Control method and system of electronic acupuncture and moxibustion equipment

Abstract

The invention provides a control method and system for electronic acupuncture equipment, and the method comprises the steps: obtaining the symptoms of a user, and recommending a matched acupoint for the user based on a symptom-acupoint database; the electronic acupuncture device scans a target area where the matched acupoint is located, skin resistance distribution and an infrared thermal image of the target area are obtained, the acupoint matching degree is calculated through a probability model, and the acupoint is positioned based on the acupoint matching degree; and monitoring whether the scanning area of the electronic acupuncture device is aligned with the recommended acupuncture point or not in real time, outputting current stimulation by the electronic acupuncture device when the scanning area is monitored to be stopped, and adjusting the output current according to the symptom of the user after the user is stimulated. By means of the method and the corresponding system, the user can be assisted in accurately positioning the acupuncture points, the stimulation parameters can be dynamically adjusted according to feedback information obtained after acupuncture and moxibustion are conducted on the user, and the optimal curative effect is kept.

Description

Technical Field

[0001] The invention provides a control method and system for electronic acupuncture equipment, and relates to the technical field of medical devices. Background Art

[0002] Electronic acupuncture is an advanced treatment method that combines traditional acupuncture therapy with modern electronic technology. Traditional acupuncture has a history of thousands of years. It regulates the body's energy flow, promotes health and relieves pain by pricking specific acupuncture points on the human body. Electronic acupuncture uses electric current stimulation to replace or assist acupuncture, and implements transcutaneous electrical stimulation on specific acupuncture points. However, existing electronic acupuncture equipment cannot assist users in accurately locating acupuncture points. Users rely on experience and feeling to locate acupuncture points, or cannot adjust the output current of electronic acupuncture according to the user's own characteristics and the effect of initial acupuncture, which greatly reduces the therapeutic effect of electronic acupuncture and easily causes abnormal skin damage to users. Summary of the invention

[0003] The present invention provides a control method and system for electronic acupuncture equipment to solve the above-mentioned problems:

[0004] The present invention provides a control method for an electronic acupuncture device, the method comprising:

[0005] Obtain user symptoms and recommend matching acupoints for the user based on the symptom-acupoint database;

[0006] The electronic acupuncture device scans the target area where the matched acupoints are located, obtains the skin resistance distribution and infrared thermal image of the target area, calculates the acupoint matching degree through a probability model, and locates the acupoints based on the acupoint matching degree.

[0007] Real-time monitoring is performed to determine whether the scanning area of ​​the electronic acupuncture device is aligned with the recommended acupuncture points and whether it is stationary. When it is detected that it is stationary, the electronic acupuncture device outputs current stimulation and adjusts the output current according to the symptoms of the user after the stimulation.

[0008] Furthermore, the user's symptoms are obtained, and matching acupoints are recommended to the user based on the symptom-acupoint database, including:

[0009] Obtain the symptoms described by the user and map the symptoms described by the user to standardized symptoms;

[0010] Acupoints matching the standardized symptoms are searched in the symptom-acupoint database, and a relevance score is generated for each matching acupoint. The relevance score sorts the acupoints according to the severity of the symptoms, and after sorting, the acupoints are initially recommended to the user. The relevance scoring model is:

[0011]

[0012] Among them, Score(ak ) indicates acupoint a k The total relevance score, w i represents the weight of the symptom, m represents the total number of standardized symptoms related to the user's symptom, and f(s i ) indicates symptoms in the symptom-acupoint database i The frequency of occurrence, f(a k ) indicates acupoint a in the symptom-acupoint database k The frequency of occurrence, f(s i ,a k )Symptoms i and acupoints k frequency of co-occurrence;

[0013] The initially recommended acupoints are filtered according to the user's personal information to generate final acupoints recommended for the user.

[0014] Furthermore, the electronic acupuncture device scans the target area where the matched acupoints are located, obtains the skin resistance distribution and infrared thermal image of the target area, calculates the acupoint matching degree through a probability model, and locates the acupoints based on the acupoint matching degree, including:

[0015] After recommending the final acupuncture point to the user, the acupuncture point selected by the user is obtained, and the specific image of the acupuncture point on the human body is automatically displayed;

[0016] The electronic acupuncture instrument scans the target area of ​​the acupuncture point on the human body, obtains the resistance distribution and infrared thermal image of the target area, and calculates the acupuncture point matching degree through the probability model. When the acupuncture point matching degree is greater than or equal to the preset threshold, the electronic acupuncture instrument starts a vibration reminder to remind the user to stop moving. Specifically, the probability model is:

[0017]

[0018] Among them, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being in the current active state, X t represents the multimodal physiological signals observed at time t, including temperature and resistance, a k Indicates acupoint k,q(X t ) indicates that when no acupoints are activated, physiological signal X is observed t The background probability distribution, P(X t |a k ) indicates that when acupoint k is in the activated state, the physiological signal X is observed t The conditional probability distribution of Represents KL divergence, which is used to measure the difference between two probability distributions;

[0019] Use deep learning to learn the distribution of physiological signals P(X) under acupoint activation t |a k ) and the background distribution q(X t ) is expressed as .

[0020] Furthermore, deep learning is used to learn the distribution of physiological signals P(X) under acupoint activation. t |a k ) and the background distribution q(X t ), including:

[0021] Collecting multimodal physiological signal data, including data in an activated state of acupoints and temperature and resistance data in an unactivated state, and preprocessing the data, wherein the preprocessing includes: removing noise and normalizing;

[0022] Construct a deep learning model, namely a variational autoencoder model. The encoder of the autoencoder model converts the input physiological signal X t Encoded as a low-dimensional latent variable z, the output z's mean μ(X t ) and variance σ(X t ), the decoder of the autoencoder model: decodes the latent variable z back to the original physiological signal dimension;

[0023] Use the preprocessed data to train the constructed variational autoencoder model;

[0024] For the background distribution q(X t ), using the temperature and resistance data of the acupoints in the unactivated state for training, for each acupoint k in the activated state, the distribution P(X t |a k ), use the data in the corresponding activation state for training, and train a separate variational autoencoder model for each acupoint;

[0025] Use the trained variational autoencoder model to calculate the KL divergence for the background distribution q(X t ), using the trained model q(X t ) type encoder output μq(X t ) and σq(X t ) is used to calculate the distribution P(X t |a k ), use the corresponding P(X t |a k ) model’s encoder output μP(X t |a k ) and σP(X t |a k ) to calculate.

[0026] Furthermore, it is monitored in real time whether the scanning area of ​​the electronic acupuncture device is aligned with the recommended acupuncture points and stops moving. When it is detected that the electronic acupuncture device stops moving, the electronic acupuncture device outputs current stimulation and adjusts the output current in real time, including:

[0027] Analyze skin resistance and infrared thermal images in real time, and after confirmation of acupoint alignment, detect the static state of the electronic acupuncture device within a preset time;

[0028] When it is determined that the electronic acupuncture device is aligned with the acupoint within the preset time and is in a static state, the output current is dynamically adjusted through the current stimulation model. Specifically, the current stimulation model is:

[0029] I(a k ,t)=I0·[1+γ·sin(2πft)·φ(a k )]

[0030] Among them, I0 represents the basic current intensity, γ represents the proportional coefficient of the amplitude, f is the stimulation frequency, which is determined by the user's symptoms, and φ(a k ) is the acupoint response function;

[0031] Specifically, the acupoint response function is:

[0032]

[0033] Among them, λ represents the sensitivity adjustment coefficient, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being currently active.

[0034] The present invention provides a control system for an electronic acupuncture device, the system comprising:

[0035] The acupoint recommendation module is used to obtain user symptoms and recommend matching acupoints to users based on the symptom-acupoint database;

[0036] The acupuncture point positioning module is used for the electronic acupuncture device to scan the target area where the matched acupuncture points are located, obtain the skin resistance distribution and infrared thermal image of the target area, calculate the acupuncture point matching degree through the probability model, and locate the acupuncture points based on the acupuncture point matching degree.

[0037] The output current module is used to monitor in real time whether the scanning area of ​​the electronic acupuncture device is aligned with the recommended acupuncture points and stops moving. When it is detected that it stops moving, the electronic acupuncture device outputs current stimulation and adjusts the output current according to the symptoms of the user after the stimulation.

[0038] Furthermore, the recommended acupoint module includes:

[0039] A standardized symptom mapping module is used to obtain the symptoms described by the user and map the symptoms described by the user to standardized symptoms;

[0040] The preliminary recommendation module is used to search for acupoints matching the standardized symptoms in the symptom-acupoint database, generate a relevance score for each matching acupoint, and sort the acupoints according to the severity of the symptoms. After sorting, the acupoints are initially recommended to the user. The relevance score model is:

[0041]

[0042] Among them, Score(a k ) indicates acupoint a k The total relevance score, w i represents the weight of the symptom, m represents the total number of standardized symptoms related to the user's symptom, and f(s i ) indicates symptoms in the symptom-acupoint database i The frequency of occurrence, f(a k ) indicates acupoint a in the symptom-acupoint database k The frequency of occurrence, f(s i , a k )Symptoms i and acupoints k frequency of co-occurrence;

[0043] The final recommendation module is used to filter the initially recommended acupoints according to the user's personal information to generate final acupoints recommended for the user.

[0044] Furthermore, the acupuncture point positioning module includes:

[0045] The module for displaying the area where the acupoints are located is used to obtain the acupoints selected by the user after recommending the final acupoints to the user, and automatically display the specific images of the acupoints on the human body;

[0046] The module for calculating the acupuncture point matching degree is used for the electronic acupuncture instrument to scan the target area of ​​the acupuncture point on the human body, obtain the resistance distribution and infrared thermal image of the target area, and calculate the acupuncture point matching degree through the probability model. When the acupuncture point matching degree is greater than or equal to the preset threshold, the electronic acupuncture instrument starts a vibration reminder to remind the user to stop moving. Specifically, the probability model is:

[0047]

[0048] Among them, P(a k |X t ) indicates that when physiological signal X is observed tAfter that, evaluate the acupoints k The probability of being in the current active state, X t represents the multimodal physiological signals observed at time t, including temperature and resistance, a k Indicates acupoint k,q(X t ) indicates that when no acupoints are activated, physiological signal X is observed t The background probability distribution, P(X t |a k ) indicates that when acupoint k is in the activated state, the physiological signal X is observed t The conditional probability distribution of Represents KL divergence, which is used to measure the difference between two probability distributions;

[0049] Deep learning module, used to use deep learning to learn the physiological signal distribution P(X t |a k ) and the background distribution q(X t ) is expressed as .

[0050] Furthermore, the deep learning module includes:

[0051] A data collection module is used to collect multimodal physiological signal data, including data in an activated state of acupoints and temperature and resistance data in an unactivated state, and preprocess the data, wherein the preprocessing includes: removing noise and normalizing;

[0052] Construct a model module for constructing a deep learning model, namely a variational autoencoder model. The encoder of the autoencoder model converts the input physiological signal X t Encoded as a low-dimensional latent variable z, the output z's mean μ(X t ) and variance σ(X t ), the decoder of the autoencoder model: decodes the latent variable z back to the original physiological signal dimension;

[0053] The training model module is used to train the constructed variational autoencoder model using the preprocessed data. t ), using the temperature and resistance data of the acupoints in the unactivated state for training, for each acupoint k in the activated state, the distribution P(X t |a k ), use the data in the corresponding activation state for training, and train a separate variational autoencoder model for each acupoint;

[0054] The KL divergence calculation module is used to calculate the KL divergence using the trained variational autoencoder model for the background distribution q(X t ), using the trained model q(X t ) type encoder output μq(Xt ) and σq(X t ) is used to calculate the distribution P(X t |a k ), use the corresponding P(X t |a k ) model’s encoder output μP(X t |a k ) and σP(X t |a k ) to calculate.

[0055] Furthermore, the output current module includes:

[0056] A real-time analysis module for analyzing skin resistance and infrared thermal images in real time, and detecting the static state of the electronic acupuncture device within a preset time after the acupoint alignment is confirmed;

[0057] The current adjustment module is used to dynamically adjust the output current through the current stimulation model when it is determined that the electronic acupuncture device is aligned with the acupoint within a preset time and is in a static state. Specifically, the current stimulation model is:

[0058] I(a k , t)=I0·[1+γ·sin(2πft)·φ(a k )]

[0059] Among them, I0 represents the basic current intensity, γ represents the proportional coefficient of the amplitude, f is the stimulation frequency, which is determined by the user's symptoms, and φ(a k ) is the acupoint response function;

[0060] Specifically, the acupoint response function is:

[0061]

[0062] Among them, λ represents the sensitivity adjustment coefficient, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being currently active.

[0063] Beneficial effects of the present invention: precise acupoint positioning, by integrating skin resistance and infrared thermal imaging, combined with probability models, can improve the accuracy of acupoint positioning, which reduces human errors and improves treatment effects; by adjusting current stimulation parameters, the system can respond to the physiological states of different users and provide more accurate and personalized treatment plans; user experience is improved, intelligent symptom matching and automated acupoint detection reduce the complexity of user operations, while also improving user trust and convenience of use; it provides a technical upgrade to traditional acupuncture therapy, making it more refined, humane and efficient. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] Figure 1 This is a schematic diagram of a control method for an electronic acupuncture device according to the present invention. DETAILED DESCRIPTION

[0065] In order to more clearly understand the above-mentioned purpose, features and advantages of the present invention, the present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the embodiments of the present application and the features in the embodiments can be combined with each other without conflict.

[0066] In the following description, many specific details are set forth to facilitate a full understanding of the present invention. The embodiments described are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0067] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0068] One embodiment of the present invention provides a control method for an electronic acupuncture device, the method comprising:

[0069] Obtain user symptoms and recommend matching acupoints for the user based on the symptom-acupoint database;

[0070] The electronic acupuncture device scans the target area where the matched acupoints are located, obtains the skin resistance distribution and infrared thermal image of the target area, calculates the acupoint matching degree through a probability model, and locates the acupoints based on the acupoint matching degree.

[0071] Real-time monitoring is performed to determine whether the scanning area of ​​the electronic acupuncture device is aligned with the recommended acupuncture points and whether it is stationary. When it is detected that it is stationary, the electronic acupuncture device outputs current stimulation and adjusts the output current according to the symptoms of the user after the stimulation.

[0072] The working principle and effect of the above technical solution are as follows: user symptom input and matching, the user inputs the symptom description through the device interface, uses natural language processing (NLP) technology to map the symptoms to specific symptoms stored in the symptom-acupoint database, and recommends a set of acupoints that are effective for the symptoms to the user based on the symptom-acupoint database. These acupoints have historical treatment data associated with the input symptoms; acupoint positioning and identification, the electronic acupuncture device scans the target area, obtains the skin resistance distribution and infrared thermal image, and uses a probability model to analyze the scanned data to calculate the acupoint matching degree. The acupoint matching degree represents the recognition accuracy of each acupoint under the current physiological conditions, and is calculated in combination with historical data and biometrics; the alignment of the acupuncture device is continuously monitored to ensure that the recommended acupoints can be accurately aimed. When the device is aligned with the recommended acupoints and stops moving, it is ready for stimulation. Once it is confirmed that the device is aligned and stopped, an appropriate amount of current is output for acupoint stimulation. While applying current stimulation, the intensity and frequency of the current stimulation are dynamically adjusted to further optimize the treatment effect. Precise acupoint positioning, by integrating skin resistance and infrared thermal imaging, combined with probability models, can improve the accuracy of acupoint positioning, which reduces human errors and improves treatment effects; by adjusting the current stimulation parameters, the system can respond to the physiological states of different users and provide more accurate and personalized treatment plans; user experience is improved, intelligent symptom matching and automated acupoint detection reduce the complexity of user operations, while also improving user trust and ease of use; it provides a technical upgrade to traditional acupuncture therapy, making it more refined, humane and efficient.

[0073] In one embodiment of the present invention, user symptoms are obtained, and matching acupoints are recommended to the user based on a symptom-acupoint database, including:

[0074] Obtain the symptoms described by the user and map the symptoms described by the user to standardized symptoms;

[0075] Acupoints matching the standardized symptoms are searched in the symptom-acupoint database, and a relevance score is generated for each matching acupoint. The relevance score sorts the acupoints according to the severity of the symptoms, and after sorting, the acupoints are initially recommended to the user. The relevance scoring model is:

[0076]

[0077] Among them, Score(a k ) indicates acupoint a k The total relevance score, w i represents the weight of the symptom, m represents the total number of standardized symptoms related to the user's symptom, and f(s i ) indicates symptoms in the symptom-acupoint database i The frequency of occurrence, f(a k) indicates acupoint a in the symptom-acupoint database k The frequency of occurrence, f(s i ,a k )Symptoms i and acupoints k frequency of co-occurrence;

[0078] The initially recommended acupoints are filtered according to the user's personal information to generate final acupoints recommended for the user.

[0079] Here is a specific numerical implementation example of the correlation scoring model. User symptoms: s1: headache (severe), s2: insomnia (mild), s3: stiff neck and shoulder (moderate), obtain candidate acupoints: a1: Baihui a2: Fengchi: a3: Hegu, obtain the correlation frequency of user symptoms and candidate acupoints in the symptom-acupoint database, symptom weights: w1=0.5 (headache), w2=0.2 (insomnia), w3=0.3 (stiff neck and shoulder), based on the above scoring calculation, Baihui acupoint (a1) has the highest correlation score, and therefore will be recommended as the most suitable acupoint.

[0080] The working principle and effect of the above technical solution are as follows: users may describe the same symptom in different ways. For example, headache may be described as "headache", "headache", "migraine", etc., and abdominal pain may be described as "stomachache" or "stomachache". In order to cope with this diversity, a standardization process is needed. Standardization helps to unify various descriptive words or phrases into specific symptom categories, so that the system can match more accurately; reduce matching errors caused by inconsistent wording, and improve the accuracy of symptom identification and processing. Users enter their own symptom descriptions through the device interface. The system parses and maps the user-described symptoms to a standardized symptom list through natural language processing (NLP) technology. This step ensures consistency of understanding and accuracy of subsequent processing; use the standardized symptoms to find matching acupoints in the symptom-acupoint database, and each matching acupoint calculates its total relevance score according to the relevance scoring model. According to the calculated relevance score, the acupoints are sorted to generate a preliminary recommended acupoint list; according to the user's personal information (such as age, gender, health status, etc.), the preliminary recommended acupoint list is filtered and adjusted to exclude acupoints that are not suitable or not in high demand. In this way, a more personalized and accurate final recommended acupoint list is formed. Precise and personalized recommendations use natural language processing and standardized symptom mapping to improve the accuracy of symptom understanding; through the relevance scoring model, the acupoint recommendations are reasonably ranked in combination with the severity of symptoms, which improves the targeted nature of the treatment; personalized acupoint screening ensures that acupuncture treatment is more in line with the user's personal health background; the symptom-acupoint database and relevance scoring mechanism automatically identify and associate the most appropriate treatment plans, improving the availability and application efficiency of expert knowledge; improving user experience, simplifying the process from user symptom input to treatment plan generation, and reducing the burden on users to manually select and identify acupoints; providing data-driven intelligent treatment recommendations, enhancing users' trust in electronic acupuncture equipment; improving efficacy and safety, effectively screening and filtering inappropriate acupoint recommendations, reducing unnecessary stimulation, and protecting user safety; because the recommended acupoints are based on a comprehensive consideration of multiple factors, the overall treatment effect is ultimately improved; committed to improving the accuracy, personalization and intelligence of electronic acupuncture treatment, and combining advanced computing models to promote the innovation of health services.

[0081] In one embodiment of the present invention, the electronic acupuncture device scans the target area where the matched acupoints are located, obtains the skin resistance distribution and infrared thermal image of the target area, calculates the acupoint matching degree through a probability model, and locates the acupoints based on the acupoint matching degree, including:

[0082] After recommending the final acupuncture point to the user, the acupuncture point selected by the user is obtained, and the specific image of the acupuncture point on the human body is automatically displayed;

[0083] The electronic acupuncture instrument scans the target area of ​​the acupuncture point on the human body, obtains the resistance distribution and infrared thermal image of the target area, and calculates the acupuncture point matching degree through the probability model. When the acupuncture point matching degree is greater than or equal to the preset threshold, the electronic acupuncture instrument starts a vibration reminder to remind the user to stop moving. Specifically, the probability model is:

[0084]

[0085] Among them, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being in the current active state, X t represents the multimodal physiological signals observed at time t, including temperature and resistance, a k Indicates acupoint k,q(X t ) indicates that when no acupoints are activated, physiological signal X is observed t The background probability distribution, P(X t |a k ) indicates that when acupoint k is in the activated state, the physiological signal X is observed t The conditional probability distribution of Represents KL divergence, which is used to measure the difference between two probability distributions;

[0086] Use deep learning to learn the distribution of physiological signals P(X) under acupoint activation t |a k ) and the background distribution q(X t ) is expressed as .

[0087] The working principle and effect of the above technical solution are as follows: after the final acupoints are recommended to the user, the user selects the specific acupoints they wish to receive acupuncture treatment, and the system automatically provides the selected acupoints with specific diagrams on the human body to help the user intuitively locate and understand the acupoints; the electronic acupuncture instrument scans the target area of ​​the human body where the acupoints selected by the user are located, collects the resistance distribution and infrared thermal image data of the area through sensors, and provides multimodal physiological signals; the matching degree of the acupoints is calculated using a probability model: the signal distribution and background distribution are learned and modeled separately using a deep learning method; when the calculated acupoint matching degree is greater than or equal to a preset threshold, the electronic acupuncture instrument will start a vibration reminder; the user can stop moving after receiving the reminder to ensure accurate positioning and effective electronic acupuncture stimulation. Improve positioning accuracy, by accurately calculating the matching degree of acupoints, and only giving reminders when there is a high degree of matching, the positioning accuracy of the acupuncture instrument is improved. This reduces the risk of misoperation and the possibility of ineffective stimulation; uses deep learning to more accurately capture and interpret multimodal physiological signals, thereby improving the accurate assessment of the activation status of acupoints; automatically displays acupoint images, making it easier for users to find and understand the targeted acupoints; vibration reminders provide real-time feedback, optimizing the smoothness of the device use process; ensures that the acupuncture instrument is only operated when it is highly matched, avoiding inefficient or erroneous stimulation; reduces position deviation during device movement, ensures accurate and effective treatment, and improves the user's sense of security; through deep learning and precise probability calculation, the operation intelligence level and user-friendliness of electronic acupuncture equipment are effectively improved.

[0088] In one embodiment of the present invention, deep learning is used to learn the distribution of physiological signals P(X) under acupoint activation conditions. t |a k ) and the background distribution q(X t ), including:

[0089] Collecting multimodal physiological signal data, including data in an activated state of acupoints and temperature and resistance data in an unactivated state, and preprocessing the data, wherein the preprocessing includes: removing noise and normalizing;

[0090] Construct a deep learning model, namely a variational autoencoder model. The encoder of the autoencoder model converts the input physiological signal X t Encoded as a low-dimensional latent variable z, the output z's mean μ(X t ) and variance σ(X t ), the decoder of the autoencoder model: decodes the latent variable z back to the original physiological signal dimension;

[0091] Use the preprocessed data to train the constructed variational autoencoder model;

[0092] For the background distribution q(X t), using the temperature and resistance data of the acupoints in the unactivated state for training, for each acupoint k in the activated state, the distribution P(X t |a k ), use the data in the corresponding activation state for training, and train a separate variational autoencoder model for each acupoint;

[0093] Use the trained variational autoencoder model to calculate the KL divergence for the background distribution q(X t ), using the trained model q(X t ) type encoder output μq(X t ) and σq(X t ) is used to calculate the distribution P(X t |a k ), use the corresponding P(X t |a k ) model’s encoder output μP(X t |a k ) and σP(X t |a k ) to calculate;

[0094] Conditional distribution P(X t |a k ) is expressed as The background distribution is expressed as q(X t ), the KL divergence is calculated by the following model,

[0095]

[0096] The working principle and effect of the above technical solution are as follows: multimodal physiological signal acquisition and preprocessing, collecting multimodal physiological signal data, including temperature and resistance data in the activated and inactivated states of acupoints, preprocessing the data, including noise removal and normalization, to ensure the quality and consistency of the input data; constructing a variational autoencoder model, in which the encoder encodes the input physiological signal into a low-dimensional latent variable z and outputs the mean and variance of z, and the decoder decodes the latent variable z back to the original physiological signal dimension, thereby realizing the characteristic expression and reconstruction of the signal; using the preprocessed data to train the variational autoencoder model, and for the background distribution, using the temperature and resistance data in the inactivated state for training. For the distribution of each acupoint k in the activated state, the corresponding data in the activated state is used for training, and a separate variational autoencoder model is trained for each acupoint; using the trained variational autoencoder model to calculate the KL divergence. For the background distribution, it is calculated by the encoder output of the model; for the distribution of each acupoint k in the activated state, the encoder output of the corresponding model is used for calculation. Efficient feature extraction, through the encoder-decoder structure of VAE, can effectively extract and learn low-dimensional feature representations of multimodal physiological signals, so as to better capture signal patterns and characteristics; use the mean and variance of the latent variable z to describe the complex distribution characteristics of the signal, and improve the understanding of the signal context and variability; accurate identification and positioning, through training a special model to obtain the characteristic distribution of each acupoint, it is helpful to more accurately identify the activation state of the acupoint; background noise and differences between different individuals are uniformly processed through the background distribution model, which improves the generalization ability of the model; using the learning ability of the VAE model, the signal patterns of different acupoints in different activation states are effectively distinguished, and the system's response ability to various physiological signal states is enhanced; intelligence and precision improvement, based on the combination of deep learning and probabilistic models, provides higher recognition accuracy and efficiency, and significantly improves the application effect of electronic acupuncture technology in intelligent operation

[0097] In one embodiment of the present invention, real-time monitoring is performed to determine whether the scanning area of ​​the electronic acupuncture device is aligned with the recommended acupuncture points and whether the acupuncture device is stationary. When the electronic acupuncture device detects that the acupuncture device is stationary, the electronic acupuncture device outputs current stimulation and adjusts the output current in real time, including:

[0098] Analyze skin resistance and infrared thermal images in real time, and after confirmation of acupoint alignment, detect the static state of the electronic acupuncture device within a preset time;

[0099] When it is determined that the electronic acupuncture device is aligned with the acupoint within the preset time and is in a static state, the output current is dynamically adjusted through the current stimulation model. Specifically, the current stimulation model is:

[0100] I(a k ,t)=I0·[1+γ·sin(2πft)·φ(ak )]

[0101] Among them, I0 represents the basic current intensity, γ represents the proportional coefficient of the amplitude, f is the stimulation frequency, which is determined by the user's symptoms, and φ(a k ) is the acupoint response function;

[0102] Specifically, the acupoint response function is:

[0103]

[0104] Among them, λ represents the sensitivity adjustment coefficient, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being currently active.

[0105] The working principle and effect of the above technical solution are as follows: the basic current intensity is used as the initial basic level of the output current, and the output current I (a k ,t) is a dynamically adjusted version of the basic current, and the adjustment is based on a sinusoidal function, which changes dynamically according to time, stimulation frequency and acupoint response function; the sinusoidal function describes the periodic changes of the current, and the frequency f controls the speed of the current oscillation; γ represents the coefficient for adjusting the current swing range, thereby controlling the sensitivity of the current to fluctuations in time and physiological state; the acupoint response function is defined by a logistic regression curve, which takes into account the skin resistance and infrared thermal image characteristics, and describes the individual sensitivity of the acupoint in an accurate quantitative way; the parameter λ adjusts the sensitivity to the deviation of the response probability from 0.5, thereby controlling the response speed or sensitivity of the current output to the detection conditions. Personalized stimulation can adaptively adjust the current intensity according to the unique response characteristics of each acupoint through real-time monitoring and dynamic adjustment to achieve personalized acupoint stimulation; it uses feedback adjustment output based on physiological response data to improve the accuracy and effectiveness of treatment; with the help of analysis of skin resistance and infrared thermal images, it can better capture the response state of acupoints, thereby optimizing the stimulation adjustment of physiological signals; through more precise current control, it may reduce discomfort or side effects caused by excessive stimulation of the human body; it combines signal analysis, control strategy and personalized parameter adjustment to achieve efficient and safe current stimulation regulation.

[0106] One embodiment of the present invention provides a control system for an electronic acupuncture device, the system comprising:

[0107] The acupoint recommendation module is used to obtain user symptoms and recommend matching acupoints to users based on the symptom-acupoint database;

[0108] The acupuncture point positioning module is used for the electronic acupuncture device to scan the target area where the matched acupuncture points are located, obtain the skin resistance distribution and infrared thermal image of the target area, calculate the acupuncture point matching degree through the probability model, and locate the acupuncture points based on the acupuncture point matching degree.

[0109] The output current module is used to monitor in real time whether the scanning area of ​​the electronic acupuncture device is aligned with the recommended acupuncture points and stops moving. When it is detected that it stops moving, the electronic acupuncture device outputs current stimulation and adjusts the output current according to the symptoms of the user after the stimulation.

[0110] The working principle and effect of the above technical solution are as follows: user symptom input and matching, the user inputs the symptom description through the device interface, uses natural language processing (NLP) technology to map the symptoms to specific symptoms stored in the symptom-acupoint database, and recommends a set of acupoints that are effective for the symptoms to the user based on the symptom-acupoint database. These acupoints have historical treatment data associated with the input symptoms; acupoint positioning and identification, the electronic acupuncture device scans the target area, obtains the skin resistance distribution and infrared thermal image, and uses a probability model to analyze the scanned data to calculate the acupoint matching degree. The acupoint matching degree represents the recognition accuracy of each acupoint under the current physiological conditions, and is calculated in combination with historical data and biometrics; the alignment of the acupuncture device is continuously monitored to ensure that the recommended acupoints can be accurately aimed. When the device is aligned with the recommended acupoints and stops moving, it is ready for stimulation. Once it is confirmed that the device is aligned and stopped, an appropriate amount of current is output for acupoint stimulation. While applying current stimulation, the intensity and frequency of the current stimulation are dynamically adjusted to further optimize the treatment effect. Precise acupoint positioning, by integrating skin resistance and infrared thermal imaging, combined with probability models, can improve the accuracy of acupoint positioning, which reduces human errors and improves treatment effects; by adjusting the current stimulation parameters, the system can respond to the physiological states of different users and provide more accurate and personalized treatment plans; user experience is improved, intelligent symptom matching and automated acupoint detection reduce the complexity of user operations, while also improving user trust and ease of use; it provides a technical upgrade to traditional acupuncture therapy, making it more refined, humane and efficient.

[0111] In one embodiment of the present invention, the acupoint recommendation module comprises:

[0112] A standardized symptom mapping module is used to obtain the symptoms described by the user and map the symptoms described by the user to standardized symptoms;

[0113] The preliminary recommendation module is used to search for acupoints matching the standardized symptoms in the symptom-acupoint database, generate a relevance score for each matching acupoint, and sort the acupoints according to the severity of the symptoms. After sorting, the acupoints are initially recommended to the user. The relevance score model is:

[0114]

[0115] Among them, Score(a k ) indicates acupoint a k The total relevance score, w i represents the weight of the symptom, m represents the total number of standardized symptoms related to the user's symptom, and f(s i ) indicates symptoms in the symptom-acupoint database i The frequency of occurrence, f(a k ) indicates acupoint a in the symptom-acupoint database k The frequency of occurrence, f(s i ,a k )Symptoms i and acupoints k frequency of co-occurrence;

[0116] The final recommendation module is used to filter the initially recommended acupoints according to the user's personal information to generate final acupoints recommended for the user.

[0117] The initially recommended acupoints are filtered according to the user's personal information to generate final acupoints recommended for the user.

[0118] Here is a specific numerical implementation example of the correlation scoring model. User symptoms: s1: headache (severe), s2: insomnia (mild), s3: stiff neck and shoulder (moderate), obtain candidate acupoints: a1: Baihui a2: Fengchi: a3: Hegu, obtain the correlation frequency of user symptoms and candidate acupoints in the symptom-acupoint database, symptom weights: w1=0.5 (headache), w2=0.2 (insomnia), w3=0.3 (stiff neck and shoulder), based on the above scoring calculation, Baihui acupoint (a1) has the highest correlation score, and therefore will be recommended as the most suitable acupoint.

[0119] The working principle and effect of the above technical solution are as follows: users may describe the same symptom in different ways. For example, headache may be described as "headache", "headache", "migraine", etc., and abdominal pain may be described as "stomachache" or "stomachache". In order to cope with this diversity, a standardization process is needed. Standardization helps to unify various descriptive words or phrases into specific symptom categories, so that the system can match more accurately; reduce matching errors caused by inconsistent wording, and improve the accuracy of symptom identification and processing. Users enter their own symptom descriptions through the device interface. The system parses and maps the user-described symptoms to a standardized symptom list through natural language processing (NLP) technology. This step ensures consistency of understanding and accuracy of subsequent processing; use the standardized symptoms to find matching acupoints in the symptom-acupoint database, and each matching acupoint calculates its total relevance score according to the relevance scoring model. According to the calculated relevance score, the acupoints are sorted to generate a preliminary recommended acupoint list; according to the user's personal information (such as age, gender, health status, etc.), the preliminary recommended acupoint list is filtered and adjusted to exclude acupoints that are not suitable or not in high demand. In this way, a more personalized and accurate final recommended acupoint list is formed. Precise and personalized recommendations use natural language processing and standardized symptom mapping to improve the accuracy of symptom understanding; through the relevance scoring model, the acupoint recommendations are reasonably ranked in combination with the severity of symptoms, which improves the targeted nature of the treatment; personalized acupoint screening ensures that acupuncture treatment is more in line with the user's personal health background; the symptom-acupoint database and relevance scoring mechanism automatically identify and associate the most appropriate treatment plans, improving the availability and application efficiency of expert knowledge; improving user experience, simplifying the process from user symptom input to treatment plan generation, and reducing the burden on users to manually select and identify acupoints; providing data-driven intelligent treatment recommendations, enhancing users' trust in electronic acupuncture equipment; improving efficacy and safety, effectively screening and filtering inappropriate acupoint recommendations, reducing unnecessary stimulation, and protecting user safety; because the recommended acupoints are based on a comprehensive consideration of multiple factors, the overall treatment effect is ultimately improved; committed to improving the accuracy, personalization and intelligence of electronic acupuncture treatment, and combining advanced computing models to promote the innovation of health services.

[0120] In one embodiment of the present invention, the acupoint positioning module includes:

[0121] The module for displaying the area where the acupoints are located is used to obtain the acupoints selected by the user after recommending the final acupoints to the user, and automatically display the specific images of the acupoints on the human body;

[0122] The module for calculating the acupuncture point matching degree is used for the electronic acupuncture instrument to scan the target area of ​​the acupuncture point on the human body, obtain the resistance distribution and infrared thermal image of the target area, and calculate the acupuncture point matching degree through the probability model. When the acupuncture point matching degree is greater than or equal to the preset threshold, the electronic acupuncture instrument starts a vibration reminder to remind the user to stop moving. Specifically, the probability model is:

[0123]

[0124] Among them, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being in the current active state, X t represents the multimodal physiological signals observed at time t, including temperature and resistance, a k Indicates acupoint k,q(X t ) indicates that when no acupoints are activated, physiological signal X is observed t The background probability distribution, P(X t |a k ) indicates that when acupoint k is in the activated state, the physiological signal X is observed t The conditional probability distribution of Represents KL divergence, which is used to measure the difference between two probability distributions;

[0125] Deep learning module, used to use deep learning to learn the physiological signal distribution P(X t |a k ) and the background distribution q(X t ) is expressed as .

[0126] The working principle and effect of the above technical solution are as follows: after the final acupoints are recommended to the user, the user selects the specific acupoints they wish to receive acupuncture treatment, and the system automatically provides the selected acupoints with specific diagrams on the human body to help the user intuitively locate and understand the acupoints; the electronic acupuncture instrument scans the target area of ​​the human body where the acupoints selected by the user are located, collects the resistance distribution and infrared thermal image data of the area through sensors, and provides multimodal physiological signals; the matching degree of the acupoints is calculated using a probability model: the signal distribution and background distribution are learned and modeled separately using a deep learning method; when the calculated acupoint matching degree is greater than or equal to a preset threshold, the electronic acupuncture instrument will start a vibration reminder; the user can stop moving after receiving the reminder to ensure accurate positioning and effective electronic acupuncture stimulation. Improve positioning accuracy, by accurately calculating the matching degree of acupoints, and only giving reminders when there is a high degree of matching, the positioning accuracy of the acupuncture instrument is improved. This reduces the risk of misoperation and the possibility of ineffective stimulation; uses deep learning to more accurately capture and interpret multimodal physiological signals, thereby improving the accurate assessment of the activation status of acupoints; automatically displays acupoint images, making it easier for users to find and understand the targeted acupoints; vibration reminders provide real-time feedback, optimizing the smoothness of the device use process; ensures that the acupuncture instrument is only operated when it is highly matched, avoiding inefficient or erroneous stimulation; reduces position deviation during device movement, ensures accurate and effective treatment, and improves the user's sense of security; through deep learning and precise probability calculation, the operation intelligence level and user-friendliness of electronic acupuncture equipment are effectively improved.

[0127] In one embodiment of the present invention, the deep learning module includes:

[0128] A data collection module is used to collect multimodal physiological signal data, including data in an activated state of acupoints and temperature and resistance data in an unactivated state, and preprocess the data, wherein the preprocessing includes: removing noise and normalizing;

[0129] Construct a model module for constructing a deep learning model, namely a variational autoencoder model. The encoder of the autoencoder model converts the input physiological signal X t Encoded as a low-dimensional latent variable z, the output z's mean μ(X t ) and variance σ(X t ), the decoder of the autoencoder model: decodes the latent variable z back to the original physiological signal dimension;

[0130] The training model module is used to train the constructed variational autoencoder model using the preprocessed data. t ), using the temperature and resistance data of the acupoints in the unactivated state for training, for each acupoint k in the activated state, the distribution P(X t |a k), use the data in the corresponding activation state for training, and train a separate variational autoencoder model for each acupoint;

[0131] The KL divergence calculation module is used to calculate the KL divergence using the trained variational autoencoder model. t ), using the trained model q(X t ) type encoder output μq(X t ) and σq(X t ) is used to calculate the distribution P(X t |a k ), use the corresponding P(X t |a k ) model’s encoder output μP(X t |a k ) and σP(X t |a k ) to calculate.

[0132] The working principle and effect of the above technical solution are as follows: multimodal physiological signal acquisition and preprocessing, collecting multimodal physiological signal data, including temperature and resistance data in the activated and inactivated states of acupoints, preprocessing the data, including noise removal and normalization, to ensure the quality and consistency of the input data; constructing a variational autoencoder model, in which the encoder encodes the input physiological signal into a low-dimensional latent variable z and outputs the mean and variance of z, and the decoder decodes the latent variable z back to the original physiological signal dimension, thereby realizing the characteristic expression and reconstruction of the signal; using the preprocessed data to train the variational autoencoder model, and for the background distribution, using the temperature and resistance data in the inactivated state for training. For the distribution of each acupoint k in the activated state, the corresponding data in the activated state is used for training, and a separate variational autoencoder model is trained for each acupoint; using the trained variational autoencoder model to calculate the KL divergence. For the background distribution, it is calculated by the encoder output of the model; for the distribution of each acupoint k in the activated state, the encoder output of the corresponding model is used for calculation. Efficient feature extraction, through the encoder-decoder structure of VAE, can effectively extract and learn low-dimensional feature representations of multimodal physiological signals, so as to better capture signal patterns and characteristics; use the mean and variance of the latent variable z to describe the complex distribution characteristics of the signal, and improve the understanding of the signal context and variability; accurate identification and positioning, through training a special model to obtain the characteristic distribution of each acupoint, it is helpful to more accurately identify the activation state of the acupoint; background noise and differences between different individuals are uniformly processed through the background distribution model, which improves the generalization ability of the model; using the learning ability of the VAE model, the signal patterns of different acupoints in different activation states are effectively distinguished, and the system's response ability to various physiological signal states is enhanced; intelligence and precision improvement, based on the combination of deep learning and probabilistic models, provides higher recognition accuracy and efficiency, and significantly improves the application effect of electronic acupuncture technology in intelligent operation

[0133] In one embodiment of the present invention, the output current module includes:

[0134] A real-time analysis module for analyzing skin resistance and infrared thermal images in real time, and detecting the static state of the electronic acupuncture device within a preset time after the acupoint alignment is confirmed;

[0135] The current adjustment module is used to dynamically adjust the output current through the current stimulation model when it is determined that the electronic acupuncture device is aligned with the acupoint within a preset time and is in a static state. Specifically, the current stimulation model is:

[0136] I(a k ,t)=I0·[1+γ·sin(2πft)·φ(a k )]

[0137] Among them, I0 represents the basic current intensity, γ represents the proportional coefficient of the amplitude, f is the stimulation frequency, which is determined by the user's symptoms, and φ(a k ) is the acupoint response function;

[0138] Specifically, the acupoint response function is:

[0139]

[0140] Among them, λ represents the sensitivity adjustment coefficient, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being currently active.

[0141] The working principle and effect of the above technical solution are as follows: the basic current intensity is used as the initial basic level of the output current, and the dynamic adjustment mechanism output current I (a k ,t) is a dynamically adjusted version of the basic current, and the adjustment is based on a sinusoidal function, which changes dynamically according to time, stimulation frequency and acupoint response function; the sinusoidal function describes the periodic changes of the current, and the frequency f controls the speed of the current oscillation; γ represents the coefficient for adjusting the current swing range, thereby controlling the sensitivity of the current to fluctuations in time and physiological state; the acupoint response function is defined by a logistic regression curve, which takes into account the skin resistance and infrared thermal image characteristics, and describes the individual sensitivity of the acupoint in an accurate quantitative way; the parameter λ adjusts the sensitivity to the deviation of the response probability from 0.5, thereby controlling the response speed or sensitivity of the current output to the detection conditions. Personalized stimulation can adaptively adjust the current intensity according to the unique response characteristics of each acupoint through real-time monitoring and dynamic adjustment to achieve personalized acupoint stimulation; it uses feedback adjustment output based on physiological response data to improve the accuracy and effectiveness of treatment; with the help of analysis of skin resistance and infrared thermal images, it can better capture the response state of acupoints, thereby optimizing the stimulation adjustment of physiological signals; through more precise current control, it may reduce discomfort or side effects caused by excessive stimulation of the human body; it combines signal analysis, control strategy and personalized parameter adjustment to achieve efficient and safe current stimulation regulation.

[0142] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. A control method for an electronic acupuncture device, characterized in that: The method comprises: Obtain user symptoms and recommend matching acupoints for the user based on the symptom-acupoint database; The electronic acupuncture device scans the target area where the matched acupoints are located, obtains the skin resistance distribution and infrared thermal image of the target area, calculates the acupoint matching degree through a probability model, and locates the acupoints based on the acupoint matching degree; Real-time monitoring is performed to determine whether the scanning area of ​​the electronic acupuncture device is aligned with the recommended acupuncture points and is stationary. When stationary acupuncture is detected, the electronic acupuncture device outputs current stimulation and adjusts the output current in real time.

2. The control method of the electronic acupuncture device according to claim 1, characterized in that: Obtain user symptoms and recommend matching acupoints for users based on the symptom-acupoint database, including: Obtain the symptoms described by the user and map the symptoms described by the user to standardized symptoms; Acupoints matching the standardized symptoms are searched in the symptom-acupoint database, and a relevance score is generated for each matching acupoint. The relevance score sorts the acupoints according to the severity of the symptoms, and after sorting, the acupoints are initially recommended to the user. The relevance scoring model is: Among them, Score(a k ) indicates acupoint a k The total relevance score, w i represents the weight of the symptom, m represents the total number of standardized symptoms related to the user's symptom, and f(s i ) indicates symptoms in the symptom-acupoint database i The frequency of occurrence, f(a k ) indicates acupoint a in the symptom-acupoint database k The frequency of occurrence, f(s i , a k )Symptoms i and acupoints k frequency of co-occurrence; The initially recommended acupoints are filtered according to the user's personal information to generate final acupoints recommended for the user.

3. The control method of the electronic acupuncture device according to claim 1, characterized in that: The electronic acupuncture device scans the target area where the matched acupoints are located, obtains the skin resistance distribution and infrared thermal image of the target area, calculates the acupoint matching degree through the probability model, and locates the acupoints based on the acupoint matching degree, including: After recommending the final acupuncture point to the user, the acupuncture point selected by the user is obtained, and the specific image of the acupuncture point on the human body is automatically displayed; The electronic acupuncture instrument scans the target area of ​​the acupuncture point on the human body, obtains the resistance distribution and infrared thermal image of the target area, and calculates the acupuncture point matching degree through the probability model. When the acupuncture point matching degree is greater than or equal to the preset threshold, the electronic acupuncture instrument starts a vibration reminder to remind the user to stop moving. Specifically, the probability model is: Among them, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being in the current activation state, n represents the total number of acupoints in the target area, X t represents the multimodal physiological signals observed at time t, including temperature and resistance, a k Indicates acupoint k,q(X t ) indicates that when no acupoints are activated, physiological signal X is observed t The background probability distribution, P(X t |a k ) indicates that when acupoint k is in the activated state, the physiological signal X is observed t The conditional probability distribution of Represents KL divergence, which is used to measure the difference between two probability distributions; Use deep learning to learn the distribution of physiological signals P(X) under acupoint activation t |a k ) and the background distribution q(X t ) is expressed as .

4. The control method of the electronic acupuncture device according to claim 3, characterized in that: Use deep learning to learn the distribution of physiological signals P(X) under acupoint activation t |a k ) and the background distribution q(X t ), including: Collecting multimodal physiological signal data, including data in an activated state of acupoints and temperature and resistance data in an unactivated state, and preprocessing the data, wherein the preprocessing includes: removing noise and normalizing; Construct a deep learning model, namely a variational autoencoder model. The encoder of the autoencoder model converts the input physiological signal X t Encoded as a low-dimensional latent variable z, the output z's mean μ(X t ) and variance σ(X t ), the decoder of the autoencoder model: decodes the latent variable z back to the original physiological signal dimension; Use the preprocessed data to train the constructed variational autoencoder model; For the background distribution q(X t ), using the temperature and resistance data of the acupoints in the unactivated state for training, for each acupoint k in the activated state, the distribution P(X t |a k ), use the data in the corresponding activation state for training, and train a separate variational autoencoder model for each acupoint; Use the trained variational autoencoder model to calculate the KL divergence for the background distribution q(X t ), using the trained model q(X t ) type encoder output μq(X t ) and σq(X t ) is used to calculate the distribution P(X t |a k ), use the corresponding P(X t |a k ) model’s encoder output μP(X t |a k ) and σP(X t |a k ) to calculate.

5. The control method of the electronic acupuncture device according to claim 1, characterized in that: Real-time monitoring of whether the scanning area of ​​the electronic acupuncture device is aligned with the recommended acupuncture points and is stationary. When it is detected that it is stationary, the electronic acupuncture device outputs current stimulation and adjusts the output current in real time, including: Analyze skin resistance and infrared thermal images in real time, and after confirmation of acupoint alignment, detect the static state of the electronic acupuncture device within a preset time; When it is determined that the electronic acupuncture device is aligned with the acupoint within the preset time and is in a static state, the output current is dynamically adjusted through the current stimulation model. Specifically, the current stimulation model is: I(a k ,t)=I0·[1+γ·sin(2πft)·φ(a k )] Among them, I0 represents the basic current intensity, γ represents the proportional coefficient of the amplitude, f is the stimulation frequency, which is determined by the user's symptoms, and φ(a k ) is the acupoint response function; Specifically, the acupoint response function is: Among them, λ represents the sensitivity adjustment coefficient, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being currently active.

6. A control system for an electronic acupuncture device, characterized in that: The system comprises: The acupoint recommendation module is used to obtain user symptoms and recommend matching acupoints to users based on the symptom-acupoint database; The acupuncture point positioning module is used for the electronic acupuncture device to scan the target area where the matched acupuncture points are located, obtain the skin resistance distribution and infrared thermal image of the target area, calculate the acupuncture point matching degree through the probability model, and locate the acupuncture points based on the acupuncture point matching degree. The output current module is used to monitor in real time whether the scanning area of ​​the electronic acupuncture device is aligned with the recommended acupuncture points and stops moving. When it is detected that it stops moving, the electronic acupuncture device outputs current stimulation and adjusts the output current according to the symptoms of the user after the stimulation.

7. The control system of the electronic acupuncture device according to claim 6, characterized in that: The recommended acupoint module includes: A standardized symptom mapping module is used to obtain the symptoms described by the user and map the symptoms described by the user to standardized symptoms; The preliminary recommendation module is used to search for acupoints matching the standardized symptoms in the symptom-acupoint database, generate a relevance score for each matching acupoint, and sort the acupoints according to the severity of the symptoms. After sorting, the acupoints are initially recommended to the user. The relevance score model is: Among them, Score(a k ) indicates acupoint a k The total relevance score, w i represents the weight of the symptom, m represents the total number of standardized symptoms related to the user's symptom, and f(s i ) indicates symptoms in the symptom-acupoint database i The frequency of occurrence, f(a k ) indicates acupoint a in the symptom-acupoint database k The frequency of occurrence, f(s i , a k )Symptoms i and acupoints k frequency of co-occurrence; The final recommendation module is used to filter the initially recommended acupoints according to the user's personal information to generate final acupoints recommended for the user.

8. The control system of the electronic acupuncture device according to claim 6, characterized in that: The acupuncture point positioning module comprises: The module for displaying the area where the acupoints are located is used to obtain the acupoints selected by the user after recommending the final acupoints to the user, and automatically display the specific images of the acupoints on the human body; The module for calculating the acupuncture point matching degree is used for the electronic acupuncture instrument to scan the target area of ​​the acupuncture point on the human body, obtain the resistance distribution and infrared thermal image of the target area, and calculate the acupuncture point matching degree through the probability model. When the acupuncture point matching degree is greater than or equal to the preset threshold, the electronic acupuncture instrument starts a vibration reminder to remind the user to stop moving. Specifically, the probability model is: Among them, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being in the current active state, X t represents the multimodal physiological signals observed at time t, including temperature and resistance, a k Indicates acupoint k,q(X t ) indicates that when no acupoints are activated, physiological signal X is observed t The background probability distribution, P(X t |a k ) indicates that when acupoint k is in the activated state, the physiological signal X is observed t The conditional probability distribution of Represents KL divergence, which is used to measure the difference between two probability distributions; Deep learning module, used to use deep learning to learn the physiological signal distribution P(X t |a k ) and the background distribution q(X t ) is expressed as .

9. The control system of the electronic acupuncture device according to claim 8, characterized in that: The deep learning module includes: A data collection module is used to collect multimodal physiological signal data, including data in an activated state of acupoints and temperature and resistance data in an unactivated state, and preprocess the data, wherein the preprocessing includes: removing noise and normalizing; Construct a model module for constructing a deep learning model, namely a variational autoencoder model. The encoder of the autoencoder model converts the input physiological signal X t Encoded as a low-dimensional latent variable z, the output z's mean μ(X t ) and variance σ(X t ), the decoder of the autoencoder model: decodes the latent variable z back to the original physiological signal dimension; The training model module is used to train the constructed variational autoencoder model using the preprocessed data. t ), using the temperature and resistance data of the acupoints in the unactivated state for training, for each acupoint k in the activated state, the distribution P(X t |a k ), use the data in the corresponding activation state for training, and train a separate variational autoencoder model for each acupoint; The KL divergence calculation module is used to calculate the KL divergence using the trained variational autoencoder model. t ), using the trained model q(X t ) type encoder output μq(X t ) and σq(X t ) is used to calculate the distribution P(X t |a k ), use the corresponding P(X t |a k ) model’s encoder output μP(X t |a k ) and σP(X t |a k ) to calculate.

10. The control system of the electronic acupuncture device according to claim 6, characterized in that: The output current module comprises: A real-time analysis module for analyzing skin resistance and infrared thermal images in real time, and detecting the static state of the electronic acupuncture device within a preset time after the acupoint alignment is confirmed; The current adjustment module is used to dynamically adjust the output current through the current stimulation model when it is determined that the electronic acupuncture device is aligned with the acupoint within a preset time and is in a static state. Specifically, the current stimulation model is: I(a k ,t)=I0·[1+γ·sin(2πft)·φ(a k )] Among them, I0 represents the basic current intensity, γ represents the proportional coefficient of the amplitude, f is the stimulation frequency, which is determined by the user's symptoms, and φ(a k ) is the acupoint response function; Specifically, the acupoint response function is: Among them, λ represents the sensitivity adjustment coefficient, P(a k |X t ) indicates that when physiological signal X is observed t After that, evaluate the acupoints k The probability of being currently active.

Images