Intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia

Abstract

The invention discloses an intelligent electroacupuncture frequency matching system for insomnia auxiliary intervention, and the system comprises a patient emotion scale obtaining module which is used for obtaining the negative emotion scale data of an insomnia patient; the patient resting state physiological signal acquisition module is used for acquiring resting state physiological signals of an insomnia patient and extracting resting state physiological signal features; the patient behavior data acquisition module is used for acquiring behavior data of an insomnia patient in an active state; the insomnia subtype classification module is used for inputting the data acquired by each acquisition module as classification basic data into a pre-trained insomnia subtype classification model so as to divide the insomnia patients into different insomnia subtypes; and the electro-acupuncture frequency matching module is used for adopting a pre-stored electro-acupuncture frequency matching model to obtain the electro-acupuncture frequency suitable for the insomnia patient according to the insomnia subtype to which the insomnia patient belongs. The subtype classification is more comprehensive and accurate, and the electroacupuncture frequency matching result is more accurate.

Description

Technical Field

[0001] The present invention relates to an electroacupuncture frequency matching technology, and in particular to an intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia. Background Art

[0002] Insomnia is one of the common sleep disorders in modern society, which seriously affects the quality of life and social function of patients. It is manifested as frequent and persistent difficulty falling asleep and / or difficulty maintaining sleep, which leads to unsatisfactory sleep. Insomnia patients usually experience problems such as difficulty falling asleep, easy waking up, and early waking up, which leads to a significant decrease in sleep quality. Insomnia not only affects nighttime rest, but also manifests symptoms such as fatigue, inattention, memory loss, and emotional instability during the day. As a common manifestation of emotional disorders, insomnia not only interferes with patients' sleep, but also causes a series of physiological and psychological problems. In the long run, insomnia can trigger or aggravate other emotional disorders such as depression and anxiety, and even affect patients' work and social functions. At present, there are many ways to treat insomnia, including drug therapy, cognitive behavioral therapy, sleep hygiene education, and lifestyle adjustments. However, the effects of these methods vary from person to person, and drug therapy is often accompanied by side effects. The compliance of cognitive behavioral therapy and lifestyle adjustments is also poor, making it difficult to achieve long-term and stable therapeutic effects. Therefore, a more effective and personalized auxiliary intervention program is urgently needed to deal with insomnia, a complex emotional disorder.

[0003] As a traditional Chinese medicine therapy, electroacupuncture intervention can regulate the patient's nervous system and endocrine system and effectively relieve insomnia symptoms by applying electrical stimulation to specific acupoints. In recent years, electroacupuncture intervention has received increasing attention and recognition due to its safety, simplicity and fewer side effects. Electroacupuncture-assisted intervention can not only improve the patient's sleep quality, but also relieve emotional disorders such as anxiety and depression. However, in practical applications, the effect of electroacupuncture-assisted intervention is affected by many factors, including electroacupuncture frequency, stimulation intensity, acupoint selection and individual differences of patients. Traditional electroacupuncture-assisted intervention methods mainly rely on the doctor's experience and lack systematic and standardized quantitative methods, making it difficult to ensure that every patient can obtain the best auxiliary intervention effect. To this end, combined with modern scientific and technological means, by establishing a scientific electroacupuncture frequency matching model, it is possible to more accurately determine the individualized electroacupuncture-assisted intervention plan, thereby improving the auxiliary intervention effect and patient satisfaction. Summary of the invention

[0004] In view of the problems existing in the prior art, the purpose of the present invention is to provide an intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia with higher electroacupuncture frequency matching accuracy.

[0005] In order to achieve the above-mentioned object of the invention, the present invention provides the following technical solutions:

[0006] An intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia, comprising:

[0007] Patient emotion scale acquisition module, used to obtain negative emotion scale data of insomnia patients;

[0008] The patient resting physiological signal acquisition module is used to collect the resting physiological signals of insomnia patients and extract the features of the resting physiological signals;

[0009] A patient behavior data acquisition module, used to acquire the behavior data of insomnia patients in their active state;

[0010] an insomnia subtype classification module, which is used to use the data acquired by the patient emotion scale acquisition module, the patient resting physiological signal acquisition module and the patient behavior data acquisition module as classification basic data, and input the data into a pre-trained insomnia subtype classification model, thereby classifying insomnia patients into different insomnia subtypes, wherein the insomnia subtype classification model is a machine learning model;

[0011] The electroacupuncture frequency matching module is used to obtain the electroacupuncture frequency suitable for the insomnia patient according to the insomnia subtype to which the insomnia patient belongs, using a pre-stored electroacupuncture frequency matching model, wherein the electroacupuncture frequency matching model is used to indicate the electroacupuncture frequency suitable for different insomnia subtypes.

[0012] Furthermore, the patient resting physiological signal acquisition module specifically includes:

[0013] An EEG data acquisition unit is used to respectively acquire EEG data of insomnia patients in the awake and sleeping states;

[0014] An EEG feature extraction unit is used to calculate the power ratio of the EEG low-frequency power and the EEG high-frequency power in the EEG data, as well as the microstate characteristics of the EEG data, and use the power ratio and the microstate characteristics as brain excitability characteristics;

[0015] An ECG data acquisition unit is used to collect ECG data of insomnia patients in the awake and sleeping states;

[0016] The ECG feature extraction unit is used to analyze the ECG data and obtain the heart rate variability characteristics of different periods;

[0017] The resting state physiological signal feature acquisition unit is used to use brain excitability features and heart rate variability features as resting state physiological signal features.

[0018] Furthermore, the patient behavior data acquisition module is specifically a portable wearable device, which is used to obtain the daytime and nighttime activities of the insomnia patient as behavior data.

[0019] Furthermore, the insomnia subtype classification model specifically includes:

[0020] The data preprocessing unit is used to clean, fill missing values, standardize and normalize the classified basic data, and use the processed data as multimodal health status data D = {X t |t=1,…,T}, where X t represents the multimodal health status data at time t, where T represents the number of time steps;

[0021] The sparse dynamic matrix decomposition unit is used to obtain the health trend matrix by matrix decomposition based on the multimodal health status data:

[0022] X t =H t ·P t +B t +ε

[0023] In the formula, H t represents the health trend matrix at time t, P t represents the health-time projection matrix at time t, B t represents the personalized deviation matrix at time t, ε represents noise, H t , P t Obtained by decomposition;

[0024] Dynamic weight update unit, used to calculate personalized dynamic weights based on the health trend matrix:

[0025] W t =α·W t-1 +(1-α)·ΔH t

[0026] Where W t , W t-1 Represents the personalized dynamic weight at time t and t-1, initial value α represents the weight smoothing factor, ΔH t Indicates H t The gradient of H t Dimensions;

[0027] The health prediction unit is used to modify the health trend matrix using personalized dynamic weights, input the modified health trend matrix into the trained prediction neural network, and predict the health status at the future time t+τ

[0028] The personalized deviation updating unit is used to update the personalized deviation matrix according to the user's health feedback:

[0029]

[0030] In the formula, B t+1represents the personalized bias matrix at time t+1, γ represents the learning rate, represents the predicted real-time health, f() represents the prediction function, F t 、F t-1 Respectively represent the user's health feedback at time t and t-1;

[0031] Classification unit, used to classify the health status at the future time t+τ Clustering is performed to obtain insomnia subtypes of insomnia patients.

[0032] Furthermore, the sparse dynamic matrix decomposition unit is used to obtain a health trend matrix by matrix decomposition based on the multimodal health status data using the following formula as the optimization target:

[0033]

[0034] In the formula, represents the Frobenius norm, ‖ ‖ 1 Indicates l 1 -norm, λ 1 , 2 , 3 represents the regularization parameter.

[0035] Furthermore, the health trend matrix is ​​modified specifically according to the following formula:

[0036]

[0037] In the formula, represents the corrected health trend matrix, and ⊙ represents element-by-element multiplication.

[0038] Furthermore, the loss function of the prediction neural network is:

[0039]

[0040] In the formula, X t+τ represents the actual multimodal health status data at time t+τ, λ represents the regularization parameter, ‖‖ 1 Indicates l 1 -norm.

[0041] Furthermore, the electroacupuncture frequency matching model is specifically constructed by the following steps:

[0042] Obtain the intervention feedback results of several insomnia patients with different insomnia subtypes after using different electroacupuncture frequencies;

[0043] The machine learning model is trained based on the feedback results and the electroacupuncture frequency, and the trained machine learning model is used as the electroacupuncture frequency matching model.

[0044] Furthermore, the system also includes an optimization module for regularly evaluating and monitoring the auxiliary intervention effect of the patient, adjusting and optimizing the plan according to the evaluation results, and adjusting the electroacupuncture frequency in time.

[0045] Furthermore, the system also includes an auxiliary program generation module for generating an auxiliary intervention program for insomnia according to the frequency of electroacupuncture.

[0046] Compared with the prior art, the present invention has the following beneficial effects: 1. The present invention combines the emotional scale data and the resting physiological signal characteristics, not only relying on the patient's subjective description, but also combining objective physiological data, so that the classification of insomnia subtypes is more comprehensive and accurate; 2. The present invention can take into account the patient's emotional needs and physiological state, and realize effective electroacupuncture frequency matching auxiliary intervention plan, and the matching result is more accurate; 3. The present invention can timely adjust the auxiliary intervention plan during the auxiliary intervention process through real-time monitoring and adjustment to ensure the best auxiliary intervention effect; 4. The present invention can enhance the patient's trust and confidence in auxiliary intervention by providing scientific and objective emotional state and physiological state assessment, thereby improving the compliance and effect of auxiliary intervention; 5. The method of the present invention has wide applicability, not only suitable for insomnia, but also can be promoted to auxiliary intervention for other emotional disorders. BRIEF DESCRIPTION OF THE DRAWINGS

[0047] Figure 1 A structural diagram of an intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia provided by an embodiment of the present invention;

[0048] Figure 2 Flowchart for assessing the emotional state of patients with insomnia;

[0049] Figure 3 To use spectrum analysis and microstate methods to extract the dynamic changes of brain networks in insomnia patients that reflect changes in brain excitability and consciousness activities;

[0050] Figure 4 Flowchart for constructing the electroacupuncture frequency matching model. DETAILED DESCRIPTION

[0051] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the accompanying drawings in the embodiments of the present invention.

[0052] Embodiment 1

[0053] The embodiment of the present invention provides an intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia, such as Figure 1 As shown, including:

[0054] Patient emotion scale acquisition module, used to obtain negative emotion scale data of insomnia patients;

[0055] The patient resting physiological signal acquisition module is used to collect the resting physiological signals of insomnia patients and extract the features of the resting physiological signals;

[0056] A patient behavior data acquisition module, used to acquire the behavior data of insomnia patients in their active state;

[0057] an insomnia subtype classification module, which is used to use the data acquired by the patient emotion scale acquisition module, the patient resting physiological signal acquisition module and the patient behavior data acquisition module as classification basic data, and input the data into a pre-trained insomnia subtype classification model, thereby classifying insomnia patients into different insomnia subtypes, wherein the insomnia subtype classification model is a machine learning model;

[0058] An electroacupuncture frequency matching module, for obtaining an electroacupuncture frequency suitable for the insomnia patient according to the insomnia subtype to which the insomnia patient belongs, using a pre-stored electroacupuncture frequency matching model, wherein the electroacupuncture frequency matching model is used to indicate the electroacupuncture frequency suitable for different insomnia subtypes;

[0059] Among them, Figure 2 As shown in the figure, the negative emotion scale data includes the anxiety-depression-stress comprehensive scale data of insomnia patients and other negative emotional state scale data. Through the objective and standardized tests of the above scales, the different types of emotional disorders and emotional fluctuations of insomnia patients can be accurately identified, and their emotional states can be comprehensively evaluated in multiple dimensions to reflect the emotional reactions and mental health conditions of patients in different situations. The emotional scale data is standardized and key features are extracted. Feature extraction can use methods such as principal component analysis to reduce data dimensions and improve data interpretability.

[0060] First, depression scales (such as the Beck Depression Inventory, BDI) can be used to assess the severity of depressive symptoms. It contains a series of questions that ask patients about their emotions, cognitions, and behaviors over the past period of time. Patients need to score each question based on their actual situation. These scores can reflect the patient's typical depressive symptoms such as low mood, loss of interest, fatigue, difficulty concentrating, and decreased self-worth. By analyzing the data from the depression scale, it can help identify whether the patient has depression and the severity of the depression.

[0061] Secondly, anxiety scales (such as the Generalized Anxiety Scale, GAD-7) can be used to assess patients' anxiety symptoms. The anxiety scale includes a series of questions that ask patients about their anxiety feelings and related physical symptoms in the past two weeks, such as tension, uncontrollable worry, irritability, muscle tension, sleep disorders, etc. Patients also need to score each question to reflect the frequency and severity of their anxiety symptoms. Through data analysis of the anxiety scale, it is possible to identify whether the patient has anxiety symptoms and the extent to which these symptoms affect their daily life, thereby providing a basis for subsequent auxiliary intervention plans.

[0062] Stress scales (such as the Perceived Stress Scale, PSS) are used to assess the level of stress felt by patients. The scale uses a series of questions to ask patients about their feelings of stress and how they cope with it in the past month. The questions involve various stressors in life, such as work, family, interpersonal relationships, etc., as well as the patient's subjective feelings and reactions to these stressors. Patients are required to score each question to reflect their perceived level of stress. The data from the stress scale can reveal the patient's psychological state when dealing with daily stress, providing an important reference for understanding their overall emotional state.

[0063] Through these emotional scales filled in by patients, the information of patients in different emotional dimensions can be fully captured. The data of these scales can not only provide the subjective emotional experience reported by patients themselves, but also quantify and standardize these emotional experiences, which is convenient for subsequent analysis and processing. Combined with the data of these emotional scales, the present invention can more accurately assess the emotional state of patients, identify their possible emotional disorders, and provide a scientific basis for formulating personalized auxiliary intervention plans.

[0064] The patient resting physiological signal acquisition module specifically includes:

[0065] An EEG data acquisition unit is used to respectively acquire EEG data of insomnia patients in the awake and sleeping states;

[0066] An EEG feature extraction unit is used to calculate the power ratio of the EEG low-frequency power and the EEG high-frequency power in the EEG data, as well as the microstate characteristics of the EEG data, and use the power ratio and the microstate characteristics as brain excitability characteristics;

[0067] An ECG data acquisition unit is used to collect ECG data of insomnia patients in the awake and sleeping states;

[0068] The ECG feature extraction unit is used to analyze the ECG data and obtain the heart rate variability characteristics of different periods;

[0069] The resting state physiological signal feature acquisition unit is used to use brain excitability features and heart rate variability features as resting state physiological signal features.

[0070] The EEG data acquisition unit is specifically a wearable physiological signal monitoring device that collects EEG data from insomnia patients and reveals the disease heterogeneity of insomnia patients through EEG data spectrum analysis. The different regulatory mechanisms of patients are identified through the resting-state physiological signal features extracted from the resting-state physiological signals. The frequency and duration of data acquisition can be adjusted according to specific needs to ensure the representativeness and accuracy of the data. For example, the acquisition frequency can be set to hundreds of times per second to capture subtle physiological changes; the acquisition time can be set to a few minutes to a few hours to obtain a sufficient sample size.

[0071] The collected physiological signal data usually contains certain noise and fluctuations. In order to improve the quality of the data and the accuracy of subsequent analysis, the data needs to be preprocessed. Preprocessing mainly involves denoising, smoothing and normalizing the collected physiological signal data. First of all, denoising is one of the important steps in data preprocessing. A variety of filters and independent component analysis algorithms can be used to eliminate noise in the data. Commonly used filters include low-pass filters, high-pass filters and band-pass filters, which can effectively remove noise in different frequency bands. In addition, advanced algorithms such as adaptive filters and wavelet transforms can be used to further improve the denoising effect. Next, smoothing can reduce the volatility of the data, making the signal more stable and easier to analyze. Moving average is a commonly used smoothing method. By averaging the data with a sliding window, short-term fluctuations can be effectively reduced. In addition, methods such as exponential smoothing and median filtering can be used to select the most suitable smoothing method according to the specific characteristics of the physiological signal. After data denoising and smoothing, normalization is also required. Normalization converts data to a uniform scale through standardization or regularization methods, so that data of different types and magnitudes can be compared and analyzed on the same scale. Common normalization methods include Z-score standardization and Min-Max standardization. Z-score standardization converts data into normally distributed data with a mean of zero and a standard deviation of one; Min-Max standardization scales the data to the interval [0,1]. Normalization can not only improve the comparability of data, but also reduce the deviation of data distribution, providing high-quality input data for subsequent machine learning algorithms.

[0072] Taking EEG signals as an example, the EEG data acquisition unit collects resting EEG signals of insomnia patients during daytime wakefulness and nighttime sleep, and divides the data of daytime wakefulness, wakefulness before falling asleep, non-rapid eye movement sleep, rapid eye movement sleep, and wakefulness after sleep. The EEG signals are converted from the time domain to the frequency domain using Fourier transform, and the power of different frequencies of resting EEG of insomnia patients in different scenarios is extracted. The low-frequency and high-frequency power of EEG corresponding to delta and beta / gamma in the selected frequency band are calculated, and the power ratio of low-frequency and high-frequency power of EEG is calculated as the ratio of high-frequency and low-frequency activity intensity. This ratio can be used as a reliable indicator to characterize the excitability (degree of wakefulness) of brain activity in a complete day and night cycle.

[0073] Further, the resting-state EEG signals of insomnia patients collected during daytime wakefulness and nighttime sleep as described above are used to extract the microstate characteristics of resting-state EEG of insomnia patients in different scenarios. By analyzing the microstate characteristics of the low-frequency and high-frequency bands of interest, the changes in consciousness activities in different frequency bands of interest are obtained, and the differences between the brain network dynamics of insomnia patients are analyzed in combination with the reflected brain activity excitability (degree of arousal), completing the heterogeneity analysis of insomnia patients, such as Figure 3 shown.

[0074] The ECG data acquisition unit was used to collect the patient's resting ECG signals when awake during the day and asleep at night, and to analyze the circadian rhythm changes of the patient's heart rate variability within 24 hours.

[0075] Different physiological signals can mainly reflect the performance of different systems of the human body. The collected EEG can represent changes in brain activity, the collected ECG signal can characterize the function of the heart and autonomic nervous system, the collected respiratory signal can characterize the functional characteristics of the respiratory system, etc. By analyzing the coupling, consistency or time lag between different physiological signals, the interaction relationship between various systems of the body can be characterized, and the change process of the dominant position or direction of action of the system during the day and night process can be judged, which can also help determine the damage to the physiological function of insomnia patients.

[0076] The patient behavior data acquisition module is specifically a portable wearable device, which is used to obtain the daytime and nighttime activities of insomnia patients, including indicator data such as gait and activity intensity, as behavior data. For example, through information such as daytime and nighttime activity types, duration, and intensity, their regular life patterns can be determined, their bad living habits and sleep habits to be improved can be discovered, and their sleep subtype classification can be assisted.

[0077] The insomnia subtype classification model specifically includes:

[0078] The data preprocessing unit is used to clean the classification basic data, fill in missing values ​​(for example, based on Gaussian process regression), standardize and normalize, and align the timestamps to generate multimodal health status data D = {X t |t=1,…,T}, where X t Represents the multimodal health status data at time t, namely, the negative emotion scale data, resting physiological signal characteristics (power ratio, microstate characteristics, heart rate variability characteristics) and behavioral data at time t, and T represents the time step;

[0079] The sparse dynamic matrix decomposition unit is used to obtain the health trend matrix by matrix decomposition based on the multimodal health status data:

[0080] X t =H t ·P t +B t +ε

[0081] In the formula, H t Represents the health trend matrix at time t, specifically X t The low-dimensional sparse basis captures the core health features of different modalities and is used to represent the intrinsic association and abstract representation between different modalities. t represents the health-time projection matrix at time t, which is used to describe the dynamic change pattern of health status over time. t represents the personalized deviation matrix at time t, which is used to compensate for the individual's specific health characteristics or baseline deviations (for example, the individual's normal heart rate level, EEG background activity, etc.), ε represents noise, which is generally assumed to have a known distribution (such as Gaussian distribution) and is used to model unexplained random errors, and H t , P t Obtained by decomposition;

[0082] Dynamic weight update unit, used to calculate personalized dynamic weights based on the health trend matrix:

[0083] W t =α·W t-1 +(1-α)·ΔH t

[0084] Where W t , W t-1 Represents the personalized dynamic weight at time t and t-1, initial value α represents the weight smoothing factor, which is used to control the balance between short-term and long-term trends. ΔH t Indicates H t The gradient of H is used to capture the trend mutation. t Dimension of W tThe weight of each modality is set in , which can adjust the degree of attention paid by the prediction model to each modality data.

[0085] The health prediction unit is used to modify the health trend matrix using personalized dynamic weights, input the modified health trend matrix into the trained prediction neural network (RNN-LSTM), and predict the health status at the future time t+τ

[0086]

[0087] In the formula, represents the corrected health trend matrix, ⊙ represents element-by-element multiplication; through the dynamic weight W t Adjust H t The importance of each modal feature in makes RNN-LSTM pay more attention to the modalities with significant changes in health trends; wherein the loss function of the prediction neural network is:

[0088]

[0089] In the formula, X t+τ represents the actual multimodal health status data at time t+τ, λ represents the regularization parameter, ‖‖ 1 Indicates l 1 -norm;

[0090] The personalized deviation updating unit is used to update the personalized deviation matrix according to the user's health feedback:

[0091]

[0092] In the formula, B t+1 represents the personalized bias matrix at time t+1, γ represents the learning rate, represents the predicted real-time health, f() represents the prediction function, such as LSTM neural network, F t 、F t-1 Respectively represent the user's health feedback (doctor's feedback and / or patient's own feedback) at time t and t-1;

[0093] Classification unit, used to classify the health status at the future time t+τ Clustering (such as K-means or DBSCAN) is performed to obtain insomnia subtypes of insomnia patients.

[0094] The sparse dynamic matrix decomposition unit is specifically used to obtain a health trend matrix by matrix decomposition based on multimodal health status data with the following formula as the optimization target:

[0095]

[0096] In the formula, represents the Frobenius norm, ‖ ‖ 1 Indicates l 1 -norm, λ 1 , 2 , 3 represents the regularization parameter.

[0097] The first term: data reconstruction error, measuring X t The difference with its reconstructed value uses the Frobenius norm (i.e. the sum of squares of the elements).

[0098] The second item: H t Apply l 1 -norm regularization to enhance sparsity.

[0099] Item 3: To B t Apply l 1 -norm regularization promotes the sparsity of the bias matrix.

[0100] The fourth item: time smoothing constraint.

[0101] The electroacupuncture frequency matching model is specifically constructed by the following steps: obtaining the intervention feedback results of several different insomnia patients after using different electroacupuncture frequencies; including negative emotion scale data (emotional state), resting physiological signal characteristics (functional characteristics) and behavioral data (behavioral patterns), intervention feedback results under different electroacupuncture frequencies, etc. These data are processed in a standardized and systematic manner to form a large and detailed database, which provides a basis for the subsequent electroacupuncture frequency matching model; negative emotion scale data, resting physiological signal characteristics and behavioral data can obtain insomnia subtypes of insomnia patients, and the intervention feedback results of different insomnia patients after using different electroacupuncture frequencies are used as training data; Figure 4 As shown in the figure, according to the feedback results and the frequency of electroacupuncture, that is, the intervention feedback results of different insomnia subtypes after using different electroacupuncture frequencies, the machine learning model is trained and the trained machine learning model is used as the electroacupuncture frequency matching model. Commonly used machine learning algorithms include support vector machines, random forests, and neural networks.

[0102] In other embodiments, the system may also include an auxiliary program generation module for generating an auxiliary intervention program for insomnia according to the frequency of electroacupuncture. The auxiliary intervention program not only includes the specific frequency of electroacupuncture, but also involves the time, frequency and acupoint selection of electroacupuncture auxiliary intervention. Through these personalized auxiliary intervention programs, it is ensured that each patient can get the most appropriate auxiliary intervention to effectively improve their sleep quality and overall health.

[0103] In other embodiments, the system may also include an optimization module for regularly evaluating and monitoring the auxiliary intervention effect of the patient, adjusting and optimizing the scheme according to the evaluation results, and adjusting the frequency of electroacupuncture in a timely manner. In this way, not only the effectiveness of auxiliary intervention can be improved, but also the adverse reactions and side effects that may occur during the auxiliary intervention process can be reduced. Thus, accurate and personalized electroacupuncture-assisted intervention schemes can be provided for patients with different subtypes of insomnia.

[0104] The device provided in the embodiment of the present invention can be used to execute the method provided in the first embodiment of the present invention, and has the corresponding functions and beneficial effects of executing the method.

[0105] It is worth noting that in the embodiment of the above-mentioned determination device, the various units and modules included are only divided according to functional logic, but are not limited to the above-mentioned division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of the present invention.

[0106] The embodiments described above are only illustrative, wherein the modules described as separate components may or may not be physically separated, and the components displayed as modules may or may not be physical modules, i.e., they may be located in one place, or they may be distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, or of course, it can be implemented only by hardware, as long as the function or effect can be achieved.

[0107] It should be understood that the above embodiments and descriptions only describe the principles, main features and advantages of the present invention. Without departing from the spirit and scope of the present invention, the present invention may be subject to various changes and improvements, and these changes and improvements all fall within the scope of protection of the present invention.

Claims

1. An intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia, characterized in that: include: Patient emotion scale acquisition module, used to obtain negative emotion scale data of insomnia patients; The patient resting physiological signal acquisition module is used to collect the resting physiological signals of insomnia patients and extract the features of the resting physiological signals; A patient behavior data acquisition module, used to acquire the behavior data of insomnia patients in their active state; an insomnia subtype classification module, which is used to use the data acquired by the patient emotion scale acquisition module, the patient resting physiological signal acquisition module and the patient behavior data acquisition module as classification basic data, and input the data into a pre-trained insomnia subtype classification model, thereby classifying insomnia patients into different insomnia subtypes, wherein the insomnia subtype classification model is a machine learning model; The electroacupuncture frequency matching module is used to obtain the electroacupuncture frequency suitable for the insomnia patient according to the insomnia subtype to which the insomnia patient belongs, using a pre-stored electroacupuncture frequency matching model, wherein the electroacupuncture frequency matching model is used to indicate the electroacupuncture frequency suitable for different insomnia subtypes.

2. The intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia according to claim 1, characterized in that: The patient resting physiological signal acquisition module specifically includes: An EEG data acquisition unit is used to respectively acquire EEG data of insomnia patients in the awake and sleeping states; An EEG feature extraction unit is used to calculate the power ratio of the EEG low-frequency power and the EEG high-frequency power in the EEG data, as well as the microstate characteristics of the EEG data, and use the power ratio and the microstate characteristics as brain excitability characteristics; An ECG data acquisition unit is used to collect ECG data of insomnia patients in the awake and sleeping states; The ECG feature extraction unit is used to analyze the ECG data and obtain the heart rate variability characteristics of different periods; The resting state physiological signal feature acquisition unit is used to use brain excitability features and heart rate variability features as resting state physiological signal features.

3. The intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia according to claim 1, characterized in that: The patient behavior data acquisition module is specifically a portable wearable device, which is used to obtain the daytime and nighttime activities of insomnia patients as behavior data.

4. The intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia according to claim 1, characterized in that: The insomnia subtype classification model specifically includes: The data preprocessing unit is used to clean, fill missing values, standardize and normalize the classified basic data, and use the processed data as multimodal health status data D = {X t |t=1,…,T}, where X t represents the multimodal health status data at time t, where T represents the number of time steps; The sparse dynamic matrix decomposition unit is used to obtain the health trend matrix by matrix decomposition based on the multimodal health status data: X t =H t ·P t +B t +ε In the formula, H t represents the health trend matrix at time t, P t represents the health-time projection matrix at time t, B t represents the personalized deviation matrix at time t, ε represents noise, H t , P t Obtained by decomposition; Dynamic weight update unit, used to calculate personalized dynamic weights based on the health trend matrix: W t =α·W t-1 +(1-a)·ΔH t Where W t , W t-1 Represents the personalized dynamic weight at time t and t-1, initial value α represents the weight smoothing factor, ΔH t Indicates H t The gradient of H t Dimensions; The health prediction unit is used to modify the health trend matrix using personalized dynamic weights, input the modified health trend matrix into the trained prediction neural network, and predict the health status at the future time t+τ The personalized deviation updating unit is used to update the personalized deviation matrix according to the user's health feedback: In the formula, B t+1 represents the personalized bias matrix at time t+1, γ represents the learning rate, represents the predicted real-time health, f() represents the prediction function, F t 、F t-1 Respectively represent the user's health feedback at time t and t-1; Classification unit, used to classify the health status at the future time t+τ Clustering is performed to obtain insomnia subtypes of insomnia patients.

5. The intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia according to claim 4, characterized in that: The sparse dynamic matrix decomposition unit is used to obtain a health trend matrix by matrix decomposition based on multimodal health status data with the following formula as the optimization target: In the formula, represents the Frobenius norm, ‖‖1 represents the l1-norm, and λ1, λ2, and λ3 represent regularization parameters.

6. The intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia according to claim 4, characterized in that: The health trend matrix is ​​specifically modified according to the following formula: In the formula, represents the corrected health trend matrix, and ⊙ represents element-by-element multiplication.

7. The intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia according to claim 4, characterized in that: The loss function of the prediction neural network is: Where, X t+τ represents the actual multimodal health status data at time t+τ, λ represents the regularization parameter, and ‖‖1 represents the l1-norm.

8. The intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia according to claim 1, characterized in that: The electroacupuncture frequency matching model is specifically constructed by the following steps: Obtain the intervention feedback results of several insomnia patients with different insomnia subtypes after using different electroacupuncture frequencies; The machine learning model is trained based on the feedback results and the electroacupuncture frequency, and the trained machine learning model is used as the electroacupuncture frequency matching model.

9. The intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia according to claim 1, characterized in that: The system further comprises: The optimization module is used to regularly evaluate and monitor the effect of auxiliary intervention on patients and adjust the frequency of electroacupuncture according to the evaluation results.

10. The intelligent electroacupuncture frequency matching system for auxiliary intervention of insomnia according to claim 1, characterized in that: The system further comprises: The auxiliary program generation module is used to generate an auxiliary intervention program for insomnia based on the frequency of electroacupuncture.

Images