Sleep-aiding mode adaptive switching method and device based on EEG signal analysis

Abstract

The present application provides a method and device for adaptively switching sleep-aid modes based on EEG signal analysis, and relates to the field of sleep-aid optimization technology, including: clustering related user data sets according to predetermined sleep scenarios to obtain multiple user feature sets, and using multiple user feature sets to train and obtain a sleep-aid mode switcher; acquiring the real-time EEG signal of the target user at a fixed point, and using the sleep-aid mode switcher to analyze and obtain an adapted sleep-aid mode, including microcurrent parameters and sound wave parameters; transmitting the microcurrent parameters and sound wave parameters to the sleep-aid control unit to complete the switching of the sleep-aid mode. This application can solve the technical problem that the sleep-aid mode cannot be set specifically according to the EEG state and sleep habits of the new user in the traditional sleep-aid method, resulting in a low degree of adaptation between the sleep-aid mode and the sleep needs, resulting in insufficient effectiveness and practicality of the sleep-aid method, and can provide users with effective and reliable sleep-aid services and enhance user experience.

Description

Technical Field

[0001] The present application relates to the field of sleep aid optimization technology, and in particular to a method and device for adaptively switching sleep aid modes based on electroencephalogram signal analysis. Background Art

[0002] In today's fast-paced life, sleep problems have become a focus of people's attention. In order to improve sleep quality, various sleep-aiding methods and devices have emerged, especially technologies based on EEG signal analysis. These methods use electroencephalogram (EEG) to monitor the user's brain electrical activity, analyze the sleep state, and provide users with personalized sleep-aiding solutions.

[0003] At present, traditional sleep-aiding methods based on EEG signal analysis often rely on fixed sleep-aiding modes, and are unable to set sleep-aiding modes specifically according to the EEG state and sleeping habits of new users. This results in a low degree of adaptability between the sleep-aiding modes and the sleep needs of new users, causing technical problems such as insufficient effectiveness and practicality of sleep-aiding for new users. Summary of the invention

[0004] The purpose of this application is to provide a method and device for adaptively switching sleep-aid modes based on EEG signal analysis, so as to solve the technical problem that the sleep-aid mode cannot be set specifically according to the EEG state and sleep habits of new users in traditional sleep-aid methods based on EEG signal analysis, resulting in a low degree of adaptability between the sleep-aid mode and the sleep needs of new users, resulting in insufficient effectiveness and practicality of sleep aid for new users.

[0005] In view of the above problems, the present application provides a method and device for adaptively switching sleep-aid modes based on EEG signal analysis.

[0006] In a first aspect, the present application provides a method for adaptively switching a sleep-aid mode based on EEG signal analysis, which is implemented by an adaptive sleep-aid mode switching device based on EEG signal analysis, including: receiving the target user's vital sign information, sleep characteristics and sleep environment characteristics through a client, and communicating them to a background processing unit; in the background processing unit, using the vital sign information, sleep characteristics and sleep environment characteristics as constraints, performing similarity association retrieval on a historical user database to obtain an associated user data set; clustering the associated user data set according to a predetermined sleep scene to obtain a plurality of user feature sets, using the plurality of user feature sets to train and obtain a scene recognition large model and a sleep-aid analysis large model, and constructing a sleep-aid mode switcher; obtaining the target user's real-time EEG signal through an EEG signal acquisition unit at a fixed point, communicating the real-time EEG signal to the background processing unit, and using the sleep-aid mode switcher to analyze and obtain an adapted sleep-aid mode, wherein the adapted sleep-aid mode includes microcurrent parameters and sound wave parameters; communicating the microcurrent parameters and sound wave parameters to a sleep-aid control unit to complete the sleep-aid mode switching.

[0007] In the second aspect, the present application also provides a sleep-aid mode adaptive switching device based on EEG signal analysis, which is used to execute a sleep-aid mode adaptive switching method based on EEG signal analysis as described in the first aspect. The device includes an EEG signal acquisition unit, a client, a background processing unit and a sleep-aid control unit, and also includes: an information receiving module, which is used to receive the target user's vital sign information, sleep characteristics and sleep environment characteristics through the client, and transmit them to the background processing unit through communication; an association retrieval module, which is used in the background processing unit to perform similar association retrieval on the historical user database with the vital sign information, sleep characteristics and sleep environment characteristics as constraints to obtain an associated user data set; a model building module A block is used to cluster the associated user data set according to a predetermined sleep scene to obtain multiple user feature sets, and use the multiple user feature sets to train to obtain a scene recognition large model and a sleep-aid analysis large model to build a sleep-aid mode switcher; a sleep-aid mode analysis module is used to obtain the real-time EEG signal of the target user at a fixed point through an EEG signal acquisition unit, and communicate the real-time EEG signal to a background processing unit, and use the sleep-aid mode switcher to analyze and obtain an adapted sleep-aid mode, wherein the adapted sleep-aid mode includes microcurrent parameters and sound wave parameters; a sleep-aid mode switching module is used to communicate the microcurrent parameters and sound wave parameters to a sleep-aid control unit to complete the sleep-aid mode switching.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0009] By taking the target user's vital sign information, sleep characteristics and sleep environment characteristics as constraints, a similar association search is performed on the historical user database to obtain an associated user data set; the associated user data set is further clustered according to the predetermined sleep scene to obtain multiple user feature sets, and the scene recognition large model and the sleep aid analysis large model are trained using multiple user feature sets to construct a sleep aid mode switcher; when providing sleep aid to the target user, the real-time EEG signal is communicated to the background processing unit, and the sleep aid mode switcher is used to analyze and obtain an adapted sleep aid mode, which includes microcurrent parameters and sound wave parameters; finally, the microcurrent parameters and sound wave parameters are communicated to the sleep aid control unit to complete the adaptive switching of the sleep aid mode; the above steps can achieve the technical goal of quickly and accurately setting the sleep aid mode based on the EEG state of the new user, and at the same time, the sleep aid mode can be adapted and optimized according to the sleep habits of the new user to improve the adaptability of the sleep aid mode to the sleep needs of the new user, provide users with effective and reliable sleep aid services, and enhance user experience.

[0010] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented according to the contents of the specification, and in order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are specifically cited below. It should be understood that the content described in this section is not intended to identify the key or important features of the embodiments of the present application, nor is it intended to limit the scope of the present application. Other features of the present application will become easy to understand through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] In order to more clearly illustrate the technical solutions in the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only exemplary, and for ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0012] Figure 1 This is a flowchart of a method for adaptively switching sleep-aid modes based on EEG signal analysis in this application;

[0013] Figure 2 This is a structural schematic diagram of a sleep-aid mode adaptive switching device based on EEG signal analysis in the present application.

[0014] Description of reference numerals:

[0015] An information receiving module 11 , an associated retrieval module 12 , a model building module 13 , a sleep-aiding mode analyzing module 14 , and a sleep-aiding mode switching module 15 . DETAILED DESCRIPTION

[0016] This application provides a method and device for adaptively switching sleep-aid modes based on EEG signal analysis, which solves the technical problem that the sleep-aid mode cannot be set specifically according to the EEG state and sleep habits of new users in traditional sleep-aid methods, resulting in a low degree of adaptability between the sleep-aid mode and the sleep needs of new users, and insufficient effectiveness and practicality of sleep-aiding for new users. The technical goal of quickly and accurately setting the sleep-aid mode based on the EEG state of new users can be achieved, and the sleep-aid mode can be adapted and optimized according to the sleep habits of new users to improve the adaptability of the sleep-aid mode to the sleep needs of new users, provide users with effective and reliable sleep-aid services, and enhance user experience.

[0017] Below, the technical solutions in the present application will be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments of the present application. It should be understood that the present application is not limited to the example embodiments described herein. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present application. It should also be noted that, for the convenience of description, only the parts related to the present application are shown in the accompanying drawings, rather than all of them.

[0018] For example, please refer to the attached Figure 1 The present application provides a method for adaptively switching a sleep-aiding mode based on EEG signal analysis, which is applied to a sleep-aiding mode adaptively switching device based on EEG signal analysis. The device includes an EEG signal acquisition unit, a client, a background processing unit, and a sleep-aiding control unit, and specifically includes the following steps:

[0019] S100: receiving the target user's vital sign information, sleep characteristics and sleep environment characteristics through the client, and communicating and transmitting them to the background processing unit.

[0020] Specifically, the present application provides a method for adaptively switching a sleep-aiding mode based on EEG signal analysis, which is specifically implemented in a sleep-aiding mode adaptive switching device based on EEG signal analysis, wherein the device includes an EEG signal acquisition unit, a client, a background processing unit, and a sleep-aiding control unit, wherein the EEG signal acquisition unit is responsible for real-time monitoring and recording of the user's EEG activity, obtaining relevant brain wave data, and analyzing the user's current sleep state; the client is a user interface, allowing the user to input personal preferences, sleep characteristics, and feedback data; the background processing unit is used to perform in-depth analysis of the received data, use machine learning algorithms to identify the user's sleep pattern, and generate personalized sleep-aiding suggestions and optimization plans; the sleep-aiding control unit is used to automatically adjust the sleep-aiding mode according to the background analysis results, and flexibly switch the stimulation mode (such as microcurrent, sound waves, etc.) to ensure that it matches the user's real-time sleep needs. Through the collaborative work of these components, the device can realize dynamic adjustment of the sleep-aiding mode according to the EEG state and sleep habits of the new user, improve the adaptability and personalization of the sleep-aiding effect, and meet the needs of new users for high-quality sleep.

[0021] The target user's physical sign information, sleep characteristics and sleep environment characteristics are received through the client, the physical sign information includes but is not limited to age, gender, height, physical signs and lifestyle; the sleep characteristics include but are not limited to sleep duration, sleep duration and awakening times; the sleep environment characteristics include but are not limited to temperature, humidity, noise characteristics and space layout. The target user's physical sign information, sleep characteristics and sleep environment characteristics are then communicated to the background processing unit for subsequent analysis.

[0022] S200: In the background processing unit, a similarity association search is performed on a historical user database based on the physical sign information, sleep characteristics and sleep environment characteristics as constraints to obtain an associated user data set.

[0023] Specifically, in the background processing unit, the vital sign information, sleep characteristics and sleep environment characteristics are used as similarity comparison conditions during retrieval, and a similarity association retrieval is performed on the historical user database, wherein the historical user database contains a multidimensional data set of multiple historical users, and the multidimensional data includes but is not limited to the user's vital sign information, sleep characteristics, environmental characteristics and historical sleep-aiding data; firstly, based on the similarity comparison conditions, a suitable similarity measurement method (such as Euclidean distance, cosine similarity, etc.) is used to calculate the similarity between the current user and each user in the historical user database; then, according to the set similarity threshold, the user data set that is most similar to the current user's vital sign information, sleep characteristics and environmental characteristics is retrieved from the historical user database to obtain the associated user data set.

[0024] S300: Clustering the associated user data set according to the predetermined sleep scene to obtain multiple user feature sets, using the multiple user feature sets to train and obtain a scene recognition model and a sleep aid analysis model, and constructing a sleep aid mode switcher.

[0025] Specifically, a predetermined sleep scene is configured, and the predetermined sleep scene includes light sleep, mild sleep, deep sleep and rapid eye movement sleep, wherein each scene has specific brain wave characteristics and user needs. Then, the associated user data set is clustered according to the predetermined sleep scene, and each clustering result should correspond to a specific sleep scene, such as a light sleep characteristic user, a deep sleep characteristic user, etc.; and then the clustering results are sorted into multiple user characteristic sets.

[0026] A scene recognition large model and a sleep-aid analysis large model are constructed. The scene recognition large model is used to perform sleep scene recognition based on the user's EEG data. The sleep-aid analysis large model is constructed based on the principles of machine learning and integrated learning. It is used to combine the user's sleep scene and perform sleep-aid mode analysis based on the user's EEG data, wherein the sleep-aid mode includes microcurrent parameters and sound wave parameters. The microcurrent parameters include current waveform, current amplitude and current frequency, and the sound wave parameters include sound signal waveform, sound signal frequency and volume. Then, the scene recognition large model and the sleep-aid analysis large model are trained using the multiple user feature sets as training data to obtain a scene recognition large model and a sleep-aid analysis large model that meet the expected conditions. Finally, the output layer of the scene recognition large model is connected to the input layer of the sleep-aid analysis large model to construct a sleep-aid mode switcher. By constructing a sleep-aid mode switcher, the technical goal of quickly and accurately setting the sleep-aid mode according to the user's EEG state can be achieved, and the automation and intelligence of the sleep-aid mode setting can be improved.

[0027] S400: acquiring the real-time EEG signal of the target user at a fixed point through an EEG signal acquisition unit, transmitting the real-time EEG signal to a background processing unit, and analyzing the sleep-aiding mode switcher to obtain an adapted sleep-aiding mode, wherein the adapted sleep-aiding mode includes microcurrent parameters and sound wave parameters.

[0028] Specifically, first, the EEG activity of the target user is continuously monitored through the EEG signal acquisition unit to obtain real-time EEG signal data. Then the collected real-time EEG signal is transmitted to the background processing unit through a secure communication channel to ensure the integrity and security of the data. Then in the background processing unit, the real-time EEG signal is analyzed using the sleep aid mode switcher, that is, first the scene recognition large model is used to determine the current user's sleep scene, and then the recognition result is combined with the user's EEG data, and the sleep aid analysis large model is used to output an adapted sleep aid mode. The adapted sleep aid mode includes microcurrent parameters and sound wave parameters. The microcurrent parameters include current waveform, current amplitude and current frequency, etc. These parameters are dynamically adjusted according to the user's EEG state; the sound wave parameters include sound signal waveform, sound signal frequency and volume, and can provide corresponding sound wave stimulation according to the user's sleep needs.

[0029] S500: The microcurrent parameters and the sound wave parameters are communicated and transmitted to the sleep aid control unit to complete the sleep aid mode switching.

[0030] Specifically, the microcurrent parameters and sound wave parameters are communicated and transmitted to the sleep aid control unit, and then the microcurrent stimulation device is controlled to output appropriate current according to the received microcurrent parameters. The output of the sound wave device, including the sound wave frequency and volume, is adjusted according to the received sound wave parameters to create a sound environment conducive to sleep. The sleep aid control unit performs adaptive switching of the sleep aid mode according to the transmitted parameters. Through the combined effects of microcurrent stimulation and sound wave therapy, effective and reliable sleep aid services are provided to users, thereby optimizing the user's sleep experience.

[0031] The method for adaptively switching sleep-aid modes based on EEG signal analysis is applied to a sleep-aid mode adaptive switching device based on EEG signal analysis, which can solve the technical problem that the sleep-aid mode cannot be set specifically according to the EEG state and sleeping habits of the new user in the traditional sleep-aid method based on EEG signal analysis, resulting in a low degree of adaptability between the sleep-aid mode and the sleep needs of the new user, resulting in insufficient effectiveness and practicality of the sleep aid for the new user. By taking the target user's vital sign information, sleep characteristics and sleep environment characteristics as constraints, a similar association search is performed on the historical user database to obtain an associated user data set; the associated user data set is further clustered according to the predetermined sleep scene to obtain multiple user feature sets, and the scene recognition large model and the sleep aid analysis large model are trained using multiple user feature sets to construct a sleep aid mode switcher; when providing sleep aid to the target user, the real-time EEG signal is communicated to the background processing unit, and the sleep aid mode switcher is used to analyze and obtain an adapted sleep aid mode, which includes microcurrent parameters and sound wave parameters; finally, the microcurrent parameters and sound wave parameters are communicated to the sleep aid control unit to complete the adaptive switching of the sleep aid mode; the above steps can achieve the technical goal of quickly and accurately setting the sleep aid mode based on the EEG state of the new user, and at the same time, the sleep aid mode can be adapted and optimized according to the sleep habits of the new user to improve the adaptability of the sleep aid mode to the sleep needs of the new user, provide users with effective and reliable sleep aid services, and enhance user experience.

[0032] Further, the associated user data set is clustered according to the predetermined sleep scene to obtain multiple user feature sets. The present application includes:

[0033] Acquire an associated user data set, wherein the associated user data includes sample EEG data, identified sleep scenes, and sample sleep-aid modes, wherein the sample sleep-aid modes include sample microcurrent parameters and sample sound wave parameters; configure multiple sleep scenes, wherein the multiple sleep scenes include shallow sleep, light sleep, deep sleep, and rapid eye movement sleep; based on the identified sleep scenes, cluster the associated user data set according to the multiple sleep scenes to obtain multiple user feature sets.

[0034] Specifically, an associated user data set is obtained, wherein the associated user data set includes sample data of multiple associated users, the associated user data includes sample EEG data, identified sleep scenes and sample sleep-aid modes, the sample sleep-aid modes include sample microcurrent parameters and sample sound wave parameters, and the identified sleep scene can be determined through subjective identification by the user or based on analysis of the user's sleep habits.

[0035] Then, multiple sleep scenes are configured, including light sleep, mild sleep, deep sleep and rapid eye movement sleep. Different sleep scenes correspond to different brain wave characteristics and user sleep needs. Then, based on the identified sleep scenes, the associated user data sets are clustered according to the multiple sleep scenes, that is, the associated user data belonging to the same sleep scene are clustered into one category to obtain multiple user feature sets, where each user feature set corresponds to a sleep scene.

[0036] Furthermore, the scene recognition model and the sleep aid analysis model are obtained by training the multiple user feature sets to construct a sleep aid mode switcher. The application includes:

[0037] The multiple user feature sets are divided into multiple EEG data sets and multiple sleep-aid mode sets; signal frequencies under the multiple sleep scenes are clustered according to the multiple EEG data sets to determine multiple signal frequency bands, and the scene recognition large model is constructed based on the mapping relationship between the multiple sleep scenes and the multiple signal frequency bands; a first sleep scene is randomly selected from the multiple sleep scenes, and a first EEG data set and a first sleep-aid mode set of the first sleep scene are obtained, wherein the first sleep-aid mode set includes a first microcurrent parameter set and a first sound wave parameter set, the microcurrent parameters include current waveform, current amplitude and current frequency, and the sound wave parameters include sound signal waveform, sound signal frequency and volume; a first sleep-aid analysis branch is obtained by training with the first EEG data set, the first microcurrent parameter set and the first sound wave parameter set, and a first sleep-aid analysis unit is constructed in combination with the first sleep scene; a sleep-aid analysis large model is constructed based on the mapping of multiple sleep scenes and multiple sleep-aid analysis units, and the output layer of the scene recognition large model is connected to the input layer of the sleep-aid analysis large model to generate the sleep-aid mode switcher.

[0038] Specifically, first, the multiple user feature sets are divided into multiple EEG data sets and multiple sleep-aid mode sets. Then, the signal frequencies under the multiple sleep scenarios are clustered according to the multiple EEG data sets, that is, for each EEG data set, relevant signal features are extracted, including spectrum data and power spectrum density, and then the signal frequencies under the same sleep scenario are clustered to determine the signal frequency bands corresponding to each sleep scene, and multiple signal frequency bands are obtained. Further based on the mapping association between sleep scenes and signal frequency bands, with signal frequency bands as child nodes and sleep scenes as leaf nodes of child nodes, a large scene recognition model is constructed based on multiple sleep scenarios and multiple signal frequency bands.

[0039] Randomly select any sleep scene from the multiple sleep scenes as the first sleep scene, obtain the first EEG data set and the first sleep-aiding mode set of the first sleep scene, wherein the first sleep-aiding mode set includes a first microcurrent parameter set and a first sound wave parameter set, the microcurrent parameters include current waveform, current amplitude and current frequency, and the sound wave parameters include sound signal waveform, sound signal frequency and volume. Then, construct a first sleep-aiding analysis branch according to multiple sleep-aiding analysis operators, and the multiple sleep-aiding analysis operators include but are not limited to BP neural network, support vector machine and random decision forest; use the first EEG data set, the first microcurrent parameter set and the first sound wave parameter set as training data, and perform supervised training on the first sleep-aiding analysis branch to obtain the first sleep-aiding analysis branch that meets the convergence conditions, establish a mapping relationship between the first sleep scene and the first sleep-aiding analysis branch, and construct the first sleep-aiding analysis unit.

[0040] Based on the decision tree principle, with sleep scenes as sub-nodes and sleep-aid analysis units as leaf nodes of sub-nodes, a sleep-aid analysis large model is constructed based on the mapping of multiple sleep scenes and multiple sleep-aid analysis units, and the output layer of the scene recognition large model is connected to the input layer of the sleep-aid analysis large model, that is, the recognition result of the scene recognition large model is used as the input data of the sleep-aid analysis large model to generate the sleep-aid mode switcher.

[0041] Further, to determine multiple signal frequency bands, the present application also includes the following steps:

[0042] Based on the multiple EEG data sets, a signal frequency fluctuation range is determined, and the signal frequency fluctuation range is divided according to a predetermined frequency step to determine a plurality of frequency intervals; scene identification is performed on the multiple EEG data sets according to the multiple sleep scenes, and the multiple identified data sets are distributed in the multiple frequency intervals, and the multiple frequency intervals are marked according to the sleep scene with the largest data proportion to determine a plurality of marked frequency intervals; adjacent fusion of the multiple marked frequency intervals is performed according to the same sleep scene to obtain the multiple signal frequency bands.

[0043] Specifically, first, extract spectrum information from multiple EEG data sets, analyze the fluctuation range of signal frequency, including low-frequency and high-frequency bands, and then calculate the frequency fluctuation in each data set, determine the minimum and maximum frequency values, and obtain the frequency fluctuation range. Then divide the signal frequency fluctuation range according to a predetermined frequency step size, which can be set according to the division accuracy requirement. The higher the accuracy requirement, the smaller the frequency step size, such as 0.05Hz or 0.1Hz, to determine multiple frequency intervals.

[0044] The multiple EEG data sets are scene-identified according to the multiple sleep scenes, each data sample is associated with its corresponding sleep scene, and the multiple identified data sets are distributed in the multiple frequency intervals, and then the proportion of the identified scenes in each frequency interval is counted, and the multiple frequency intervals are marked with the sleep scene with the largest data proportion, and multiple marked frequency intervals are determined. Finally, the multiple marked frequency intervals are adjacently fused according to the same sleep scene, and the frequency intervals are merged to obtain the multiple signal frequency bands.

[0045] Further, the first EEG data set, the first microcurrent parameter set and the first sound wave parameter set are used to train a first sleep aid analysis branch, and the present application further includes the following steps:

[0046] Based on multiple sleep-aid analysis operators, an initial microcurrent analysis channel and an initial sound wave analysis channel are constructed; the initial microcurrent analysis channel is supervised trained and verified using the first EEG data set and the first microcurrent parameter set to obtain a microcurrent analysis channel; the initial sound wave analysis channel is supervised trained and verified using the first EEG data set and the first sound wave parameter set to obtain a sound wave analysis channel; the input layer, the microcurrent analysis channel, the sound wave analysis channel and the output layer are fused to obtain the first sleep-aid analysis branch.

[0047] Specifically, multiple sleep-aid analysis operators are configured, and the multiple sleep-aid analysis operators include at least BP neural network, support vector machine and random decision forest, and then an initial microcurrent analysis channel and an initial sound wave analysis channel are constructed based on the multiple sleep-aid analysis operators. Then, the initial microcurrent analysis channel is supervised trained and verified using the first EEG data set and the first microcurrent parameter set to obtain a microcurrent analysis channel. The initial sound wave analysis channel is supervised trained and verified using the first EEG data set and the first sound wave parameter set to obtain a sound wave analysis channel, wherein the construction and training method of the sound wave analysis channel is the same as that of the microcurrent analysis channel, and those skilled in the art may refer to the construction and training method of the microcurrent analysis channel.

[0048] Finally, the input layer, microcurrent analysis channel, sound wave analysis channel and output layer are integrated, that is, the microcurrent analysis channel and the sound wave analysis channel share the input layer and output layer to construct the first sleep-aid analysis branch.

[0049] Further, to obtain a microcurrent analysis channel, the present application also includes the following steps:

[0050] Configure multiple sleep-aid analysis operators, which include at least BP neural network, support vector machine and random decision forest; construct a first microcurrent analysis model, a second microcurrent analysis model and a third microcurrent analysis model based on BP neural network, support vector machine and random decision forest respectively; form a first training set according to the first EEG data set and the first microcurrent parameter set, and divide the first training set into three equal parts to obtain three training sets; use the three training sets to perform supervised training and cross-validation training on the first microcurrent analysis model, the second microcurrent analysis model and the third microcurrent analysis model respectively to obtain a first convergence analysis model, a second convergence analysis model and a third convergence analysis model that meet the convergence constraints, as well as the corresponding first analysis accuracy, second analysis accuracy and third analysis accuracy; configure a weight ratio according to the first analysis accuracy, the second analysis accuracy and the third analysis accuracy, integrate the first convergence analysis model, the second convergence analysis model and the third convergence analysis model based on the weight ratio, and construct the microcurrent analysis channel.

[0051] Specifically, multiple sleep-aid analysis operators are configured, and the multiple sleep-aid analysis operators include at least BP neural network, support vector machine and random decision forest. Those skilled in the art can also select additional sleep-aid analysis operators according to actual needs. Then, based on BP neural network, support vector machine and random decision forest, a first microcurrent analysis model, a second microcurrent analysis model and a third microcurrent analysis model are constructed respectively. On the other hand, a first training set is formed according to the first EEG data set and the first microcurrent parameter set, and the first training set is divided into three equal parts to obtain three training sets.

[0052] The three training sets are used to perform supervised training on the first microcurrent analysis model, the second microcurrent analysis model and the third microcurrent analysis model respectively. For example, when training the first microcurrent analysis model, multiple iterations are performed through forward propagation and back propagation algorithms to optimize weights and biases. When a predetermined number of training times is met, the three training sets are exchanged, and the first microcurrent analysis model, the second microcurrent analysis model and the third microcurrent analysis model are subjected to verification training to obtain the first convergence analysis model, the second convergence analysis model, the third convergence analysis model that meet the convergence constraints, and the corresponding first analysis accuracy, the second analysis accuracy and the third analysis accuracy, where the analysis accuracy is the output accuracy of the model.

[0053] Then, the weight ratio is configured according to the first analysis accuracy, the second analysis accuracy and the third analysis accuracy, wherein the greater the accuracy, the greater the corresponding weight, and the weight configuration can be performed based on the existing coefficient of variation method; ensemble learning is a machine learning technology that combines the prediction results of multiple base learners to improve the performance and accuracy of the overall model, and finally, based on the ensemble learning principle, the first convergence analysis model, the second convergence analysis model and the third convergence analysis model are integrated according to the weight ratio to construct a microcurrent analysis channel, and the output of the microcurrent analysis channel is the weighted calculation result of the output of the first convergence analysis model, the second convergence analysis model and the third convergence analysis model. The construction and training method of the acoustic wave analysis channel is the same as that of the microcurrent analysis channel, and those skilled in the art can refer to the construction and training method of the microcurrent analysis channel.

[0054] Furthermore, after building the sleep-aid mode switcher, it also includes:

[0055] Monitor the sleep characteristics of the target user, divide the sleep scenes and evaluate the sleep quality according to the sleep characteristics, and determine the actual sleep scenes and sleep quality deviations; perform sleep aid mode compensation analysis according to the sleep quality deviations, and determine the microcurrent compensation parameters and the sound wave compensation parameters; time-align the EEG signal of the target user with the actual sleep scene, the microcurrent compensation parameters, and the sound wave compensation parameters to obtain first compensation data, second compensation data, and third compensation data; use the first compensation data to perform incremental learning on the scene recognition large model of the sleep aid mode switcher, and use the second compensation data and the third compensation data to perform incremental learning on the sleep aid analysis large model of the sleep aid mode switcher.

[0056] Specifically, the sleep characteristics of the target user are monitored, including the duration of falling asleep, the duration of sleep, the number of awakenings, etc., and then the sleep scenes are divided and the sleep quality is evaluated based on the sleep characteristics. For example, the overall sleep quality is evaluated by combining the monitored duration of falling asleep, the duration of sleep, and the number of awakenings, and the gap between the actual sleep scene and the ideal sleep scene is calculated to determine the deviation between the actual sleep scene and the sleep quality. Further, a compensation analysis of the sleep aid mode is performed based on the sleep quality deviation. The machine learning model can be used to analyze historical data for compensation analysis to obtain microcurrent compensation parameters and sound wave compensation parameters.

[0057] Then, the EEG signal of the target user is time-aligned and data-fused with the actual sleep scene, microcurrent compensation parameters, and sound wave compensation parameters to obtain first compensation data, second compensation data, and third compensation data. Incremental learning is a machine learning method that allows the model to gradually learn new data without retraining from scratch. The model can update its parameters when receiving new data without the need for a complete data set, thereby saving computing resources and time.

[0058] Then, the first compensation data is used to perform incremental learning on the scene recognition large model of the sleep aid mode switch, and the second compensation data and the third compensation data are used to perform incremental learning on the sleep aid analysis large model of the sleep aid mode switch. Through incremental learning of the model, the adaptability and effectiveness of the sleep aid mode switch can be continuously improved, and it can more accurately adapt to the changing needs of users, improve the personalization level of the sleep aid mode, and ensure that the best sleep aid service is provided to users.

[0059] Furthermore, the present application also includes:

[0060] In the background processing unit, a signal deviation between the real-time EEG signal and the EEG signal of the previous adjacent monitoring node is identified, and if the signal deviation meets a predetermined deviation threshold, the real-time EEG signal is input into the sleep-aiding mode switcher.

[0061] Specifically, in the background processing unit, the signal deviation between the real-time EEG signal and the EEG signal of the previous adjacent monitoring node is identified; a predetermined deviation threshold is configured, and the predetermined deviation threshold is used to judge the significant change of the user's EEG state, which can be set according to the actual scenario and needs; the signal deviation is further judged according to the predetermined deviation threshold, if the signal deviation is greater than the predetermined deviation threshold, the real-time EEG signal is input into the sleep aid mode switcher, and the sleep aid mode switcher automatically adjusts the sleep aid mode according to the newly input data to better meet the user's current sleep needs; if the signal deviation is less than or equal to the predetermined deviation threshold, the mode switching analysis is not performed.

[0062] In summary, the sleep-aiding mode adaptive switching method based on EEG signal analysis provided by the present application has the following technical effects:

[0063] By taking the target user's vital sign information, sleep characteristics and sleep environment characteristics as constraints, a similar association search is performed on the historical user database to obtain an associated user data set; the associated user data set is further clustered according to the predetermined sleep scene to obtain multiple user feature sets, and the scene recognition large model and the sleep aid analysis large model are trained using multiple user feature sets to construct a sleep aid mode switcher; when providing sleep aid to the target user, the real-time EEG signal is communicated to the background processing unit, and the sleep aid mode switcher is used to analyze and obtain an adapted sleep aid mode, which includes microcurrent parameters and sound wave parameters; finally, the microcurrent parameters and sound wave parameters are communicated to the sleep aid control unit to complete the adaptive switching of the sleep aid mode; the above steps can achieve the technical goal of quickly and accurately setting the sleep aid mode based on the EEG state of the new user, and at the same time, the sleep aid mode can be adapted and optimized according to the sleep habits of the new user to improve the adaptability of the sleep aid mode to the sleep needs of the new user, provide users with effective and reliable sleep aid services, and enhance user experience.

[0064] Embodiment 2, based on the sleep-aiding mode adaptive switching method based on EEG signal analysis in the aforementioned embodiment, the present application also provides a sleep-aiding mode adaptive switching device based on EEG signal analysis, the device includes an EEG signal acquisition unit, a client, a background processing unit and a sleep-aiding control unit, please refer to the attached Figure 2 , also includes:

[0065] The information receiving module 11 is used to receive the target user's vital sign information, sleep characteristics and sleep environment characteristics through the client, and transmit them to the background processing unit by communication; the association retrieval module 12 is used to perform similar association retrieval on the historical user database in the background processing unit, with the vital sign information, sleep characteristics and sleep environment characteristics as constraints, to obtain the associated user data set; the model building module 13 is used to cluster the associated user data set according to the predetermined sleep scene to obtain multiple user feature sets, and use the multiple user feature sets to train and obtain the scene recognition large model and the sleep aid analysis large model to build a sleep aid mode switcher; the sleep aid mode analysis module 14 is used to obtain the real-time EEG signal of the target user at a fixed point through the EEG signal acquisition unit, communicate the real-time EEG signal to the background processing unit, and use the sleep aid mode switcher to analyze and obtain the adapted sleep aid mode, wherein the adapted sleep aid mode includes microcurrent parameters and sound wave parameters; the sleep aid mode switching module 15 is used to communicate the microcurrent parameters and sound wave parameters to the sleep aid control unit to complete the sleep aid mode switching.

[0066] Furthermore, the sleep-aiding mode adaptive switching device based on EEG signal analysis is also used for:

[0067] Acquire an associated user data set, wherein the associated user data includes sample EEG data, identified sleep scenes, and sample sleep-aid modes, wherein the sample sleep-aid modes include sample microcurrent parameters and sample sound wave parameters; configure multiple sleep scenes, wherein the multiple sleep scenes include shallow sleep, light sleep, deep sleep, and rapid eye movement sleep; based on the identified sleep scenes, cluster the associated user data set according to the multiple sleep scenes to obtain multiple user feature sets.

[0068] Furthermore, the sleep-aiding mode adaptive switching device based on EEG signal analysis is also used for:

[0069] The multiple user feature sets are divided into multiple EEG data sets and multiple sleep-aid mode sets; signal frequencies under the multiple sleep scenes are clustered according to the multiple EEG data sets to determine multiple signal frequency bands, and the scene recognition large model is constructed based on the mapping relationship between the multiple sleep scenes and the multiple signal frequency bands; a first sleep scene is randomly selected from the multiple sleep scenes, and a first EEG data set and a first sleep-aid mode set of the first sleep scene are obtained, wherein the first sleep-aid mode set includes a first microcurrent parameter set and a first sound wave parameter set, the microcurrent parameters include current waveform, current amplitude and current frequency, and the sound wave parameters include sound signal waveform, sound signal frequency and volume; a first sleep-aid analysis branch is obtained by training with the first EEG data set, the first microcurrent parameter set and the first sound wave parameter set, and a first sleep-aid analysis unit is constructed in combination with the first sleep scene; a sleep-aid analysis large model is constructed based on the mapping of multiple sleep scenes and multiple sleep-aid analysis units, and the output layer of the scene recognition large model is connected to the input layer of the sleep-aid analysis large model to generate the sleep-aid mode switcher.

[0070] Furthermore, the sleep-aiding mode adaptive switching device based on EEG signal analysis is also used for:

[0071] Based on the multiple EEG data sets, a signal frequency fluctuation range is determined, and the signal frequency fluctuation range is divided according to a predetermined frequency step to determine a plurality of frequency intervals; scene identification is performed on the multiple EEG data sets according to the multiple sleep scenes, and the multiple identified data sets are distributed in the multiple frequency intervals, and the multiple frequency intervals are marked according to the sleep scene with the largest data proportion to determine a plurality of marked frequency intervals; adjacent fusion of the multiple marked frequency intervals is performed according to the same sleep scene to obtain the multiple signal frequency bands.

[0072] Furthermore, the sleep-aiding mode adaptive switching device based on EEG signal analysis is also used for:

[0073] Based on multiple sleep-aid analysis operators, an initial microcurrent analysis channel and an initial sound wave analysis channel are constructed; the initial microcurrent analysis channel is supervised trained and verified using the first EEG data set and the first microcurrent parameter set to obtain a microcurrent analysis channel; the initial sound wave analysis channel is supervised trained and verified using the first EEG data set and the first sound wave parameter set to obtain a sound wave analysis channel; the input layer, the microcurrent analysis channel, the sound wave analysis channel and the output layer are fused to obtain the first sleep-aid analysis branch.

[0074] Furthermore, the sleep-aiding mode adaptive switching device based on EEG signal analysis is also used for:

[0075] Configure multiple sleep-aid analysis operators, which include at least BP neural network, support vector machine and random decision forest; construct a first microcurrent analysis model, a second microcurrent analysis model and a third microcurrent analysis model based on BP neural network, support vector machine and random decision forest respectively; form a first training set according to the first EEG data set and the first microcurrent parameter set, and divide the first training set into three equal parts to obtain three training sets; use the three training sets to perform supervised training and cross-validation training on the first microcurrent analysis model, the second microcurrent analysis model and the third microcurrent analysis model respectively to obtain a first convergence analysis model, a second convergence analysis model and a third convergence analysis model that meet the convergence constraints, as well as the corresponding first analysis accuracy, second analysis accuracy and third analysis accuracy; configure a weight ratio according to the first analysis accuracy, the second analysis accuracy and the third analysis accuracy, integrate the first convergence analysis model, the second convergence analysis model and the third convergence analysis model based on the weight ratio, and construct the microcurrent analysis channel.

[0076] Furthermore, the sleep-aiding mode adaptive switching device based on EEG signal analysis is also used for:

[0077] Monitor the sleep characteristics of the target user, divide the sleep scenes and evaluate the sleep quality according to the sleep characteristics, and determine the actual sleep scenes and sleep quality deviations; perform sleep aid mode compensation analysis according to the sleep quality deviations, and determine the microcurrent compensation parameters and the sound wave compensation parameters; time-align the EEG signal of the target user with the actual sleep scene, the microcurrent compensation parameters, and the sound wave compensation parameters to obtain first compensation data, second compensation data, and third compensation data; use the first compensation data to perform incremental learning on the scene recognition large model of the sleep aid mode switcher, and use the second compensation data and the third compensation data to perform incremental learning on the sleep aid analysis large model of the sleep aid mode switcher.

[0078] Furthermore, the sleep-aiding mode adaptive switching device based on EEG signal analysis is also used for:

[0079] In the background processing unit, a signal deviation between the real-time EEG signal and the EEG signal of the previous adjacent monitoring node is identified, and if the signal deviation meets a predetermined deviation threshold, the real-time EEG signal is input into the sleep-aiding mode switcher.

[0080] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The method and specific examples of an adaptive switching method of a sleep-aiding mode based on EEG signal analysis in the aforementioned embodiment 1 are also applicable to an adaptive switching device of a sleep-aiding mode based on EEG signal analysis in this embodiment. Through the aforementioned detailed description of an adaptive switching method of a sleep-aiding mode based on EEG signal analysis, those skilled in the art can clearly know the adaptive switching device of a sleep-aiding mode based on EEG signal analysis in this embodiment, so for the sake of brevity of the specification, it will not be described in detail here. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the description of the method part.

[0081] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

[0082] Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application belong to the scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. A sleep-aiding mode adaptive switching method based on EEG signal analysis, characterized in that: The device is applied to the sleep-aiding mode adaptive switching device based on EEG signal analysis, and includes an EEG signal acquisition unit, a client, a background processing unit and a sleep-aiding control unit, including: Receive the target user's vital sign information, sleep characteristics and sleep environment characteristics through the client, and transmit them to the background processing unit; In the background processing unit, similarity association retrieval is performed on the historical user database based on the physical sign information, sleep characteristics and sleep environment characteristics as constraints to obtain an associated user data set; Clustering the associated user data set according to the predetermined sleep scene to obtain multiple user feature sets, using the multiple user feature sets to train and obtain a scene recognition model and a sleep aid analysis model, and constructing a sleep aid mode switcher; The real-time EEG signal of the target user is acquired at a fixed point by an EEG signal acquisition unit, the real-time EEG signal is transmitted to a background processing unit, and the sleep-aiding mode switcher is used to analyze and obtain an adapted sleep-aiding mode, wherein the adapted sleep-aiding mode includes microcurrent parameters and sound wave parameters; The microcurrent parameters and the sound wave parameters are communicated and transmitted to the sleep aid control unit to complete the sleep aid mode switching; The associated user data set is clustered according to the predetermined sleep scene to obtain multiple user feature sets, including: Acquire an associated user data set, wherein the associated user data includes sample EEG data, identified sleep scenes, and sample sleep-aiding modes, wherein the sample sleep-aiding modes include sample microcurrent parameters and sample sound wave parameters; Configure multiple sleep scenes, including light sleep, light sleep, deep sleep and rapid eye movement sleep; Based on the identified sleep scene, clustering the associated user data set according to the multiple sleep scenes to obtain multiple user feature sets; The method of using the plurality of user feature sets to train and obtain a scene recognition large model and a sleep aid analysis large model to construct a sleep aid mode switcher includes: Dividing the multiple user feature sets into multiple EEG data sets and multiple sleep-aiding mode sets; Performing signal frequency clustering under the multiple sleep scenarios according to the multiple EEG data sets to determine multiple signal frequency bands, and constructing the scene recognition large model based on the mapping relationship between the multiple sleep scenarios and the multiple signal frequency bands; Randomly selecting a first sleep scene from the multiple sleep scenes, and obtaining a first EEG data set and a first sleep-aiding mode set of the first sleep scene, wherein the first sleep-aiding mode set includes a first microcurrent parameter set and a first sound wave parameter set, the microcurrent parameters include a current waveform, a current amplitude, and a current frequency, and the sound wave parameters include a sound signal waveform, a sound signal frequency, and a volume; Using the first EEG data set, the first microcurrent parameter set, and the first sound wave parameter set to train a first sleep-aid analysis branch, and combining the first sleep scene to construct a first sleep-aid analysis unit; A sleep-aid analysis large model is constructed based on mapping of multiple sleep scenes and multiple sleep-aid analysis units, and the output layer of the scene recognition large model is connected to the input layer of the sleep-aid analysis large model to generate the sleep-aid mode switcher; The first sleep aid analysis branch is obtained by training the first EEG data set, the first microcurrent parameter set and the first sound wave parameter set, including: Construct initial microcurrent analysis channel and initial sound wave analysis channel based on multiple sleep-aid analysis operators; Performing supervised training and verification on the initial microcurrent analysis channel using the first EEG data set and the first microcurrent parameter set to obtain a microcurrent analysis channel; Performing supervised training and verification on the initial sound wave analysis channel using the first EEG data set and the first sound wave parameter set to obtain a sound wave analysis channel; The input layer, the microcurrent analysis channel, the sound wave analysis channel and the output layer are integrated to obtain the first sleep-aiding analysis branch.

2. The method for adaptively switching sleep-aiding modes based on EEG signal analysis according to claim 1, characterized in that: Identify multiple signal frequency bands, including: Determine a signal frequency fluctuation range based on the multiple EEG data sets, divide the signal frequency fluctuation range according to a predetermined frequency step, and determine multiple frequency intervals; Performing scene identification on the multiple EEG data sets according to the multiple sleep scenes, distributing the multiple identified data sets in the multiple frequency intervals, and marking the multiple frequency intervals with the sleep scene with the largest data proportion, to determine multiple marked frequency intervals; Adjacent fusion is performed on the multiple marked frequency intervals according to the same sleep scene to obtain the multiple signal frequency bands.

3. The method for adaptively switching sleep-aiding modes based on EEG signal analysis according to claim 1, characterized in that: Get microcurrent analysis channels, including: Configure multiple sleep-aid analysis operators, wherein the multiple sleep-aid analysis operators at least include BP neural network, support vector machine and random decision forest; The first microcurrent analysis model, the second microcurrent analysis model and the third microcurrent analysis model were constructed based on BP neural network, support vector machine and random decision forest respectively; A first training set is formed according to the first EEG data set and the first microcurrent parameter set, and the first training set is equally divided into three parts to obtain three training sets; Using the three training sets to perform supervised training and cross-validation training on the first microcurrent analysis model, the second microcurrent analysis model and the third microcurrent analysis model, respectively, to obtain a first convergence analysis model, a second convergence analysis model, a third convergence analysis model that meet the convergence constraints, and corresponding first analysis accuracy, second analysis accuracy and third analysis accuracy; The weight ratio is configured according to the first analysis accuracy, the second analysis accuracy and the third analysis accuracy, and the first convergence analysis model, the second convergence analysis model and the third convergence analysis model are integrated based on the weight ratio to construct the microcurrent analysis channel.

4. The method for adaptively switching sleep-aiding modes based on EEG signal analysis according to claim 1, characterized in that: Build a sleep mode switcher, and then include: Monitor the sleep characteristics of the target user, divide the sleep scenes and evaluate the sleep quality according to the sleep characteristics, and determine the actual sleep scenes and sleep quality deviations; Performing a sleep aid mode compensation analysis according to the sleep quality deviation, and determining microcurrent compensation parameters and sound wave compensation parameters; Time-aligning the target user's EEG signal with the actual sleep scene, the microcurrent compensation parameter, and the sound wave compensation parameter to obtain first compensation data, second compensation data, and third compensation data; The first compensation data is used to perform incremental learning on the scene recognition large model of the sleep-aid mode switch, and the second compensation data and the third compensation data are used to perform incremental learning on the sleep-aid analysis large model of the sleep-aid mode switch.

5. The method for adaptively switching sleep-aiding modes based on EEG signal analysis according to claim 1, characterized in that: In the background processing unit, a signal deviation between the real-time EEG signal and the EEG signal of the previous adjacent monitoring node is identified, and if the signal deviation meets a predetermined deviation threshold, the real-time EEG signal is input into the sleep-aiding mode switcher.

6. A sleep-aiding mode adaptive switching device based on EEG signal analysis, characterized in that: The method for implementing the sleep-aiding mode adaptive switching method based on EEG signal analysis according to any one of claims 1 to 5 comprises an EEG signal acquisition unit, a client, a background processing unit and a sleep-aiding control unit, and further comprises: An information receiving module is used to receive the target user's vital sign information, sleep characteristics and sleep environment characteristics through the client, and transmit the information to the background processing unit through communication; An associated retrieval module is used to perform similar associated retrieval on a historical user database in a background processing unit based on the physical sign information, sleep characteristics and sleep environment characteristics as constraints to obtain an associated user data set; A model building module, used to cluster the associated user data set according to the predetermined sleep scene to obtain multiple user feature sets, use the multiple user feature sets to train and obtain a scene recognition model and a sleep aid analysis model, and build a sleep aid mode switcher; A sleep-aiding mode analysis module is used to obtain the real-time EEG signal of the target user at a fixed point through an EEG signal acquisition unit, transmit the real-time EEG signal to a background processing unit, and use the sleep-aiding mode switcher to analyze and obtain an adapted sleep-aiding mode, wherein the adapted sleep-aiding mode includes microcurrent parameters and sound wave parameters; The sleep aid mode switching module is used to communicate the microcurrent parameters and the sound wave parameters to the sleep aid control unit to complete the sleep aid mode switching.

Images