Active psychological problem symptom detection method and system based on electroencephalogram data detection
Through the preprocessing of EEG data and optimization of neural network models, the problems of diagnosis accuracy and large-scale detection of mental illnesses in the existing technology are solved, more accurate mental health monitoring is achieved, and the detection success rate and system reliability are improved. sex.
Patent Information
- Application Number
- CN202411082741.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-08
- Publication Date
- 2024-12-06
AI Technical Summary
The prior art has problems with accuracy in the diagnosis of patients with insufficient early or psychological symptoms, which depends on the patient's complaint and the therapist's subjective judgment, and cannot support the detection of large-scale psychological problems.
By obtaining the subject's EEG data in the preset monitoring mode for preprocessing, and using neural network algorithms to establish a predictive model, identifying the symptoms of mental health problems, and optimizing the model with the subject's feedback to reduce subjective judgment dependence.
It provides a more comprehensive and accurate mental health monitoring method, improves the success rate of detection of mental health symptoms, meets the timeliness and accuracy requirements of large-scale screening, reduces detection time, and improves intelligence. and reliability.
Smart Images

Figure CN119073990A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of active psychological problem symptom detection based on electroencephalogram data detection, and in particular to a method and system for active psychological problem symptom detection based on electroencephalogram data detection. Background Art
[0002] Usually, doctors in psychiatric hospitals usually use the diagnostic criteria of the tenth revision of the International Statistical Classification of Diseases and Related Health Problems (ICD-10) or the third edition of the Chinese Classification and Diagnostic Criteria for Mental Disorders (CCMD-3). Patients are required to fill out questionnaires or inquiries, and the diagnosis is made based on the patient's symptoms and medical history. This leads to the following problems: patients with early or unclear symptoms of mental illness cannot be diagnosed in a timely manner; the diagnosis of mental illness depends on the patient's chief complaint - if the patient is too young to fill out the questionnaire accurately or his or her guardian fills out the questionnaire, the result may deviate greatly from the patient's actual problem; if the patient lacks the ability to express himself or herself for some reason and cannot express himself or herself clearly or fill out the questionnaire accurately, the result may deviate greatly; the patient's chief complaint is the patient's own feeling, which is non-figurative and deviates from the actual problem of brain function expressed by the data of the nervous system; the accuracy of the data cannot be guaranteed in large-scale mental health screening.
[0003] QEEG can quantify brain data and provide difference values, but in the standard QEEG process, the detection of specific psychological problems is highly dependent on the therapist's subjective judgment, so it requires a very high level of experience and ability from the therapist, and it also cannot support the detection of large-scale or batch psychological problems.
[0004] Therefore, the prior art needs to be further developed. Summary of the invention
[0005] The purpose of the present invention is to overcome the above-mentioned technical deficiencies and provide a method and system for proactively detecting symptoms of psychological problems based on EEG data detection, so as to solve the problems existing in the prior art.
[0006] To achieve the above technical objectives, according to a first aspect of the present invention, the present invention provides a method for proactively detecting symptoms of psychological problems based on EEG data detection, comprising:
[0007] S100, obtaining EEG data of a subject in a preset monitoring mode, and performing a first preprocessing on the obtained EEG data;
[0008] S200, using a neural network algorithm to train the EEG data after the first preprocessing, establishing a neural network prediction model, and inputting the data to be tested into the neural network prediction model to identify preset mental health problem symptoms;
[0009] S300. Correct the preset mental health problem symptoms identified by the neural network prediction model using the actual mental health problem symptoms fed back by the subjects, and input the corrected preset mental health problem symptom data into the neural network prediction model for retraining to complete the optimization of the neural network prediction model.
[0010] Specifically, obtaining the EEG data of the subject in a preset monitoring mode includes:
[0011] Collect EEG data of subjects with their eyes open and closed in a preset monitoring mode;
[0012] The acquisition time of the EEG data in the two states of eyes being open and eyes being closed is set as the first preset time.
[0013] Specifically, the preset monitoring modes include 4-channel monitoring and 19-channel monitoring;
[0014] The brain points monitored by the 4 channels include: Fz, Pz, C3, and C4;
[0015] The brain points monitored by the 19 channels include: Fp1, Fp2, Fz, F3, F4, F7, F8, Cz, C3, C4, P3, P4, Pz, T3, T4, T5, T6, O1, and O2.
[0016] Specifically, the first preprocessing of the acquired EEG data includes:
[0017] Point elimination processing, eliminating outliers or noise through a first preset algorithm;
[0018] Eliminating trend items, removing systematic trends or changes through a second preset algorithm;
[0019] Further eliminating artifacts by using a third preset algorithm, wherein the sources of the artifacts include electrooculography, electromyography, and electrocardiography;
[0020] The EEG data is frequency-divided according to the first frequency to obtain EEG data of each frequency band.
[0021] Specifically, the first preset algorithm includes the Dixon criterion;
[0022] The second preset algorithm includes a least squares fitting method and a wavelet decomposition method;
[0023] The third preset algorithm includes a filter method and a deep learning method;
[0024] The first frequencies include Delta, Theta, Alpha, Low Beta, Beta, High Beta, Gamma, Alpha1, and Alpha2.
[0025] Specifically, the first preprocessing of the acquired EEG data further includes:
[0026] Quantify the absolute value of the EEG data of each preset brain area and the relative value between each preset brain area, and calculate the difference with the standard value of the same gender and age in the preset brain area;
[0027] The EEG data of each preset brain area is intercepted according to the preset recording time.
[0028] Specifically, the method of using a neural network algorithm to train the EEG data after the first preprocessing to establish a neural network prediction model includes:
[0029] Classify and group the EEG data of each preset brain region and frequency band according to their relevance to the preset mental health problem symptoms;
[0030] Performing logical relationship processing on the EEG data through a first association analysis and a first weight analysis, respectively calculating a first weighted sum of absolute values and a first weighted sum of relative values of the EEG data of each preset brain region and each frequency band in the eyes-open state and the eyes-closed state;
[0031] Determine whether the first weighted sum of the absolute values and the first weighted sum of the relative values of the EEG data of each preset brain area and each frequency band in the eyes open and eyes closed states are greater than or equal to a first preset threshold, and determine whether the subject has preset mental health problem symptoms based on the judgment result.
[0032] Specifically, the determining whether the first weighted sum of the absolute values and the first weighted sum of the relative values of the electroencephalogram data of each preset brain region and each frequency band in the eyes-open state and the eyes-closed state is greater than or equal to a first preset threshold value, and determining whether the subject has a preset mental health problem symptom according to the determination result, includes:
[0033] If any first weighted sum of the absolute value and the relative value of the EEG data of each preset brain region and each frequency band in the eyes-open and eyes-closed states is greater than or equal to a first preset threshold, it is determined that the subject has a preset mental health problem symptom;
[0034] If the first weighted sum of the absolute value and relative value of the EEG data of each preset brain area and each frequency band in the eyes open and eyes closed states are both less than the first preset threshold, it is determined that the subject does not have the preset mental health problem symptoms.
[0035] Specifically, the S300 includes:
[0036] Compare the pre-set mental health problem symptoms identified by the neural network prediction model with the actual mental health problem symptoms reported by the subjects;
[0037] If the comparison results are inconsistent, the actual mental health problem symptom data reported by the subjects will be input into the neural network prediction model for the first association analysis and the first weight analysis to optimize the neural network prediction model.
[0038] According to a second aspect of the present invention, there is provided a proactive psychological problem symptom detection system based on EEG data detection, comprising:
[0039] Acquisition module: used for acquiring EEG data of the subject in a preset monitoring mode, and performing a first preprocessing on the acquired EEG data;
[0040] A control module is used to perform a first preprocessing on the acquired EEG data, or to use a neural network algorithm to train the EEG data after the first preprocessing, establish a neural network prediction model, and input the test data into the neural network prediction model to identify preset mental health problem symptoms; or to use the actual mental health problem symptoms fed back by the subjects to correct the preset mental health problem symptoms identified by the neural network prediction model, and input the corrected preset mental health problem symptom data into the neural network prediction model for retraining to complete the optimization of the neural network prediction model.
[0041] Beneficial effects:
[0042] The present invention obtains EEG data of a subject in a preset monitoring mode, performs a first preprocessing on the obtained EEG data, uses a neural network algorithm to train the EEG data after the first preprocessing, establishes a neural network prediction model, inputs the data to be tested into the neural network prediction model to identify preset mental health problem symptoms, uses the actual mental health problem symptoms fed back by the subject to correct the preset mental health problem symptoms identified by the neural network prediction model, and inputs the corrected preset mental health problem symptom data into the neural network prediction model for retraining to complete the optimization of the neural network prediction model. Compared with traditional scale detection and QEEG detection technology, the present invention provides a more comprehensive and accurate scientific mental health monitoring method, which is not limited by the subjective expression ability of the subject, but scientifically reflects the real mental health status of the subject through big data analysis, replaces manual detection of mental health problem symptoms, reduces reliance on therapists' subjective judgment, improves the detection success rate of mental health problem symptoms through algorithm optimization, and can meet the timeliness and accuracy requirements of large-scale mental health screening, greatly saves detection time, and greatly improves the intelligence, usability and reliability of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS
[0043] Figure 1 It is a flow chart of a method for proactively detecting symptoms of psychological problems based on EEG data detection provided in a specific embodiment of the present invention;
[0044] Figure 2 It is a schematic diagram of the system composition of the proactive psychological problem symptom detection system based on EEG data detection provided in a specific embodiment of the present invention;
[0045] Figure 3 It is a schematic diagram of a proactive psychological problem symptom detection process based on EEG data detection provided in a specific embodiment of the present invention;
[0046] Figure 4 is a schematic diagram of electroencephalogram (EEG) waves of artifacts in EEG data provided in a specific embodiment of the present invention;
[0047] Figure 5 It is a schematic diagram of the international 10-20 brain region division provided in a specific embodiment of the present invention. DETAILED DESCRIPTION
[0048] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is clearly and completely described below in conjunction with the accompanying drawings of the present invention. Based on the embodiments in this application, other similar embodiments obtained by ordinary technicians in this field without making creative work should all fall within the scope of protection of this application. In addition, the directional words mentioned in the following embodiments, such as "up", "down", "left", "right", etc., are only reference to the directions of the accompanying drawings. Therefore, the directional words used are used to illustrate rather than limit the invention.
[0049] The present invention will be further described below in conjunction with the accompanying drawings and preferred embodiments.
[0050] See also Figure 1 This embodiment provides a method for proactively detecting symptoms of psychological problems based on EEG data detection, including:
[0051] S100, obtaining EEG data of a subject in a preset monitoring mode, and performing a first preprocessing on the obtained EEG data;
[0052] It should be noted here that before step S100, the first preset time, the preset recording time, the first preset threshold, and the preset mental health problem symptoms are preset.
[0053] It can be understood that the present invention does not limit the specific values or specific contents of the first preset time, the preset recording time, the first preset threshold, and the preset mental health problem symptoms, as long as the active psychological problem symptom detection method based on EEG data detection proposed by the present invention can be applied.
[0054] Preferably, the present invention sets the first preset time to 2 minutes, the present invention sets the preset recording time to 10 seconds to 3 minutes and 50 seconds, the present invention sets the first preset threshold to 3, and the present invention sets the preset mental health problem symptom to attention problem. The above settings can further improve the accuracy of the data and the accuracy of the model prediction, and greatly improve the accuracy and reliability of the present invention.
[0055] Specifically, obtaining the EEG data of the subject in a preset monitoring mode includes:
[0056] Collect EEG data of subjects with their eyes open and eyes closed in a preset monitoring mode;
[0057] The acquisition time of the EEG data in the two states of eyes being open and eyes being closed is set as the first preset time.
[0058] Specifically, the preset monitoring modes include 4-channel monitoring and 19-channel monitoring;
[0059] The brain points monitored by the 4 channels include: Fz, Pz, C3, and C4;
[0060] The brain points monitored by the 19 channels include: Fp1, Fp2, Fz, F3, F4, F7, F8, Cz, C3, C4, P3, P4, Pz, T3, T4, T5, T6, O1, and O2.
[0061] It is understandable that the detection of mental health problems requires the use of bilateral ears as reference electrodes. Mental health monitoring provides two monitoring modes, from 4 channels to 19 channels. The 4-channel mode is suitable for preliminary screening or daily monitoring, and the 19-channel mode provides more comprehensive EEG data, which is suitable for in-depth analysis and professional diagnosis. The EEG data is collected from 4 or 19 brain points of the subjects with their eyes open and closed. The EEG data is collected for 2 minutes each when the subjects have their eyes open and closed. The eyes open state helps detect activities related to vision and attention, while the eyes closed state can reflect basic EEG activity. Comparative analysis can reveal possible mental health problems, greatly improving the reliability of the data.
[0062] Specifically, the first preprocessing of the acquired EEG data includes:
[0063] Point elimination processing, eliminating outliers or noise through a first preset algorithm;
[0064] Eliminating trend items, removing systematic trends or changes through a second preset algorithm;
[0065] Further eliminating artifacts by using a third preset algorithm, wherein the sources of the artifacts include electrooculography, electromyography, and electrocardiography;
[0066] The EEG data is frequency-divided according to the first frequency to obtain EEG data of each frequency band.
[0067] It should be further explained that, regarding the point elimination processing, in the QEEG (quantitative electroencephalogram) signal, due to various reasons (such as poor electrode contact, patient movement, equipment noise, etc.), outliers or noise may appear, and these outliers will interfere with the analysis and diagnosis of the EEG, and therefore need to be automatically eliminated by an algorithm. The present invention uses the Dixon criterion to eliminate weaker interference, and then eliminate outliers or noise; regarding the elimination of trend items, in the QEEG data preprocessing, mathematical or statistical methods are used to remove those systematic trends or changes that are not directly related to brain activity itself but may affect the data. The main reason for eliminating trend items is that when wearing an electrode cap, the skin and the electrode are not in sufficient contact, resulting in signal deviation. Common methods are least squares fitting and wavelet decomposition. It should be noted that the above two steps of point elimination processing and trend item elimination are implemented in mature software for processing EEG data.
[0068] Regarding further eliminating artifacts, EEG signals (electroencephalogram signals) are very susceptible to artifacts generated by experimental equipment and participants themselves during the acquisition process. Frequent blinking, tightening of jaw muscles, and violent beating of the heart after high-intensity exercise will all generate high-intensity bioelectricity, which will interfere with the accuracy of EEG data and belong to the category of artifacts. Compared with normal EEG data, Figure 4 As shown, this type of artifact has distinct characteristics: high frequency and large amplitude, which can be eliminated after data comparison by an algorithm. Common sources of artifacts include electrooculogram (EOG), electromyography (EMG), electrocardiogram (ECG), etc., which are removed by filters and deep learning. The brain data is divided according to the following frequencies: Delta, Theta, Alpha, Low Beta, Beta, HighBeta, Gamma, Alpha1, Alpha2. The above frequencies represent different states of the brain, which further improves the reliability of the data source. The above preprocessing of EEG data can further improve the accuracy of the data, which greatly improves the usability and reliability of the present invention.
[0069] Specifically, the first preset algorithm includes the Dixon criterion;
[0070] The second preset algorithm includes a least squares fitting method and a wavelet decomposition method;
[0071] The third preset algorithm includes a filter method and a deep learning method;
[0072] The first frequencies include Delta, Theta, Alpha, Low Beta, Beta, High Beta, Gamma, Alpha1, and Alpha2.
[0073] It is understandable that Delta waves (0.5-4Hz) are usually associated with deep sleep states and are a sign of brain relaxation and unconsciousness. In the awake state, it may also appear in moments of high concentration, such as in meditation; Theta waves (4-8Hz) are associated with light sleep, dream states, and meditation. In the awake state, theta waves may also appear when the attention is distracted or creative thinking; Alpha waves (8-12Hz) are usually produced by the brain when a person is awake but relaxed, such as closing his eyes, deeply relaxing, or thinking internally. Alpha waves are divided into two subtypes: Alpha1 (8-10Hz) is usually associated with a lighter state of relaxation and attention; Alpha2 (10-12Hz): may indicate deeper relaxation and more focused internal thinking; Low Beta waves (12-20Hz): associated with logical thinking, concentration and decision-making processes. The brain usually produces waves of this frequency during daily waking activities; Beta waves (20-30Hz) represent wakefulness, high concentration and daily thinking activities. The frequency of Beta waves will increase in high activity states, such as anxiety or tension; High Beta waves (30-40Hz) are usually associated with strong emotional reactions, tension or overstimulation; Gamma waves (30-100Hz) are associated with rapid information processing, perceptual integration and high cognitive activities, such as problem solving and learning new information. Gamma waves are believed to be associated with synchronous activity in different areas of the brain. These brain wave frequencies help scientists and researchers understand the activity patterns of the brain in different states. They are of great significance in neuroscience, psychology and clinical applications.
[0074] Specifically, the first preprocessing of the acquired EEG data further includes:
[0075] Quantify the absolute value of the EEG data of each preset brain area and the relative value between each preset brain area, and calculate the difference with the standard value of the same gender and age in the preset brain area;
[0076] It is understandable that these differences are used to calculate and judge whether the data of each brain region corresponding to each mental health problem symptom is abnormal, that is, to judge whether the brain region data corresponding to the preset mental health problem is abnormal. The calculation here is completed by comparing with the standard data provided by the brain database. It should be noted that the main data sources of the brain database are from Europe and the United States. Although many studies have shown that brain problems are not related to race and culture, it has been found through a lot of practice that factors such as diet structure and habits, social relationships, family relationships, and other races and cultures are closely related to brain problems. Therefore, in the subsequent training of the neural network prediction model:
[0077] (1) For the preset symptoms of mental health problems, it is necessary to add a frequency band threshold (K value) for each frequency band in each brain region to determine the logical relationship between the EEG data of each brain point and between the points;
[0078] (2) For the preset symptoms of mental health problems, it is necessary to add a first weight for each brain region and each frequency band to determine the actual impact of abnormal EEG data at that brain point on the brain;
[0079] (3) For the preset symptoms of mental health problems, a first preset threshold needs to be added to comprehensively judge the actual impact of abnormal EEG data of all relevant brain points on the brain;
[0080] The EEG data of each preset brain area is intercepted according to the preset recording time.
[0081] It needs to be further explained that if Figure 5 As shown, the preset brain regions are divided according to the international 10-20 brain region system and the brain points monitored by 4 channels or 19 channels, that is, the preset brain regions are brain points, and for the EEG data of each brain region, its absolute value relative to a reference point (usually the entire brain or a reference brain region) is calculated, and this reference point can be regarded as the 0 point; the relative values between the brain regions can be calculated by calculating the ratio of the data of one brain region to the data of another brain region. Brain region A is relative to all brain regions. For example, the absolute difference of brain region A is 2, and the average value of all brain regions is 1, then the relative value of A is 1, and from Existing studies or databases obtain standard values of brain regions for people of the same gender and age group. These standard values are usually obtained by statistically analyzing the EEG data of a large number of people. The EEG data of an individual is compared with the standard values of the same gender and age, and the difference between each brain region is calculated. This can be achieved through simple subtraction, that is, individual value - standard value; by intercepting the EEG data at a preset recording time, it can be ensured that the data of different brain regions are recorded under the same experimental conditions or cognitive state, which can improve the synchronization and comparability between data, greatly improve the data quality, and make subsequent analysis more accurate and reliable.
[0082] S200, using a neural network algorithm to train the EEG data after the first preprocessing to establish a neural network prediction model,
[0083] It should be further explained that the specific steps of using the neural network algorithm to train the first preprocessed EEG data are as follows:
[0084] (1) Data preparation
[0085] Step 1: Organize EEG data and mental health problem symptom data into a structured format;
[0086] Step 2: Define the relationship between symptoms and EEG data waveforms of each brain region based on known information. Taking the four-channel detection of depression as an example, by setting the frequency band threshold (K value) of each frequency band in each brain region, the corresponding values of each brain region are screened out:
[0087] The Alpha value of the left hemisphere of the brain is greater than that of the right hemisphere, that is, if C3>C4+K;
[0088] Theta value in the left hemisphere is greater than that in the right hemisphere, that is, C3>C4+K;
[0089] Alpha inversion in the anterior and posterior brain regions, i.e. Pz>Fz+K;
[0090] The parietal values are high, that is, the sum of all values of Fz, Pz, C3, and C4 is > K;
[0091] Among them, the K value is the frequency band threshold of each frequency band in each brain area. The above K values are different, and the conclusions of the above 4 items also need to be weighted to finally determine the symptoms of mental health problems. At the same time, the data observability of the two states of eyes closed and eyes open is different. For example, when the eyes are closed, the alpha value will increase significantly.
[0092] (2) Feature Engineering
[0093] Features are created to capture the mathematical relationship between each symptom and the EEG waveform at each point in the brain, namely, a first preset threshold and a first weight.
[0094] (3) Model construction
[0095] Step 1: Build an RNN model using LSTM units to learn the relationship between mental health problem symptoms and EEG data;
[0096] Step 2: Define the input shape of the neural network prediction model based on the number of EEG sites and waveform types;
[0097] Step 3: training the model to generate a frequency band threshold (K value) for each frequency band in each brain region, a first weight for each frequency band in each brain region, and a first preset threshold;
[0098] Step 4: Model training and evaluation: divide the data into training set and test set, train the RNN model on the training set, and evaluate its performance on the test set;
[0099] Step 5: Modify the model based on the real feedback from the subjects in the actual test;
[0100] Step 6: Input the test data into the neural network prediction model to identify the preset mental health problem symptoms.
[0101] Specifically, the method of using a neural network algorithm to train the EEG data after the first preprocessing to establish a neural network prediction model includes:
[0102] Classify and group the EEG data of each preset brain region and frequency band according to their relevance to the preset mental health problem symptoms;
[0103] Performing logical relationship processing on the EEG data through a first association analysis and a first weight analysis, respectively calculating a first weighted sum of absolute values and a first weighted sum of relative values of the EEG data of each preset brain region and each frequency band in the eyes-open state and the eyes-closed state;
[0104] It can be understood that the logical relationship processing of the EEG data through the first association analysis and the first weight analysis includes: firstly setting the frequency band threshold (K value) of each frequency band of each brain region, filtering out the corresponding values of each brain region through the frequency band threshold (K value) of each frequency band of each brain region, and then setting the first weight of each frequency band of each brain region, weighted summing the filtered values according to the frequency band, and multiplying the absolute value by the first weight to obtain 4 groups of data of two states of eyes open and eyes closed as well as absolute value and relative value;
[0105] Specifically, the calculation method of the first weighted sum of the absolute values is:
[0106] Calculate the absolute values of all values of each frequency band in each brain area respectively, and sum the absolute values of all values of each frequency band;
[0107] Specifically, it is recorded as the first summation value of each frequency band in each brain region;
[0108] Obtaining the first weight corresponding to each frequency band in each brain region;
[0109] The product of the first sum value and the first weight corresponding to each frequency band is calculated in sequence, and recorded as the first product;
[0110] Sum the first products.
[0111] Specifically, the calculation method of the first weighted sum of the relative values is:
[0112] Calculate the relative values of all numerical values of each frequency band corresponding to each brain region respectively, and sum the relative values of all numerical values of each frequency band;
[0113] Specifically, it is recorded as the second summation value of each frequency band in each brain region;
[0114] Obtaining the first weight corresponding to each frequency band in each brain region;
[0115] The products of the second sum value and the first weight corresponding to each frequency band are calculated in sequence, and recorded as second products;
[0116] Sum the second products.
[0117] Determine whether the first weighted sum of the absolute values and the first weighted sum of the relative values of the EEG data of each preset brain area and each frequency band in the eyes open and eyes closed states are greater than or equal to a first preset threshold, and determine whether the subject has preset mental health problem symptoms based on the judgment result.
[0118] Specifically, the determining whether the first weighted sum of the absolute values and the first weighted sum of the relative values of the electroencephalogram data of each preset brain region and each frequency band in the eyes-open state and the eyes-closed state is greater than or equal to a first preset threshold value, and determining whether the subject has a preset mental health problem symptom according to the determination result, includes:
[0119] If any first weighted sum of the absolute value and the relative value of the EEG data of each preset brain region and each frequency band in the eyes-open and eyes-closed states is greater than or equal to a first preset threshold, it is determined that the subject has a preset mental health problem symptom;
[0120] If the first weighted sum of the absolute value and the relative value of the EEG data of each preset brain region and each frequency band in the eyes-open and eyes-closed states are both less than the first preset threshold, it is determined that the subject does not have the preset mental health problem symptoms;
[0121] It can be understood that the first preset threshold is set according to the preset mental health problem symptoms. The four sets of data calculated are compared with the first preset thresholds of the preset mental health problem symptoms respectively. If the first weighted sum value of any set of data is greater than or equal to 3, it is considered that the preset mental health problem symptoms exist. This means that if in any state (eyes open or eyes closed), whether the first weighted sum calculated using absolute value or relative value, any first weighted sum is greater than or equal to 3, it can be determined that the preset mental health problem symptoms exist, that is, in this preferred embodiment, if the first weighted sum value of any set of data is greater than or equal to 3, it is considered that there is an attention problem; if the first weighted sum values of all data are less than 3, it is considered that there is no attention problem.
[0122] S300. Correct the preset mental health problem symptoms identified by the neural network prediction model using the actual mental health problem symptoms fed back by the subjects, and input the corrected preset mental health problem symptom data into the neural network prediction model for retraining to complete the optimization of the neural network prediction model.
[0123] Specifically, the S300 includes:
[0124] Compare the pre-set mental health problem symptoms identified by the neural network prediction model with the actual mental health problem symptoms reported by the subjects;
[0125] If the comparison results are inconsistent, the actual mental health problem symptom data reported by the subjects will be input into the neural network prediction model for the first association analysis and the first weight analysis to optimize the neural network prediction model.
[0126] It should be further explained here that the preset mental health problem symptoms and conclusions obtained through monitoring are confirmed by the subjects, and then the training weights are set for the data based on the subjects' feedback. After a period of accumulation, they are re-invested in the neural network to optimize the first preset threshold for the detection of mental health problem symptoms, the first weight of each frequency band, and even the brain region correlation. As time goes by, the system will collect more data, including new subjects and new symptom manifestations. The accumulation of data helps to improve the generalization ability and prediction accuracy of the model, and greatly improves the accuracy and range of symptom detection of mental health problems. The specific steps are as follows:
[0127] (1) Result confirmation
[0128] The system first analyzes the subject’s mental health data through a neural network to draw conclusions about potential symptoms;
[0129] These conclusions need to be communicated and confirmed with the subjects to ensure that the interpretation of the data is accurate and consistent with the subjects’ actual experience;
[0130] At the same time, the actual experiences of other unmanifested symptoms can also be confirmed with the subjects to make the data more complete.
[0131] (2) Data weighting
[0132] Based on the subjects’ responses, the data is weighted, meaning that some data points or features may be more important than others, and therefore have a greater influence in the model;
[0133] The purpose of weighting is to make the model focus more on information that is more critical for diagnosing specific mental health conditions.
[0134] (3) Re-enter the neural network
[0135] Re-input the weighted and accumulated data into the neural network;
[0136] This step is to allow the model to learn and adjust its internal parameters from new data.
[0137] (4) Optimizing the first preset threshold, the first weight of each frequency band, and the brain region correlation
[0138] By training on new data, the system can optimize the first preset threshold for detecting symptoms of mental health problems, that is, determining which signals are strong enough to be considered signs of symptoms;
[0139] At the same time, the first weights of each frequency band of different data features can also be adjusted to ensure that the model considers all relevant information in a more balanced manner;
[0140] In addition, models of connectivity between different brain regions can be optimized to better understand the distribution and interaction of mental health problem symptoms in the brain.
[0141] The working principle of the present invention is described below by means of specific examples:
[0142] Figure 3 The implementation process of the technical solution of the present invention is as follows:
[0143] 1. Data Collection
[0144] Detection of mental health problems requires the use of bilateral ears as reference electrodes. Mental health monitoring provides two monitoring modes, from 4 channels to 19 channels, which collect the electroencephalogram (EEG) of 4 or 19 brain points of the subjects with eyes open (EO) and eyes closed (EC). All brain points include: Fp1, Fp2, Fz, F3, F4, F7, F8, Cz, C3, C4, P3, P4, Pz, T3, T4, T5, T6, O1, O2. The brain points monitored by 4 channels are Fz, Pz, C3, C4, and 19 channels cover all brain points. In Table 1, 1-11 is the 4-channel monitoring range, and 1-34 is the 19-channel monitoring range.
[0145] Table 1 Classification of symptoms of mental health problems
[0146]
[0147] 2. Data Extraction
[0148] The brain data collection time is 4 minutes, and the brain bioelectric data is collected for 2 minutes each time when the subject opens his eyes and closes his eyes. The system will trim the data, including:
[0149] (1) Point removal: In the QEEG signal, outliers or noise may appear due to various reasons (such as poor electrode contact, patient movement, equipment noise, etc.). These outliers will interfere with the analysis and diagnosis of the EEG and need to be automatically removed by the algorithm. The common method is to use the Dixon criterion to remove weak interference;
[0150] (2) Eliminating trend items: In QEEG data preprocessing, mathematical or statistical methods are used to remove systematic trends or changes that are not directly related to brain activity itself but may affect the data. Common methods include least squares fitting and wavelet decomposition.
[0151] (3) Further eliminate artifacts: EEG signals are easily affected by artifacts generated by experimental equipment and participants themselves during the acquisition process. Common sources of artifacts include electrooculogram (EOG), electromyography (EMG), electrocardiogram (ECG), etc., which can be removed through filters and deep learning;
[0152] (4) Brain data were divided into the following frequencies: Delta, Theta, Alpha, Low Beta, Beta, High Beta, Gamma, Alpha 1, Alpha 2;
[0153] (5) Quantify the data of each brain region relative to the zero point (i.e., absolute value), as well as the relative value between each brain region, and calculate the difference with the standard value of the brain of the same gender and age;
[0154] (6) The data is intercepted according to the recording time, generally taking data from 10 seconds to 3 minutes and 50 seconds;
[0155] (7) The final extracted data set contains 160 or 760 sets of data.
[0156] 3. Data Analysis
[0157] The analysis of brain data in mental health monitoring is based on deep learning. Therefore, by training a large amount of brain data using neural networks, the logical relationship between various symptoms and brain regions and frequencies can be derived.
[0158] Taking the detection of attention problems in item 4 in Table 1 as an example, the specific steps are as follows:
[0159] (1) The data of each brain region and frequency band were classified and grouped according to the correlation of symptoms in Table 1;
[0160] The brain areas and frequency bands associated with attention problems are Fz (Theta, Al pha 2, High Beta), Pz (Alpha), C3 (Theta, Al pha), C4 (Theta, Al pha);
[0161] (2) performing logical relationship processing on the collected EEG data through a first correlation analysis and a first weight relationship;
[0162] First, the thresholds of each frequency band are clearly defined as Theta (1), Alpha 2 (2), Alpha (-2), and High Beta (1). The corresponding values of each brain region are screened out through the frequency band thresholds (K values) of each brain region.
[0163] Set the first weights of the above frequency bands to 1, 1, 0.5, and 1 respectively, sum the values filtered out in the previous step according to the frequency bands, and multiply the absolute value with each first weight, and there are 4 sets of data in total, including the eyes open and eyes closed states, as well as the absolute value and relative value;
[0164] (3) The four sets of data are compared with the first preset threshold value of attention problem respectively. If the final sum total value is greater than or equal to 3, it is the criterion for brain attention problem.
[0165] (4) Detection of symptoms of mental health problems: Detection can be done after the data processing process and results are manually confirmed;
[0166] (5) Result confirmation and optimization: After the symptoms and conclusions of mental health problems obtained through monitoring are confirmed by the subjects, training weights are set for the data based on the subjects’ feedback. After a period of accumulation, the data is re-invested in the neural network to optimize the first preset threshold for detecting mental health problem symptoms, the first weights of each brain region and each frequency band, and even the correlation between brain regions, to improve the accuracy and range of symptom detection. Currently, this system can monitor 34 mental health problems and symptoms, with a detection accuracy rate of 95%.
[0167] It should be noted here that the present invention obtains the EEG data of the subject in a preset monitoring mode, performs a first preprocessing on the acquired EEG data, uses a neural network algorithm to train the EEG data after the first preprocessing, establishes a neural network prediction model, and inputs the test data into the neural network prediction model to identify preset mental health problem symptoms, uses the actual mental health problem symptoms fed back by the subject to correct the preset mental health problem symptoms identified by the neural network prediction model, and inputs the corrected preset mental health problem symptom data into the neural network prediction model for retraining to complete the optimization of the neural network prediction model. Compared with traditional scale detection and QEEG detection technology, the present invention provides a more comprehensive and accurate scientific mental health monitoring method, which is not limited by the subjective expression ability of the subject, but scientifically reflects the real mental health status of the subject through big data analysis, replaces manual detection of mental health problem symptoms, reduces the reliance on the subjective judgment of the therapist, and can improve the detection success rate of mental health problem symptoms through algorithm optimization, and can meet the timeliness and accuracy requirements of large-scale mental health screening, greatly saves detection time, and greatly improves the intelligence, usability and reliability of the present invention.
[0168] See also Figure 2 This embodiment provides a proactive psychological problem symptom detection system based on EEG data detection, the system comprising:
[0169] Acquisition module 100: used to acquire EEG data of a subject in a preset monitoring mode, and perform a first preprocessing on the acquired EEG data;
[0170] Control module 200: used to perform a first preprocessing on the acquired EEG data, or to use a neural network algorithm to train the EEG data after the first preprocessing, establish a neural network prediction model, and input the test data into the neural network prediction model to identify preset mental health problem symptoms; or to use the actual mental health problem symptoms fed back by the subject to correct the preset mental health problem symptoms identified by the neural network prediction model, and input the corrected preset mental health problem symptom data into the neural network prediction model for retraining, thereby completing the optimization of the neural network prediction model.
[0171] It should be noted here that the present invention obtains the EEG data of the subject in a preset monitoring mode, performs a first preprocessing on the acquired EEG data, uses a neural network algorithm to train the EEG data after the first preprocessing, establishes a neural network prediction model, and inputs the test data into the neural network prediction model to identify preset mental health problem symptoms, uses the actual mental health problem symptoms fed back by the subject to correct the preset mental health problem symptoms identified by the neural network prediction model, and inputs the corrected preset mental health problem symptom data into the neural network prediction model for retraining to complete the optimization of the neural network prediction model. Compared with traditional scale detection and QEEG detection technology, the present invention provides a more comprehensive and accurate scientific mental health monitoring method, which is not limited by the subjective expression ability of the subject, but scientifically reflects the real mental health status of the subject through big data analysis, replaces manual detection of mental health problem symptoms, reduces the reliance on the subjective judgment of the therapist, and can improve the detection success rate of mental health problem symptoms through algorithm optimization, and can meet the timeliness and accuracy requirements of large-scale mental health screening, greatly saves detection time, and greatly improves the intelligence, usability and reliability of the present invention.
[0172] In a preferred embodiment, the present application further provides an electronic device, the electronic device comprising:
[0173] A memory; and a processor, wherein the memory stores computer-readable instructions, and when the computer-readable instructions are executed by the processor, the method for detecting active psychological problem symptoms based on EEG data detection is implemented. The computer device can be broadly a server, a terminal, or any other electronic device with necessary computing and / or processing capabilities. In one embodiment, the computer device may include a processor, a memory, a network interface, a communication interface, etc. connected via a system bus. The processor of the computer device can be used to provide necessary computing, processing and / or control capabilities. The memory of the computer device may include a non-volatile storage medium and an internal memory. An operating system, a computer program, etc. may be stored in or on the non-volatile storage medium. The internal memory can provide an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface and the communication interface of the computer device can be used to connect and communicate with external devices via a network. When the computer program is executed by the processor, the steps of the method of the present invention are executed.
[0174] The present invention may be implemented as a computer-readable storage medium having a computer program stored thereon, which causes the steps of the method of an embodiment of the present invention to be executed when executed by a processor. In one embodiment, the computer program is distributed on a plurality of computer devices or processors coupled to a network so that the computer program is stored, accessed and executed in a distributed manner by one or more computer devices or processors. A single method step / operation, or two or more method steps / operations, may be performed by a single computer device or processor or by two or more computer devices or processors. One or more method steps / operations may be performed by one or more computer devices or processors, and one or more other method steps / operations may be performed by one or more other computer devices or processors. One or more computer devices or processors may perform a single method step / operation, or perform two or more method steps / operations.
[0175] It will be appreciated by those skilled in the art that the method steps of the present invention may be performed by instructing related hardware such as a computer device or a processor through a computer program, and the computer program may be stored in a non-temporary computer-readable storage medium, which causes the steps of the present invention to be performed when the computer program is executed. Depending on the circumstances, any reference to memory, storage, database, or other media herein may include non-volatile and / or volatile memory. Examples of non-volatile memory include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), flash memory, magnetic tape, floppy disk, magneto-optical data storage device, optical data storage device, hard disk, solid state disk, etc. Examples of volatile memory include random access memory (RAM), external cache memory, etc.
[0176] It should be noted here that the present invention obtains the EEG data of the subject in a preset monitoring mode, performs a first preprocessing on the acquired EEG data, uses a neural network algorithm to train the EEG data after the first preprocessing, establishes a neural network prediction model, and inputs the test data into the neural network prediction model to identify preset mental health problem symptoms, uses the actual mental health problem symptoms fed back by the subject to correct the preset mental health problem symptoms identified by the neural network prediction model, and inputs the corrected preset mental health problem symptom data into the neural network prediction model for retraining to complete the optimization of the neural network prediction model. Compared with traditional scale detection and QEEG detection technology, the present invention provides a more comprehensive and accurate scientific mental health monitoring method, which is not limited by the subjective expression ability of the subject, but scientifically reflects the real mental health status of the subject through big data analysis, replaces manual detection of mental health problem symptoms, reduces the reliance on the subjective judgment of the therapist, and can improve the detection success rate of mental health problem symptoms through algorithm optimization, and can meet the timeliness and accuracy requirements of large-scale mental health screening, greatly saves detection time, and greatly improves the intelligence, usability and reliability of the present invention.
[0177] The various technical features described above can be combined arbitrarily. Although all possible combinations of these technical features are not described, any combination of these technical features should be considered to be covered by this specification as long as there is no contradiction in such combination.
[0178] The specific implementation of the present invention described above does not constitute a limitation on the protection scope of the present invention. Any other corresponding changes and modifications made based on the technical concept of the present invention should be included in the protection scope of the claims of the present invention.
Claims
1. A method for detecting active psychological problem symptoms based on EEG data detection, characterized in that: include: S100, obtaining EEG data of a subject in a preset monitoring mode, and performing a first preprocessing on the obtained EEG data; S200, using a neural network algorithm to train the EEG data after the first preprocessing, establishing a neural network prediction model, and inputting the data to be tested into the neural network prediction model to identify preset mental health problem symptoms; S300. Correct the preset mental health problem symptoms identified by the neural network prediction model using the actual mental health problem symptoms fed back by the subjects, and input the corrected preset mental health problem symptom data into the neural network prediction model for retraining to complete the optimization of the neural network prediction model.
2. The method for detecting active psychological problem symptoms based on EEG data detection according to claim 1, characterized in that: The step of obtaining the EEG data of the subject in a preset monitoring mode includes: Collect EEG data of subjects with their eyes open and closed in a preset monitoring mode; The acquisition time of the EEG data in the two states of eyes being open and eyes being closed is set as the first preset time.
3. The method for detecting active psychological problem symptoms based on EEG data detection according to claim 2, characterized in that: The preset monitoring modes include 4-channel monitoring and 19-channel monitoring; The brain points monitored by the 4 channels include: Fz, Pz, C3, and C4; The brain points monitored by the 19 channels include: Fp1, Fp2, Fz, F3, F4, F7, F8, Cz, C3, C4, P3, P4, Pz, T3, T4, T5, T6, O1, and O2.
4. The method for proactively detecting symptoms of psychological problems based on EEG data detection according to claim 1, characterized in that: The first preprocessing of the acquired EEG data includes: Point elimination processing, eliminating outliers or noise through a first preset algorithm; Eliminating trend items, removing systematic trends or changes through a second preset algorithm; Further eliminating artifacts by using a third preset algorithm, wherein the sources of the artifacts include electrooculography, electromyography, and electrocardiography; The EEG data is frequency-divided according to the first frequency to obtain EEG data of each frequency band.
5. The method for proactively detecting symptoms of psychological problems based on EEG data detection according to claim 4, characterized in that: The first preset algorithm includes the Dixon criterion; The second preset algorithm includes a least squares fitting method and a wavelet decomposition method; The third preset algorithm includes a filter method and a deep learning method; The first frequencies include Delta, Theta, Alpha, Low Beta, Beta, High Beta, Gamma, Alpha1, and Alpha2.
6. The method for proactively detecting symptoms of psychological problems based on EEG data detection according to claim 4, characterized in that: The first preprocessing of the acquired EEG data further includes: Quantify the absolute value of the EEG data of each preset brain area and the relative value between each preset brain area, and calculate the difference with the standard value of the same gender and age in the preset brain area; The EEG data of each preset brain area is intercepted according to the preset recording time.
7. The method for proactively detecting symptoms of psychological problems based on EEG data detection according to claim 6, characterized in that: The method of using a neural network algorithm to train the EEG data after the first preprocessing to establish a neural network prediction model includes: Classify and group the EEG data of each preset brain region and frequency band according to their relevance to the preset mental health problem symptoms; Performing logical relationship processing on the EEG data through a first association analysis and a first weight analysis, respectively calculating a first weighted sum of absolute values and a first weighted sum of relative values of the EEG data of each preset brain region and each frequency band in the eyes-open state and the eyes-closed state; Determine whether the first weighted sum of the absolute values and the first weighted sum of the relative values of the EEG data of each preset brain area and each frequency band in the eyes open and eyes closed states are greater than or equal to a first preset threshold, and determine whether the subject has preset mental health problem symptoms based on the judgment result.
8. The method for proactively detecting symptoms of psychological problems based on EEG data detection according to claim 7, characterized in that: The determining whether the first weighted sum of the absolute values and the first weighted sum of the relative values of the electroencephalogram data of each preset brain region and each frequency band in the eyes-open and eyes-closed states is greater than or equal to a first preset threshold, and determining whether the subject has a preset mental health problem symptom according to the determination result, includes: If any first weighted sum of the absolute value and the relative value of the EEG data of each preset brain region and each frequency band in the eyes-open and eyes-closed states is greater than or equal to a first preset threshold, it is determined that the subject has a preset mental health problem symptom; If the first weighted sum of the absolute value and relative value of the EEG data of each preset brain area and each frequency band in the eyes open and eyes closed states are both less than the first preset threshold, it is determined that the subject does not have the preset mental health problem symptoms.
9. The method for proactively detecting symptoms of psychological problems based on EEG data detection according to claim 7, characterized in that: The S300 includes: Compare the pre-set mental health problem symptoms identified by the neural network prediction model with the actual mental health problem symptoms reported by the subjects; If the comparison results are inconsistent, the actual mental health problem symptom data reported by the subjects will be input into the neural network prediction model for the first association analysis and the first weight analysis to optimize the neural network prediction model.
10. A proactive psychological problem symptom detection system based on EEG data detection, characterized in that: include: Acquisition module: used for acquiring EEG data of the subject in a preset monitoring mode, and performing a first preprocessing on the acquired EEG data; A control module is used to perform a first preprocessing on the acquired EEG data, or to use a neural network algorithm to train the EEG data after the first preprocessing, establish a neural network prediction model, and input the test data into the neural network prediction model to identify preset mental health problem symptoms; or to use the actual mental health problem symptoms fed back by the subjects to correct the preset mental health problem symptoms identified by the neural network prediction model, and input the corrected preset mental health problem symptom data into the neural network prediction model for retraining to complete the optimization of the neural network prediction model.
