows:

[0036] C1. Use one-way analysis of variance to calculate the influence of the feature vector of the clustering result on this microstate category; the specific calculation formula for the one-way analysis of variance is: Y(ij) = u(i) + ε(ij), where Y(ij) represents the jth observation value in the ith treatment, u(i) represents the average value of the ith treatment, and ε(ij) represents the random error of the jth observation value in the ith treatment;

[0037] C2. Calculate the correlation coefficient between the resting state microstates of each group according to the correlation coefficient calculation formula. Where η represents the correlation coefficient, xi and yi represent the corresponding elements in the variables, i.e., the feature vectors, and x and y represent the sample mean corresponding to x and y;

[0038] C3. Based on the data obtained in steps C1 and C2, perform a significance test.

[0039] In a preferred embodiment, the data obtained from steps C1 and C2 are subjected to a significance test, and the significance test steps are as follows:

[0040] C31. Assume that the two variables have no significant linear relationship, that is, the two variables are zero-correlation;

[0041] C32. Construct a new statistic t. Where η represents the correlation coefficient and n represents the sample size;

[0042] C33. Calculate the probability P value corresponding to the η distribution.

[0043] In a preferred embodiment, the specific formula for calculating the efficacy assessment index is as follows: Where б represents the efficacy assessment index, Ti, Mi, fi, 1 / a Let represent the average duration, coverage, frequency of occurrence, amplitude of microstate D, and probability of transition from one microstate to another, respectively, and ω represent the emotion factor.

[0044] Based on the adolescent's micro-state D over a week, and combined with the efficacy evaluation index calculation formula, the efficacy is evaluated, and an efficacy trend graph is plotted over time. If the efficacy trend graph shows an upward trend, the adolescent's efficacy is good; if the efficacy trend graph shows a downward trend, the adolescent's efficacy has a risk of relapse, and the treatment plan needs to be adjusted in a timely manner.

[0045] The technical effects and advantages of this invention are as follows:

[0046] This invention collects brainwave signals from adolescents wearing a non-invasive EEG device; preprocesses the signals using a data preprocessing module; analyzes microstates using a microstate analysis module; maps microstates to emotional states based on a correlational prognostic model using a prognostic model construction module and an emotion recognition module; analyzes and predicts relapse risk and calculates an efficacy assessment index based on assessment parameters and emotional factors using an efficacy assessment module to obtain efficacy assessment results for adolescent depression with suicidal attempts; and presents the mapped emotional states to the user through a human-computer interaction module. Users can understand the adolescent's real-time emotional state, efficacy assessment index, and efficacy assessment effect, and can also view historical emotional state trends, thus comprehensively, accurately, and conveniently assessing the brain stability and adaptability of adolescents with depression in the maintenance phase of treatment. Attached Figure Description

[0047] Figure 1 This is an overall block diagram of the present invention. Detailed Implementation

[0048] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0049] This invention provides, for example Figure 1 The system shown is an efficacy evaluation system for adolescent depression patients based on EEG microstates, comprising a data acquisition module, a data preprocessing module, a data processing module, a microstate analysis module, a prognostic model construction module, an emotion recognition module, an efficacy evaluation module, and a human-computer interaction module. The data acquisition module is connected to the data preprocessing module, the data preprocessing module is connected to the data processing module, the data processing module is connected to the microstate analysis module, the microstate analysis module is connected to the prognostic model construction module, the prognostic model construction module is connected to the emotion recognition module, the emotion recognition module is connected to the efficacy evaluation module, and the efficacy evaluation module is connected to the human-computer interaction module.

[0050] The data acquisition module is used to collect the brain signals of adolescents using a non-invasive EEG signal acquisition device and transmit the data to the data preprocessing module.

[0051] Specifically, the data acquisition module uses non-invasive equipment to collect EEG data from 28 healthy adolescents (TD group), 38 adolescents with severe depressive disorder (MIK group), and 54 adolescents with suicide attempts (UK group) in a quiet, dark room with minimal electromagnetic interference.

[0052] In this embodiment, it should be specifically explained that the data acquisition module acquires data in the following manner: During the acquisition period, in a quiet, light-proof room with minimal electromagnetic interference, the adolescent being tested is placed in a resting state, and according to the international 10-20 system electrode placement standard, wears a non-invasive EEG signal acquisition device on their head for 5 minutes to acquire EEG signals from the central and occipital regions; the acquisition frequency is a fixed time point every day, and the acquisition period is one week; the resting state specifically refers to the adolescent being awake, quiet, and with their eyes closed.

[0053] The data preprocessing module is used to filter and denoise the collected EEG signals to obtain clean signals, which are then transmitted to the data processing module.

[0054] In this embodiment, it should be specifically explained that the data preprocessing module performs the following steps: Bandpass filtering of 1-30Hz is applied to the signals acquired from 64 channels to obtain four frequency bands for each subject's EEG signal: δ (0.5-4Hz), θ (4-8Hz), α (8-13Hz), and β (13-30Hz). Wavelet denoising is then used to denoise the data. Specifically, the wavelet denoising method is based on the characteristic that noise and signal have different intensity distributions in different frequency bands. The wavelet coefficients corresponding to noise in each frequency band are removed, retaining the wavelet coefficients of the original signal. Then, wavelet reconstruction is performed on the processed coefficients to obtain a clean signal. The specific formula for wavelet denoising is: τ represents the time shift, a represents the scaling factor, and f(t) represents the input time-domain signal. The wavelet basis function is represented here. Considering that each person may experience fluctuations in their psychological state at the beginning and end of the instrument measurement and fail to immediately enter the corresponding state, and that the time and data sequence length of the EEG signals measured by each adolescent are different, in order to obtain more stable EEG information, the value of the pure signal is taken as follows: 80 seconds are taken before and after the center of the EEG signal time sequence, and the first and last segments of the data are removed, leaving only the effective middle part, finally obtaining 160 seconds of data, so that the EEG signal time sequence length of all adolescents is equal.

[0055] The data processing module is used to detect the clean signal after data preprocessing, evaluate the denoising effect based on the signal-to-noise ratio and mean square error, thereby obtaining the expected clean EEG signal and transmitting it to the microstate analysis module.

[0056] In this embodiment, it should be specifically explained that the data processing module processes data as follows: The signal-to-noise ratio (SNR) is calculated using the SNR calculation formula, and the mean square error (MSE) is calculated using the MSE calculation formula; the SNR calculation formula is: The formula for calculating the mean square error is: Where Sn represents the signal-to-noise ratio, Sm represents the mean square error, and fi represents the clean signal. This represents the estimated signal after denoising; the larger the signal-to-noise ratio Sn and the smaller the mean square error Sm, the better the signal denoising effect.

[0057] Based on the signal-to-noise ratio Sn and the mean square error Sm, the signal-to-noise ratio threshold Sn corresponding to the pre-set denoising effect is determined. 阈 and mean squared error threshold Sm 阈 Perform a comparison and judgment; if Sn is satisfied... <Sn 阈 And Sm <Sm 阈 This indicates that the denoising effect meets expectations, and the clean EEG signal is transmitted to the microstate analysis module; if Sn is not satisfied... <Sn 阈 And Sm <Sm 阈 This indicates that the denoising effect is not as expected and needs to be repeated until the denoising effect meets expectations; where the pre-set signal-to-noise ratio threshold Sn corresponds to the denoising effect. 阈 and mean squared error threshold Sm 阈 The specific settings can be configured according to the specific circumstances, and this embodiment does not impose any specific limitations on them;

[0058] The microstate analysis module is used to analyze and process pure EEG signals using an improved k-means clustering algorithm. Based on the clustering analysis results, it extracts the average duration, coverage, frequency of occurrence, amplitude of the microstate, and the transition probability from one microstate to another as feature vectors of the clustering results, which are then transmitted to the prognostic model construction module.

[0059] In this embodiment, it should be specifically explained that the specific analysis process of the micro-state analysis module is as follows:

[0060] A1. Determine the sample set; the processing procedure is as follows:

[0061] A11. For each adolescent's EEG signal, the global field power (GFP) of 160 seconds of data is extracted to obtain a continuous topographic map of GFP. K-means clustering initialization analysis is performed to obtain the initial cluster centers. Here, GFP represents the standard deviation of all electrode voltage values ​​of the topographic map at a certain time point, which is used to describe the intensity of the electric field of a certain topographic map. The initialization analysis is random, so 20 different initializations were performed, and the optimal result was taken. After obtaining the optimal cluster, each time window can be assigned to a cluster, and each cluster is a microstate category.

[0062] A12. After obtaining the initial cluster centers of each adolescent, put all the initial cluster centers of adolescents together, and each initial cluster center is a sample set L, L = {x1, x2, ..., xn};

[0063] A2. Perform spatial clustering analysis on the GFP peak topographic map of the entire sample set; the processing process is as follows:

[0064] A21. Calculate the sample information entropy, and select the cluster that can minimize the information entropy and divide it into two clusters; the calculation formula of the sample information entropy is: where D kl (f m (x)g q (x)) can measure the difference in the point distribution between cluster m and cluster q. When the information entropy is used in EEG data classification, the similarity between the two clusters can be judged by the magnitude of the information entropy. When the information entropy is large, the distribution difference between the two clusters is large, and the classification effect is good;

[0065] A22. Determine the global clustering center, and select the point with a larger joint probability as the global clustering center point;

[0066] A23. Calculate the goodness S of the clustering effect for the existing clusters; the calculation formula of the goodness of the clustering effect is: where X is the longitude, Y is the latitude, p(x, y) is the longitude and latitude of the peak point, p(x) is the probability of the longitude occurrence, p(y) is the probability of the latitude occurrence, p(x, y), p(x), and p(y) can be obtained from the original EEG data, D kl (f m (x)g q (x)) represents the sample information entropy, w represents the proportionality coefficient, and the magnitude of the proportionality coefficient is a specific value obtained by quantifying each parameter for subsequent comparison. Regarding the magnitude of the proportionality coefficient, as long as it does not affect the proportional relationship between the parameter and the quantified value;

[0067] A24. Set the judgment value H and make a judgment comparison with S; the judgment value H can be set specifically according to the specific situation, and this embodiment does not make a specific description on this;

[0068] A25. Repeat steps A23, A24, and A25. If S < H, it means that the clustering effect this time is better than the previous clustering effect, and continue to iterate; otherwise, stop the iteration and return K;

[0069] A3. Determine the clustering result according to the optimal clustering number K value; among them, the clustering result includes microstate A, microstate B, microstate C, microstate D, microstate λ, λ = 1, 2... n;

[0070] A4. After assigning a microstate category to each time window, a one-dimensional discrete sequence of microstate categories changing over time is obtained; the average duration, coverage, frequency of occurrence, amplitude of microstates, and the transition probability from a certain microstate to other microstates are extracted from the microstates as feature vectors of the clustering results and transmitted to the prognostic model construction module.

[0071] The prognostic model construction module is used to receive data transmitted by the microstate analysis module, use one-way ANOVA and Spearman correlation to compare the resting state microstate parameters between groups, and construct a correlation prognostic model; wherein the resting state microstate parameters specifically include the feature vector of the clustering results.

[0072] In this embodiment, it should be specifically explained that the use of one-way ANOVA and Spearman correlation to compare the resting state micro-state parameters among the groups further analyzes the correlation between the significantly different micro-state parameters in the patient groups and the severity of depression and self-harm data.

[0073] In this embodiment, it should be specifically explained that the specific processing procedure of the prognostic model construction module is as follows:

[0074] C1. Use one-way ANOVA to calculate the influence of the clustering result's feature vector on the microstate category; the specific calculation formula for one-way ANOVA is: Y(ij)=u(i)+ε(ij), where Y(ij) represents the j-th observation in the i-th treatment, u(i) represents the mean of the i-th treatment, and ε(ij) represents the random error of the j-th observation in the i-th treatment;

[0075] C2. Calculate the correlation coefficient between the resting state microstates of each group according to the correlation coefficient calculation formula. Where η represents the correlation coefficient, xi and yi represent the corresponding elements in the variables, i.e., the feature vectors, and x and y represent the sample means corresponding to x and y;

[0076] C3. Based on the data obtained in steps C1 and C2, perform a significance test; generally, P < 0.05 is considered significant, and P < 0.01 is considered highly significant; the steps for the significance test are as follows:

[0077] C31. Assume that the two variables have no significant linear relationship, that is, the two variables are zero-correlation;

[0078] C32. Construct a new statistic t. Where η represents the correlation coefficient and n represents the sample size;

[0079] C33. Calculate the probability P value corresponding to the η distribution. If P < 0.05, it indicates that there is a significant linear correlation between the two variables. The calculation of the probability P value is a prior art technique, so it is not specifically explained in this embodiment.

[0080] Specifically, the correlation between significantly different microstate parameters in the patient group and depression severity and self-harm data was significant. Depression severity and self-harm data were significantly correlated, and positively correlated with the occurrence of microstate D, but not with the occurrence of other microstates. Therefore, microstate D can be used as assessment data, and the average duration, coverage, frequency of occurrence, amplitude of microstate D, and probability of transition from microstate to other microstates can be used as assessment parameters.

[0081] The emotion recognition module is used to convert microstates into emotional states based on the correlation prognostic model, associate specific microstates or microstate sequences with a certain emotional state to obtain emotional factors, and transmit them to the efficacy evaluation module.

[0082] In this embodiment, it should be specifically explained that the specific recognition method of the emotion recognition module is as follows: based on the correlation prognostic model of the prognostic model construction module, the microstate D is substituted into the emotional state and transformed into an emotional state. The specific microstate or microstate sequence is associated with a certain emotional state to obtain the emotion factor ω. The emotion factor ω is obtained by analyzing the emotional state based on the microstate D through statistical analysis.

[0083] The efficacy assessment module is used to analyze and predict the risk of relapse based on assessment parameters and emotional factors, calculate the efficacy assessment index, and obtain the efficacy assessment results of adolescent depression with suicidal attempts in the adolescents being tested.

[0084] In this embodiment, it should be specifically explained that the specific calculation formula for the efficacy evaluation index is as follows: Where б represents the efficacy assessment index, Ti, Mi, fi, 1 / a Let represent the average duration, coverage, frequency of occurrence, amplitude of microstate D, and probability of transition from one microstate to another, respectively, and ω represent the emotion factor.

[0085] Based on the adolescent's micro-state D over a week, and combined with the efficacy evaluation index calculation formula, the efficacy is evaluated, and an efficacy trend graph is plotted over time. If the efficacy trend graph shows an upward trend, the adolescent's efficacy is good; if the efficacy trend graph shows a downward trend, the adolescent's efficacy has a risk of relapse, and the treatment plan needs to be adjusted in time.

[0086] The human-computer interaction module is used to intuitively present the emotional state, efficacy evaluation index, and efficacy evaluation results to the user through the terminal's user interface; the user interface includes a real-time display of the emotional state, efficacy evaluation index, efficacy evaluation results, and a trend chart of the historical emotional state; the user includes the adolescent himself or his guardian.

[0087] It should be specifically noted in this embodiment that the human-computer interaction module interacts through the user interface of the terminal, which is a traditional information transmission process. The information transmission methods and information security methods involved are existing technologies, so they are not specifically described in this embodiment. In addition, all information data designed in this embodiment is stored in the information storage module.

[0088] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A treatment efficacy assessment system for adolescent depressive patients based on EEG microstates, characterized in that: include: Data acquisition module: Used to collect brain signals from adolescents using non-invasive EEG signal acquisition equipment and transmit the data to the data preprocessing module; Data preprocessing module: used to filter and denoise the acquired EEG signals to obtain clean signals, and then transmit them to the data processing module; Data processing module: used to detect the clean signal after data preprocessing, evaluate the denoising effect based on the signal-to-noise ratio and mean square error, thereby obtaining the expected clean EEG signal and transmitting it to the microstate analysis module; Microstate analysis module: Used to analyze and process pure EEG signals using an improved k-means clustering algorithm. Based on the clustering analysis results, it extracts the average duration, coverage, frequency of occurrence, amplitude of different EEG microstates, and the transition probability from one microstate to another as feature vectors of the clustering results, and transmits them to the prognostic model construction module. Prognostic model building module: This module receives data from the microstate analysis module, uses one-way ANOVA and Spearman correlation to compare the resting state microstate parameters among groups, and builds a prognostic model. The emotion recognition module is used to convert microstates into emotional states based on the correlation prognostic model, associate specific microstates or microstate sequences with a certain emotional state to obtain emotional factors, and transmit them to the efficacy evaluation module. The efficacy assessment module is used to analyze and predict the risk of relapse based on assessment parameters and emotional factors, calculate the efficacy assessment index, and obtain the efficacy assessment results of adolescent depression with suicidal attempts in the test adolescents. The human-computer interaction module is used to intuitively present the emotional state, efficacy assessment index, and efficacy assessment results to the user through the terminal's user interface.

2. The system for evaluating the efficacy of treatment in adolescent depressive patients based on EEG microstates according to claim 1, characterized in that: The specific data acquisition method of the data acquisition module is as follows: During the acquisition period, in a quiet, light-proof room with minimal electromagnetic interference, the adolescent to be tested is placed in a resting state, and a non-invasive EEG signal acquisition device is worn on the head for 5 minutes according to the international 10-20 system electrode placement standard, to collect EEG signals from the central and occipital regions; the acquisition frequency is a fixed time node every day, and the acquisition period is one week; the resting state specifically refers to the adolescent being awake, quiet, and with eyes closed.

3. The efficacy evaluation system for adolescent depressive patients based on EEG microstates according to claim 1, characterized in that: The specific processing procedure of the data preprocessing module is as follows: Bandpass filtering of 1-30Hz is applied to the signals acquired from 64 channels to obtain four frequency bands for each subject's EEG signal: δ (0.5-4Hz), θ (4-8Hz), α (8-13Hz), and β (13-30Hz). Wavelet denoising is then used to denoise the data. Specifically, the wavelet denoising method is based on the characteristic that noise and signal have different intensity distributions in wavelet decomposition coefficients across different frequency bands. The wavelet coefficients corresponding to noise in each frequency band are removed, retaining the wavelet decomposition coefficients of the original signal. Then, wavelet reconstruction is performed on the processed coefficients to obtain a clean signal. The specific formula for wavelet denoising is: τ represents the time shift, a represents the scaling factor, and f(t) represents the input time-domain signal. This represents the wavelet basis function.

4. The system for evaluating the efficacy of treatment in adolescent depressive patients based on EEG microstates according to claim 1, characterized in that: The pure signal was specifically measured by taking 80 seconds forward and backward from the center of the EEG signal time series, removing the first and last segments of the data, and retaining only the middle effective part, resulting in 160 seconds of data.

5. The system for evaluating the efficacy of treatment in adolescent depressive patients based on EEG microstates according to claim 1, characterized in that: The specific processing method of the data processing module is as follows: calculate the signal-to-noise ratio (SNR) according to the SNR calculation formula, and calculate the mean square error (MSE) according to the MSE calculation formula; the SNR calculation formula is: The formula for calculating the mean square error is: Where Sn represents the signal-to-noise ratio, Sm represents the mean square error, and fi represents the clean signal. This represents the estimated signal after denoising; the larger the signal-to-noise ratio Sn and the smaller the mean square error Sm, the better the signal denoising effect. Based on the signal-to-noise ratio Sn and the mean square error Sm, the signal-to-noise ratio threshold Sn corresponding to the pre-set denoising effect is determined. 阈 and mean squared error threshold Sm 阈 Perform a comparison and judgment; if Sn is satisfied... <Sn 阈 And Sm <Sm 阈 This indicates that the denoising effect meets expectations, and the clean EEG signal is transmitted to the microstate analysis module; if Sn is not satisfied... <Sn 阈 And Sm <Sm 阈 This indicates that the noise reduction effect is not as expected and noise reduction processing needs to be performed again until the noise reduction effect meets expectations.

6. The system for evaluating the efficacy of treatment in adolescent depressive patients based on EEG microstates according to claim 1, characterized in that: The specific analysis process of the micro-state analysis module is as follows: A1. Determine the sample set; the processing procedure is as follows: A11. For the EEG signals of each adolescent, the global field power (GFP) of 160 s of data is intercepted, a continuous topographic map of the GFP is obtained, and K-means clustering initialization analysis is performed to obtain the initial clustering centers. Here, GFP represents the standard deviation of the voltage values of all electrodes in the topographic map at a certain time point, which is used to describe the intensity of the electric field of a certain topographic map. Since the initialization analysis is random, 20 different initializations are performed, and the optimal result is taken. After obtaining the optimal clustering, each time window can be assigned to a cluster, and each cluster is a microstate category. A12. After obtaining the initial clustering centers of each adolescent, the initial clustering centers of all adolescents are put together, and each initial clustering center serves as a sample set L, where L = {x1, x2, ……, xn}. A2. Perform spatial clustering analysis on the GFP peak topographic map of the entire sample set. The processing process is as follows: A21. Calculate the sample information entropy, and select the cluster that minimizes the information entropy to divide the sample into two clusters; the formula for calculating the sample information entropy is: Where D kl (f m (x)g q (x) can measure the difference between the distribution of midpoints in cluster m and cluster q. When information entropy is used in EEG data classification, the similarity between two clusters can be judged by the magnitude of information entropy. When the information entropy is large, the difference between the distributions of the two clusters is large, and the classification effect is better. A22. Determine the global clustering center, and select the point with a relatively large joint probability as the global clustering center point. A23. Calculate the quality S of the clustering effect for the existing clusters; the formula for calculating the quality S of the clustering effect is: Where X is longitude, Y is latitude, p(x, y) are the longitude and latitude of the peak point, p(x) is the probability of longitude occurring, and p(y) is the probability of latitude occurring. p(x, y), p(x), and p(y) can be obtained from the original EEG data. kl (f m (x)g q (x) represents the sample information entropy, and w represents the scaling factor; A24. Set a judgment value H and compare it with S for judgment. A25. Repeat steps A23, A24, and A25. If S < H, it means that the clustering effect this time is better than the previous time, and continue the iteration; otherwise, stop the iteration and return K. A3. According to the optimal clustering number K value, determine the clustering result. The clustering result includes microstate A, microstate B, microstate C, microstate D, microstate λ, where λ = 1, 2 …… n. A4. After assigning a microstate category to each time window, a one-dimensional discrete sequence of the change of the microstate category over time is obtained. The average duration, coverage range, occurrence frequency, amplitude of the EEG microstate, and the transition probability from one microstate to other microstates are extracted as the feature vectors of the clustering result and transmitted to the prognosis model construction module.

7. The efficacy evaluation system for adolescent depressive patients based on EEG microstates according to claim 1, characterized in that: The specific processing process of the prognosis model construction module is as follows: C1. Use one-way analysis of variance to calculate the influence of the feature vectors of the clustering result on this microstate category. The specific calculation formula for one-way analysis of variance is: Y(ij) = u(i) + ε(ij), where Y(ij) represents the jth observation value in the ith treatment, u(i) represents the average value of the ith treatment, and ε(ij) represents the random error of the jth observation value in the ith treatment. C2. Calculate the correlation coefficient between the resting state microstates of each group according to the correlation coefficient calculation formula. Where η represents the correlation coefficient, and xi and yi represent the corresponding elements in the variables, i.e., the feature vectors. Let x and y represent the sample mean. C3. Perform a significance test based on the data obtained in steps C1 and C2.

8. The efficacy evaluation system for adolescent depressive patients based on EEG microstates according to claim 7, characterized in that: The significance test based on the data obtained in steps C1 and C2 is as follows: C31. Assume that there is no significant linear relationship between the two variables, that is, the two variables have zero correlation. C32. Construct a new statistic t. Where η represents the correlation coefficient and n represents the sample size; C33. Calculate the probability P value corresponding to the η distribution.

9. The efficacy evaluation system for adolescent depressive patients based on EEG microstates according to claim 1, characterized in that: The specific formula for calculating the efficacy assessment index is as follows: Where б represents the efficacy assessment index, Ti, Mi, fi, 1 / a Let represent the average duration, coverage, frequency of occurrence, amplitude of microstate D, and probability of transition from one microstate to another, respectively, and ω represent the emotion factor. According to the microstate D of adolescents in a week, combined with the efficacy evaluation index calculation formula, the efficacy is evaluated, and a efficacy trend chart is drawn with time as the unit. If the efficacy trend chart shows an upward trend, the efficacy of the adolescent is good; if the efficacy trend chart shows a downward trend, the adolescent has a risk of recurrence of the disease, and the treatment plan needs to be adjusted in time.

Images