Intelligent sleep state monitoring method and device based on brain waves

Abstract

The invention relates to the technical field of sleep state monitoring, and particularly discloses an intelligent sleep state monitoring method and device based on brain waves, brain wave signals of a user are collected through EEG equipment, and a signal processing and analysis algorithm based on artificial intelligence and deep learning is introduced at the rear end to analyze the brain wave signals, so that the sleep state monitoring accuracy is improved. And the sleep state of the user is monitored based on full-frequency aggregation of the brain wave signals. In this way, the sleep state of the user can be automatically identified and displayed in real time based on the brain waves of the user, and improvement suggestions are provided when the sleep state is identified to be poor. For example, a brain balance regulator is used for sleep regulation, and a non-invasive and non-invasive weak stimulation technology is adopted for regulation. According to the method, the prior data table between the sleep state and the current mode is constructed by using an experiment, so that after the sleep state is obtained, the current mode can be adaptively adjusted based on prior knowledge to adjust cerebral cortex neural activity, and the sleep improvement of a user can be better assisted.

Description

Technical Field

[0001] The present application relates to the technical field of sleep state monitoring, and more specifically, to an intelligent sleep state monitoring method and device based on brain waves. Background Art

[0002] Sleep is an important part of human physiological activities and is essential for maintaining physical health, emotional stability and cognitive function. Good sleep quality helps improve the quality of life, while sleep disorders may lead to a variety of health problems, such as cardiovascular disease, diabetes, obesity, and mental health problems. Therefore, accurately monitoring an individual's sleep state and making targeted adjustments based on different sleep states to assist sleep are of great significance for promoting health management and disease prevention.

[0003] However, traditional sleep monitoring methods mainly rely on some relatively simple monitoring methods. For example, sleep time and quality are estimated by detecting wrist movement; pressure sensors placed under the mattress are used to monitor body movements; or the built-in accelerometer and microphone of the mobile phone are used to collect data to analyze sleep cycles. These solutions mainly rely on externally monitored behavioral indicators rather than directly measuring brain activity, so they cannot accurately distinguish different sleep stages. At the same time, they are also easily affected by changes in the external environment, such as a partner turning over or other noises, which reduces the consistency and reliability of sleep status results, and thus affects the decision-making of sleep adjustment and improvement plans.

[0004] Therefore, an optimized sleep state monitoring solution is desired. Summary of the invention

[0005] The present application provides an intelligent sleep status monitoring method and device based on brain waves, which can not only monitor the user's sleep status in real time, but also automatically adjust the current mode of non-invasive weak electrical stimulation of specific brain areas according to the detection results and prior knowledge to optimize sleep quality.

[0006] In a first aspect, a method for intelligent sleep state monitoring based on brain waves is provided, comprising: Collect brain wave signals of the target object through EEG equipment; Performing signal preprocessing on the brain wave signal of the target object and decomposing the signal into a set of brain wave signal components; Performing signal component feature extraction on each brain wave signal component in the set of brain wave signal components to obtain a set of brain wave signal component waveform coding features; The set of brain wave signal component waveform coding features is subjected to feature distinguishable enhancement processing to obtain a set of distinguishable enhanced brain wave signal component coding features, including: performing field mapping on each brain wave signal component waveform coding feature in the set of brain wave signal component waveform coding features to obtain a set of entry brain wave signal component waveform coding features; based on the essential features of the set of entry brain wave signal component waveform coding features as conditional features, the set of brain wave signal component waveform coding features is subjected to feature enhancement processing to obtain the set of distinguishable enhanced brain wave signal component coding features; Sleep state identification is performed based on the set of distinguishable enhanced brain wave signal component encoding features to determine a sleep state type label.

[0007] In a second aspect, an intelligent sleep state monitoring device based on brain waves is provided, comprising: The brain wave signal acquisition module is used to collect the brain wave signal of the target object through the EEG device; A brain wave signal processing module, used for performing signal preprocessing on the brain wave signal of the target object and decomposing it into a set of brain wave signal components; An electroencephalogram signal component feature extraction module is used to extract signal component features of each electroencephalogram signal component in the set of electroencephalogram signal components to obtain a set of waveform coding features of the electroencephalogram signal components; The brain wave signal feature distinguishable enhancement processing module is used to perform feature distinguishable enhancement processing on the set of brain wave signal component waveform coding features to obtain a set of distinguishable enhanced brain wave signal component coding features, wherein the brain wave signal feature distinguishable enhancement processing module includes: a field mapping processing unit, used to perform field mapping on each brain wave signal component waveform coding feature in the set of brain wave signal component waveform coding features to obtain a set of entry brain wave signal component waveform coding features; a feature enhancement processing unit, used to perform feature enhancement processing on the set of brain wave signal component waveform coding features based on the essential features of the set of entry brain wave signal component waveform coding features as conditional features to obtain the set of distinguishable enhanced brain wave signal component coding features; The sleep state identification module is used to identify the sleep state based on the set of encoding features of the distinguishable enhanced brain wave signal components to determine the sleep state type label.

[0008] The present application provides a method and device for intelligent sleep state monitoring based on brain waves, which collects the user's brain wave signals through EEG equipment, and introduces a signal processing and analysis algorithm based on artificial intelligence and deep learning in the back end to analyze the brain wave signals, so as to capture the full-frequency significant aggregation of the user's brain waves, and monitor the user's sleep state based on the full-frequency aggregation of this brain wave signal. In this way, the user's sleep state can be automatically identified and displayed in real time based on the user's brain waves, and improvement suggestions can be provided when the sleep state is not good. For example, a brain balance controller is used for sleep regulation, which uses non-invasive weak electrical stimulation technology of specific brain areas for regulation. In other words, experiments are used to construct a priori data table between sleep state and current mode, so that after obtaining the sleep state, the current mode can be adaptively adjusted based on prior knowledge to regulate the cerebral cortex neural activity, which helps to better assist the user in improving sleep. BRIEF DESCRIPTION OF THE DRAWINGS

[0009] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings of the embodiments of the present application are briefly introduced below. Obviously, the drawings described below only relate to some embodiments of the present application, and are not intended to limit the present application.

[0010] Figure 1 This is a schematic flowchart of an intelligent sleep state monitoring method based on brain waves according to an embodiment of the present application.

[0011] Figure 2 This is a data flow diagram of the brain wave-based intelligent sleep state monitoring method according to an embodiment of the present application.

[0012] Figure 3 This is a schematic flowchart of step S2 in the brain wave-based intelligent sleep state monitoring method of an embodiment of the present application.

[0013] Figure 4 This is a schematic flowchart of step S4 in the brain wave-based intelligent sleep state monitoring method of an embodiment of the present application.

[0014] Figure 5 This is a schematic flowchart of step S42 in the brain wave-based intelligent sleep state monitoring method of an embodiment of the present application.

[0015] Figure 6 This is a schematic block diagram of an intelligent sleep state monitoring device based on brain waves according to an embodiment of the present application. DETAILED DESCRIPTION

[0016] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments of the present application, all other embodiments obtained by ordinary technicians in this field without creative work also fall within the scope of protection of the present application.

[0017] With the development of technology, EEG (Electroencephalogram) monitoring technology plays an important role in sleep status monitoring due to its non-invasiveness and real-time nature, and is widely used in clinical and personal health management. By analyzing the brain's specific frequency of brain wave activity generated in different sleep stages, a person's current sleep state can be accurately judged, and personalized intervention measures can be provided to improve sleep quality. Therefore, the new EEG-based monitoring technology is gradually changing the understanding of sleep status and the management of sleep patterns with its unique advantages.

[0018] In order to further improve the recognition accuracy of sleep state and realize personalized sleep regulation scheme, this scheme proposes a new method of transcranial non-invasive brain stimulation (NiBS) based on brain balance targeted regulation technology. NiBS is a non-invasive transcranial weak brain stimulation technology that regulates the excitability or inhibition of nerve cells by applying weak currents to specific areas of the brain, thereby affecting cognitive function, emotional state, and promoting the recovery process. For patients with sleep disorders, appropriate adjustment of the excitation level of the cerebral cortex can help relieve insomnia symptoms and improve sleep efficiency.

[0019] Based on this, in the technical solution of the present application, a method for intelligent sleep state monitoring based on brain waves is proposed, which can not only monitor the user's sleep condition in real time, but also automatically adjust the current mode of non-invasive weak electrical stimulation of specific brain regions according to the detection results and prior knowledge to optimize sleep quality. Specifically, the technical concept of the present application is to collect the user's brain wave signal through an EEG device, and introduce a signal processing and analysis algorithm based on artificial intelligence and deep learning in the back end to analyze the brain wave signal, so as to capture the user's full-frequency significant aggregation of the brain wave, and monitor the user's sleep state based on the full-frequency aggregation of this brain wave signal. In this way, the user's sleep state can be automatically identified and displayed in real time based on the user's brain waves, and improvement suggestions can be provided when the sleep state is not good. For example, a brain balance controller is used for sleep regulation, which uses non-invasive weak electrical stimulation technology of specific brain regions for regulation. In other words, an experiment is used to construct a priori data table between sleep state and current mode, so that after obtaining the sleep state, the current mode can be adaptively adjusted based on prior knowledge to regulate the cerebral cortex neural activity, thereby helping to better assist the user in improving sleep.

[0020] Specifically, in the technical solution of the present application, if Figure 1 and Figure 2 As shown, the brain wave-based intelligent sleep state monitoring method includes: S1, collecting brain wave signals of the target object through an EEG device; S2, performing signal preprocessing on the brain wave signals of the target object and decomposing them into a set of brain wave signal components; S3, performing signal component feature extraction on each brain wave signal component in the set of brain wave signal components to obtain a set of brain wave signal component waveform coding features; S4, performing feature distinguishable enhancement processing on the set of brain wave signal component waveform coding features to obtain a set of distinguishable enhanced brain wave signal component coding features; S5, performing sleep state identification based on the set of distinguishable enhanced brain wave signal component coding features to determine a sleep state type label.

[0021] Exemplarily, in step S1, the EEG device is used to collect the brain wave signal of the target object. It should be understood that the activity of the human brain in different sleep stages and wakefulness is mainly reflected in a specific frequency range. For example: delta waves (0.5-4Hz) mainly appear during deep sleep; theta waves (4-8Hz) are common in the light sleep stage; alpha waves (8-12Hz) are associated with a relaxed but awake state; beta waves (12-30Hz) are usually associated with active thinking and concentration. These frequency bands are essential for analyzing sleep states and brain functions. Therefore, EEG signals can provide detailed information about an individual's current sleep state, which is essential for accurately assessing sleep quality and structure. In addition, traditional sleep monitoring methods mainly rely on external behavioral indicators (such as wrist activity, body movement, etc.), which cannot accurately distinguish different sleep stages and are easily affected by changes in the external environment. In contrast, EEG monitoring technology plays an important role in sleep state monitoring due to its non-invasiveness and real-time nature. It can more accurately judge a person's current sleep state and provide personalized intervention measures to improve sleep quality accordingly.

[0022] In one embodiment, the EEG device is used to collect the brain wave signals of the target subject, including: first, the EEG device needs to be prepared, which usually includes one or more electrodes, which are placed at specific locations on the scalp to record the electrical signals generated by the brain. Modern EEG equipment may also include wireless transmission capabilities, so that data can be sent directly to a computer or other analysis device for processing. The electrodes are positioned and placed according to the international 10-20 system or a modified version of the standard. This standard defines a series of standardized positions to ensure comparability between different studies. Electrodes are usually wet (using conductive paste to increase contact), dry (no conductive paste required) or gel type. Once the electrodes are correctly placed, the EEG device begins to continuously collect weak electrical signals from the brain. In order to ensure the quality of the data, a calibration period may be performed first to allow the subject to be in a resting state in order to determine the baseline level.

[0023] Exemplarily, in step S2, the EEG signal of the target object is preprocessed and then decomposed into a set of EEG signal components. It should be understood that the original EEG signal usually contains noise and artifacts from various sources, such as muscle activity, eye movement, electrocardiogram (ECG) interference, and environmental electromagnetic interference. These signal components from non-brain sources will confuse the real brain activity information and reduce the accuracy of subsequent analysis. Therefore, high-frequency noise and other types of interference can be effectively reduced or eliminated through the preprocessing step to improve the signal-to-noise ratio. EEG signals are essentially nonlinear and non-stationary, containing multiple frequency components and time-varying patterns. It is very difficult to directly analyze such a composite signal. By decomposing the signal, the present application disassembles it into a series of relatively simple signal components, each of which represents the activity characteristics within different frequency ranges. This method makes subsequent feature extraction and analysis easier, and can better capture the unique EEG changes in different sleep stages.

[0024] In one embodiment, Figure 3 As shown, the brain wave signal of the target object is decomposed into a set of brain wave signal components after signal preprocessing, including: S21, signal preprocessing the brain wave signal of the target object to filter out high-frequency noise in the brain wave signal to obtain a denoised brain wave signal; S22, signal decomposition based on CEEMDAN on the denoised brain wave signal to obtain a set of brain wave signal components.

[0025] Exemplarily, in step S21, the brain wave signal of the target object is subjected to signal preprocessing to filter out high-frequency noise in the brain wave signal to obtain a brain wave signal after noise reduction. It should be understood that high-frequency noise such as external electromagnetic interference, power line frequency (such as 50Hz or 60Hz) and its harmonics may be mixed into the EEG signal, and these noises are usually higher than the above-mentioned useful brain wave frequency range. In addition, electromyographic signals may also generate high-frequency noise, especially when the facial and neck muscles are active, these signals are usually above 30Hz, and sometimes even higher. Therefore, in order to be able to remove high-frequency noise, so as to significantly improve the signal-to-noise ratio, make subsequent signal analysis more accurate and reliable, and help to better identify different sleep stages and states, in the technical solution of the present application, the brain wave signal of the target object is subjected to signal preprocessing to filter out high-frequency noise in the brain wave signal to obtain a brain wave signal after noise reduction.

[0026] In one embodiment, the brain wave signal of the target object is subjected to signal preprocessing to filter out high-frequency noise in the brain wave signal to obtain the noise-reduced brain wave signal, including: using a low-pass filter to filter the brain wave signal of the target object to filter out high-frequency noise in the brain wave signal to obtain the noise-reduced brain wave signal.

[0027] Exemplarily, in step S22, the denoised brain wave signal is subjected to signal decomposition based on CEEMDAN to obtain a set of brain wave signal components. It should be understood that, considering that the brain wave signal is highly nonlinear and non-stationary, it contains multiple frequency components and time-varying patterns. Therefore, in order to decompose the complex brain wave signal into a series of simpler and easier to analyze signal components, so as to capture the characteristics of brain wave signals in different frequency ranges, in the technical solution of the present application, the denoised brain wave signal is further subjected to signal decomposition based on CEEMDAN to obtain a set of brain wave signal components. It is worth mentioning that CEEMDAN is an adaptive signal decomposition method, which does not require a fixed basis function to be set in advance, but automatically decomposes according to the characteristics of the signal itself. This makes it particularly suitable for processing nonlinear and non-stationary data.

[0028] Exemplarily, in step S3, signal component feature extraction is performed on each brainwave signal component in the set of brainwave signal components to obtain a set of brainwave signal component waveform coding features. It should be understood that in order to reveal the different frequency components and time-varying features hidden in the original brainwave signal, in one embodiment, signal component feature extraction is performed on each brainwave signal component in the set of brainwave signal components to obtain a set of brainwave signal component waveform coding features, including: inputting each brainwave signal component in the set of brainwave signal components into a signal component feature extractor based on the RNN model to obtain a set of brainwave signal component waveform coding feature vectors as the set of brainwave signal component waveform coding features. Through the processing of the signal component feature extractor based on the RNN model, the waveform feature information of each brainwave signal component can be captured separately, which helps to better understand and encode the contextual information and dynamics of each signal component.

[0029] Exemplarily, in step S4, the set of the brain wave signal component waveform coding features is subjected to feature distinguishable enhancement processing to obtain a set of distinguishable enhanced brain wave signal component coding features. It should be understood that since each vector in the set of the brain wave signal component waveform coding feature vectors contains different brain wave signal component waveform frequency domain feature information, these signal component frequency domain features can reflect the different potential physiological significance and sleep state of the user. However, in the set of the brain wave signal component waveform coding feature vectors, not every brain wave signal component waveform coding feature vector has the same importance for monitoring and type identification of the user's sleep state. In order to capture the significant key signal component features in the set and optimize the overall signal component feature expression with this feature, in the technical solution of the present application, the set of the brain wave signal component waveform coding features is further subjected to feature distinguishable enhancement processing to obtain a set of distinguishable enhanced brain wave signal component coding features.

[0030] In one embodiment, Figure 4 As shown, the set of brain wave signal component waveform coding features is subjected to feature distinguishable enhancement processing to obtain a set of distinguishable enhanced brain wave signal component coding features, including: S41, field mapping of each brain wave signal component waveform coding feature in the set of brain wave signal component waveform coding features to obtain a set of entry brain wave signal component waveform coding features; S42, based on the essential characteristics of the set of entry brain wave signal component waveform coding features as conditional characteristics, feature enhancement processing is performed on the set of brain wave signal component waveform coding features to obtain the set of distinguishable enhanced brain wave signal component coding features.

[0031] In one embodiment, in step S41, field mapping is performed on each brain wave signal component waveform coding feature in the set of brain wave signal component waveform coding features to obtain a set of entry brain wave signal component waveform coding features, including: using a field mapper to field map each brain wave signal component waveform coding feature vector in the set of brain wave signal component waveform coding feature vectors to obtain a set of entry brain wave signal component waveform coding feature vectors as the set of entry brain wave signal component waveform coding features. Specifically, the process can be expressed by a formula:

[0032] in, is the first one in the set of waveform encoding feature vectors of the brain wave signal components EEG signal component waveform encoding feature vector, and are the first weight matrix and the second weight matrix respectively, yes The corresponding incoming EEG signal component waveform encoding feature vector.

[0033] That is, through field mapping, the position and direction of the waveform encoding feature vector of each brain wave signal component can be adjusted so that they better reflect the physiological significance they represent in the new space. For example, some features may be more closely associated with the light sleep period, while other features appear more in the deep sleep stage. After mapping, the differences between these features become more obvious, thereby improving the effect of subsequent classification or clustering analysis. Field mapping is not just a simple coordinate transformation, it can also reveal complex relationships that are not easily perceived in the original feature space. By properly mapping the feature vectors, originally hidden related patterns can be found in the new space, which helps to deeply understand the connection between brain wave signals and specific sleep states. For example, some seemingly unrelated frequency components may show synergy in the mapped field, providing a new perspective for explaining the sleep mechanism. When the present application uses the waveform encoding feature vectors of the mapped entry brain wave signal components for sleep state recognition, since these features have been optimized, the performance of the classifier can be significantly improved. Compared with directly using the unmapped original features, the feature set that has been mapped by the field usually has higher separability and lower risk of dimensional disaster, which means that the classifier can more accurately divide different sleep stages.

[0034] In one embodiment, Figure 5 As shown, in step S42, based on the essential characteristics of the set of waveform coding features of the entry brain wave signal component as conditional characteristics, the set of waveform coding features of the brain wave signal component is subjected to feature enhancement processing to obtain the set of distinguishable and enhanced brain wave signal component coding features, including: S421, calculating the essential characteristics of the set of waveform coding feature vectors of the entry brain wave signal component to obtain the essential feature vector of the brain wave signal component waveform field; S422, taking the essential feature vector of the brain wave signal component waveform field as the conditional feature vector, the set of waveform coding feature vectors of the brain wave signal component is subjected to feature enhancement processing to obtain the set of distinguishable and enhanced brain wave signal component coding feature vectors as the set of distinguishable and enhanced brain wave signal component coding features.

[0035] In one embodiment, in step S421, the essential characteristics of the set of the incoming brain wave signal component waveform coding feature vectors are calculated to obtain the brain wave signal component waveform field essential characteristic vectors, including: calculating the field depth factor of each incoming brain wave signal component waveform coding feature vector in the set of the incoming brain wave signal component waveform coding feature vectors to obtain the set of brain wave signal component waveform field depth factors; based on the set of brain wave signal component waveform field depth factors, calculating the essential characteristics of the set of the incoming brain wave signal component waveform coding feature vectors to obtain the brain wave signal component waveform field essential characteristic vectors. Specifically, the process can be expressed as follows:

[0036]

[0037]

[0038] in, yes The corresponding incoming EEG signal component waveform encoding feature vector, To calculate the square of the vector-norm, represents the logarithmic function value with base 2, For calculation The depth of field factor, for The corresponding EEG signal component waveform field depth factor, It means to calculate the value of the natural exponential function with the natural constant e as the base. for The corresponding standardized EEG signal component waveform field depth factor, It is the essential characteristic vector of the waveform field of the brain wave signal component.

[0039] That is, the field depth factor is an indicator used to measure the importance of the position of each incoming brain wave signal component waveform encoding feature vector in its field. Simply put, it reflects the "depth" or "significance" of the information carried by the feature vector in the new space (i.e., the space after field mapping). By calculating the field depth factor, the present application can determine which features are more critical to describing a specific sleep state. For example, in the light sleep stage, some frequency components may show a higher level of activity; while during deep sleep, other frequency components are more active. Through the field depth factor, these differences can be quantified, allowing the system to more accurately capture the unique signs of different sleep stages. Once the field depth factor set of all incoming brain wave signal component waveform encoding feature vectors is obtained, the next step is to calculate the essential features of these feature vectors to form the essential feature vector of the brain wave signal component waveform field. The core of this step is to integrate the information of each feature vector and extract the core content they express together, that is, the key features that best represent different sleep states. In this way, not only the distinction between features is enhanced, but also the impact of redundant information is reduced. This means that in subsequent classification tasks, the classifier can focus more on those truly meaningful features, thereby improving the accuracy of classification.

[0040] In one embodiment, in step S422, the set of the brain wave signal component waveform coding feature vectors is subjected to feature enhancement processing using the intrinsic feature vector of the brain wave signal component waveform field as a conditional feature vector to obtain a set of distinguishable enhanced brain wave signal component coding feature vectors as the set of distinguishable enhanced brain wave signal component coding features, including: using the intrinsic feature vector of the brain wave signal component waveform field as a conditional feature vector, and distinguishably enhancing each brain wave signal component waveform coding feature vector in the set of the brain wave signal component waveform coding feature vectors to obtain the set of distinguishable enhanced brain wave signal component coding feature vectors. Specifically, the process can be expressed as:

[0041] in, is the essential characteristic vector of the waveform field of the brain wave signal component, is the intrinsic weight matrix of the EEG signal component waveform, is the waveform weight matrix of the local EEG signal component, is the bias vector, yes function, It is the point multiplication by position. yes The corresponding distinguishable enhanced EEG signal component encoding feature vector.

[0042] That is, by introducing the intrinsic feature vector of the waveform field of the EEG signal component as a conditional feature vector, a global guide can be provided for each original encoding feature vector. This guidance is to emphasize the key features that best represent different sleep stages, so that each feature vector can better reflect its corresponding physiological significance. For example, in the light sleep period (NREM N1), some frequency components may be particularly prominent; while in the deep sleep period (SWS), other frequency components are more significant. By strengthening these features, they can be more easily distinguished in subsequent classification tasks. In practical applications, due to individual differences, environmental factors or other interferences, the original EEG signal may contain certain noise or variation. Through distinguishable strengthening processing based on the intrinsic feature vector of the field, these uncertainties can be eliminated to a certain extent, making the feature vector more stable and reliable. This means that even in the face of complex real-life situations, the system can maintain a high recognition accuracy and will not make misjudgments due to minor changes.

[0043] In summary, the feature distinguishable enhancement processing can optimize the feature representation in the set of the waveform encoding features of the brain wave signal component through a multi-level abstraction and enhancement process. It realizes the refined processing of the set of the input brain wave signal component waveform encoding feature vectors and the dynamic reconstruction of the feature space through the steps of field mapping, calculating the field depth factor, refining the field essential feature vector and distinguishable enhancement. Through such a multi-stage processing mechanism, not only can the deep information structure be effectively mined from the original brain wave signal component waveform, but also the quality of the feature representation of each brain wave signal component can be significantly improved, making it more in line with the needs of actual application scenarios. For example, in sleep state monitoring, this enhanced feature vector can more accurately reflect the characteristics of different sleep stages, thereby improving the accuracy of sleep state recognition. In addition, the processing process of this feature distinguishable enhancement processing has strong flexibility and adaptability, and can flexibly adjust the parameter configuration of each link according to the specific requirements of the sleep state monitoring and recognition tasks of different users, thereby achieving the best performance and providing strong support for the sleep state monitoring task.

[0044] Exemplarily, in step S5, sleep state identification is performed based on the set of distinguishable and enhanced brain wave signal component coding features to determine the sleep state type label. It should be understood that the set of distinguishable and enhanced brain wave signal component coding feature vectors is feature spliced ​​into a brain wave full-frequency domain significant aggregation coding feature vector, so as to integrate the feature information of brain wave signal components in different frequency domains. The full-frequency domain significant aggregation coding feature vector formed in this way can more comprehensively reflect the overall state of brain activity and enhance the expression ability of the feature. Furthermore, a classifier-based sleep state identifier is used to classify the spliced ​​brain wave full-frequency domain significant aggregation coding feature vector, so as to accurately identify the user's current sleep state and obtain an identification result for representing the sleep state type label.

[0045] In one embodiment, sleep state recognition is performed based on the set of distinguishable and enhanced brain wave signal component coding features to determine the sleep state type label, including: feature splicing the set of distinguishable and enhanced brain wave signal component coding feature vectors into a brain wave full-frequency domain significant aggregation coding feature vector, and then inputting the brain wave full-frequency domain significant aggregation coding feature vector into a classifier-based sleep state identifier to obtain a recognition result, and the recognition result is used to represent the sleep state type label.

[0046] Specifically, the above-constructed EEG full-frequency domain significant aggregation encoding feature vector is input into a pre-trained classifier. The task of the classifier is to analyze the input feature vector and finally output a label representing the current sleep state type. The classifier is fully trained with a large amount of labeled sample data to ensure that the model has good generalization ability and accuracy. The classifier will establish decision boundaries between different sleep states based on the knowledge learned during the training process. When a new feature vector is input, the classifier will determine which known category it is closest to based on these boundaries, and assign the corresponding sleep state label accordingly. Finally, the result of the classifier output is a specific sleep state type label, such as light sleep (NREM N1), deep sleep (SWS), rapid eye movement (REM), etc. These labels directly reflect the user's current sleep stage, providing an important reference for subsequent sleep quality assessment and intervention measures. Specifically, the normal human sleep cycle is usually divided into rapid eye movement (REM) and non-rapid eye movement (NREM), the latter of which is further divided into three stages: N1, N2 and deep slow wave sleep (SWS). Each stage has its own unique EEG characteristics, such as delta waves (0.5-4Hz) mainly appear during SWS; theta waves (4-8Hz) are common in the light sleep stage; alpha waves (8-12Hz) are associated with the state of wakefulness and relaxation. Therefore, by accurately identifying these pattern changes, a person's sleep structure and quality can be effectively assessed.

[0047] In a specific embodiment of the present application, the classifier-based sleep state identifier includes a fully connected layer and a Softmax function. First, the fully connected layer receives the input feature vector from the previous layer (such as a feature extraction layer or a hidden layer) and performs a linear transformation on these features. This process maps the input features to a new space by adjusting the weights and bias terms, and each dimension corresponds to a potential sleep state category. During the training process, the fully connected layer learns which features are most critical for a specific sleep stage, thereby optimizing the feature representation. Next, the Softmax function converts the score of each category into a probability distribution. Softmax ensures that all output values ​​are positive and add up to 1, so they can be directly interpreted as probabilities. This means that after Softmax processing, we get a probability distribution vector in which each element represents the probability of belonging to a certain category. Finally, in order to determine the final recognition result, the system selects the label with the highest probability. In other words, whichever category has the highest probability is considered to belong to that category. For example, in the sleep state classifier, if the probability distribution obtained after processing indicates that the probability of the deep slow-wave sleep stage is the highest, then the system will use this stage as the final recognition result.

[0048] In particular, considering that each brain wave signal component waveform encoding feature vector in the set of brain wave signal component waveform encoding feature vectors respectively represents the frequency semantic features of the brain wave signal component, when performing feature distinguishability enhancement based on sequence condition description, taking into account the sequence condition descriptive differences caused by the signal semantic differences between each brain wave signal component, the brain wave full-frequency domain significant aggregation encoding feature vector obtained by feature splicing has the global richness of global feature splicing of sequence condition descriptive differences, resulting in repetition of unstructured probability mapping, affecting the accuracy of the classification results.

[0049] Preferably, inputting the EEG full-frequency domain significant aggregation encoding feature vector into a classifier-based sleep state identifier to obtain the identification result comprises: Determine the number of non-zero eigenvalues ​​in the EEG full-frequency domain significant aggregation coding feature vector, and subtract the number of non-zero eigenvalues ​​from the length of the EEG full-frequency domain significant aggregation coding feature vector to obtain the zero-dimensional value of the EEG full-frequency domain significant aggregation coding ; Calculate the zero-differentiable value of the brain wave full frequency domain significant aggregation coding after subtracting one from the zero-dimensional value of the brain wave full frequency domain significant aggregation coding The first brain wave full frequency domain significant aggregation coding field fitting value obtained by multiplying and dividing by the zero-dimensional value of the brain wave full frequency domain significant aggregation coding and the second EEG full frequency domain significant aggregation encoding field fitting value ; Calculate the square root of the sum of squares of all eigenvalues ​​of the EEG full-frequency domain significant aggregation coding feature vector to obtain the EEG full-frequency domain significant aggregation coding mode representation value ,in, Represents each eigenvalue of the EEG full-frequency domain significant aggregation encoding feature vector, Indicates the representation value of the significant aggregate coding pattern in the full frequency domain of EEG; The zero differentiable value is encoded by the significant aggregation of the full frequency domain of the EEG Each eigenvalue of the EEG full-frequency domain significant aggregate coding feature vector is calculated as an index The power function , and multiply it by the first EEG full-frequency domain significant aggregate coding field fitting value To obtain the local representation value of the significant aggregation encoding of the whole frequency domain of the EEG ; Calculate each eigenvalue of the EEG full-frequency domain significant aggregation encoding eigenvector The second EEG full-frequency domain significant aggregation coding field fitting value And the EEG full frequency domain significant aggregation coding pattern representation value The product of is used to obtain the center-directed value of the significant aggregation coding of the whole frequency domain of the EEG ; The brain wave full frequency domain significant aggregation encoding local representation value The EEG full frequency domain significantly aggregates the encoding center-directed value The weighted difference The composition is optimized EEG full-frequency domain significant aggregation encoding feature vector, where and represents the weighted hyperparameter; The optimized brain wave full-frequency domain significant aggregation encoding feature vector is passed through a classifier-based sleep state identifier to obtain a recognition result.

[0050] Accordingly, based on the vector field fitting law of high-dimensional manifold, the isolated zeros in the EEG full-frequency domain significant aggregation coding feature vector are used as the differentiable dimension of the high-dimensional manifold, so as to fix the normal direction pointing to the distribution center near the local coordinates represented by the eigenvalues ​​of the EEG full-frequency domain significant aggregation coding feature vector, thereby avoiding invalid repetition in the probability mapping process caused by the unstructured characteristics of the semantically rich distribution of the EEG full-frequency domain significant aggregation coding feature vector by effectively aligning the characteristic distribution pattern of the EEG full-frequency domain significant aggregation coding feature vector to the probability density differentiable field, thereby improving the accuracy of the recognition results obtained by the classifier-based sleep state identifier.

[0051] Furthermore, experiments are used to construct a priori data table between sleep state and current mode, so that after obtaining the sleep state, the current mode can be adaptively adjusted to better assist sleep. Specifically, a priori data table can be constructed by collecting the corresponding brain wave characteristics of a large number of users in different sleep states and the corresponding non-invasive weak current modes of specific brain regions through experiments. This data table records the effective current modes in different sleep states, providing a basis for subsequent adaptive adjustments. In addition, since the brain activity patterns of each user may be different, the data collected through experiments can help the system learn and adapt to the specific needs of different users, thereby providing more personalized assistance solutions. In this way, the user's sleep state can be automatically identified and monitored in real time based on his or her brain waves, and improvement suggestions can be provided when the sleep state is not good. For example, a brain balance controller is used for sleep regulation, which can automatically adjust the non-invasive weak current mode of a specific brain region to regulate the cerebral cortical neural activity based on the test results and prior knowledge, thereby helping to better assist users in improving sleep.

[0052] Furthermore, based on the theory of targeted regulation of brain balance, the present application also provides a brain balance regulator, which adopts non-invasive weak electrical stimulation technology for specific brain areas and is a portable non-invasive regulation device that selects regulation parameters through simple operation. The brain balance regulator has 2-8 channels, and uses electrodes to apply specific weak currents to specific targeted brain areas to regulate the neural activity of the cerebral cortex. It effectively meets the regulatory intervention of different intensities and times for different brain areas. The brain balance regulator mainly includes the following contents: (1) Regulator (1 unit); (2) Sensor (1 set); (3) Positioning cap (1 unit); (4) Accessories (including cotton swabs, saline, head circumference tester, positioning patch, electrode box, etc.).

[0053] The following preparations are required before using the brain balance controller: (1) Ensure that the device is fully charged. If the device cannot be started due to low power, charge or replace it as soon as possible. (2) The controller uses saline electrodes to reduce the contact impedance of the human body to achieve the best conditioning effect. You need to prepare 0.9% sodium chloride solution (normal saline) or saturated saline.

[0054] The steps for using the brain balance regulator include: (1) soaking the sponge inside the electrode sheet with saline and inserting the sponge into the electrode sheet; (2) inserting the columnar electrode of the electrode wire into the jack of the electrode sheet, and inserting the other end of the electrode wire into the output port of the brain balance regulator; (3) putting on a matching hat (large, medium, or small size is available) for the user to ensure that it is worn correctly and to determine the specific brain area to be intervened; (4) adjusting the electrode sheet to reduce the contact impedance with the human body. To reduce the contact impedance, the concentration of saline in the sponge of the electrode sheet can be appropriately increased, the fixing force of the positioning cap can be appropriately increased, the degree of wetting of the sponge can be increased, and sufficient saline can be ensured between the electrode sheet and the head; (5) turning on the power switch and selecting the intensity and time as required; (6) after the conditioning is completed, unplug the electrode plug, remove the positioning cap, and turn off the power.

[0055] The precautions for the brain balance controller include: (1) When wetting the sponge, do not let the controller host be immersed in saline to avoid short circuit damage; (2) Wet the sponge sheet before inserting it into the electrode; (3) Do not use tap water to wet the sponge sheet, but be sure to use 0.9% sodium chloride solution or saturated saline; (4) If the sponge sheet has not been used for a long time, soak it in saline for about 5 minutes before using it again; (5) When not in use for a long time, please unplug the electrode wire; (6) Stop using it immediately if the user feels a burning sensation; (7) When using it, ensure that the surrounding skin is not damaged or ulcerated, and there are no metal objects such as jewelry on the head; (8) When using it, ensure that the electrode sheet is in full contact with the user, otherwise the contact area will be reduced, the current density will be too high, and burns may occur.

[0056] Contraindications for the use of the brain balance controller include: (1) patients with metal implants in the brain; (2) patients with increased intracranial pressure; (3) patients with fever, electrolyte imbalance or unstable vital signs; (4) patients with acute large-area cerebral infarction; (5) patients with epilepsy and those taking drugs that can cause epilepsy; (6) patients with implants with metal parts in the treatment area; (7) patients with hyperalgesia in the stimulation area; (8) patients with bleeding tendency; (9) patients using implanted electronic devices (such as pacemakers); (10) patients with severe heart disease or other medical diseases; and (11) pregnant women and children.

[0057] The technical advantages of the brain balance control device include: (1) easy operation; (2) effectiveness: confirmed by research and evidence-based medicine; (3) safety: only transient sensations such as slight tingling, itching, and dizziness; (4) a wide range of users: from children as young as 3 years old to those with severe conditions and those who wake up from coma; and (5) high acceptance.

[0058] In summary, the brain wave-based intelligent sleep state monitoring method according to the embodiment of the present application is explained, which collects the user's brain wave signal through an EEG device, and introduces a signal processing and analysis algorithm based on artificial intelligence and deep learning in the back end to analyze the brain wave signal, so as to capture the user's full-frequency significant aggregation of brain waves, and monitor the user's sleep state based on the full-frequency aggregation of this brain wave signal. In this way, the user's sleep state can be automatically identified and displayed in real time based on the user's brain waves, and improvement suggestions can be provided when the sleep state is not good. For example, a brain balance controller is used for sleep regulation, which uses non-invasive weak electrical stimulation technology of specific brain areas for regulation. In other words, experiments are used to construct a priori data table between sleep state and current mode, so that after obtaining the sleep state, the current mode can be adaptively adjusted based on prior knowledge to regulate the cerebral cortex neural activity, which helps to better assist the user in improving sleep.

[0059] Figure 6 Schematic block diagram of an intelligent sleep state monitoring device based on brain waves according to an embodiment of the present application. Figure 6 As shown, the brainwave-based intelligent sleep state monitoring device 100 includes: a brainwave signal acquisition module 110, which is used to collect brainwave signals of a target object through an EEG device; a brainwave signal processing module 120, which is used to perform signal preprocessing on the brainwave signals of the target object and decompose them into a set of brainwave signal components; a brainwave signal component feature extraction module 130, which is used to perform signal component feature extraction on each brainwave signal component in the set of brainwave signal components to obtain a set of brainwave signal component waveform coding features; a brainwave signal feature distinguishable enhancement processing module 140, which is used to perform feature distinguishable enhancement processing on the set of brainwave signal component waveform coding features to obtain a set of distinguishable enhanced brainwave signal component coding features; a sleep state recognition module 150, which is used to perform sleep state recognition based on the set of distinguishable enhanced brainwave signal component coding features to determine a sleep state type label.

[0060] In one embodiment, the EEG signal feature distinguishable enhancement processing module includes: a field mapping processing unit, which is used to perform field mapping on each EEG signal component waveform coding feature in the set of EEG signal component waveform coding features to obtain a set of entry EEG signal component waveform coding features; a feature enhancement processing unit, which is used to perform feature enhancement processing on the set of EEG signal component waveform coding features based on the essential features of the set of entry EEG signal component waveform coding features as conditional features to obtain the set of distinguishable enhanced EEG signal component coding features.

[0061] Here, those skilled in the art can understand that the specific operations of the various modules and units in the above-mentioned intelligent sleep state monitoring device based on brain waves have been described in the above reference. Figures 1 to 6 The description of the brain wave-based intelligent sleep state monitoring method has been introduced in detail, and therefore, its repeated description will be omitted.

[0062] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0063] It should be understood that the specific examples in this article are only intended to help those skilled in the art better understand the embodiments of the present application, rather than to limit the scope of the embodiments of the present application.

[0064] It should also be understood that in the various embodiments of the present application, the size of the serial number of each process does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0065] It should also be understood that the various implementation methods described in this specification can be implemented individually or in combination, and the embodiments of the present application are not limited to this.

[0066] Unless otherwise stated, all technical and scientific terms used in the embodiments of the present application are the same as the meanings generally understood by those skilled in the art of the technical field of the present application. The terms used in the present application are only for the purpose of describing specific embodiments and are not intended to limit the scope of the present application. The term "and / or" used in the present application includes any and all combinations of one or more related listed items. The singular forms of "a kind of", "above" and "the" used in the embodiments of the present application and the appended claims are also intended to include majority forms, unless the context clearly indicates other meanings. In addition, the terms "first", "second" etc. are only used for description purposes and cannot be understood as indicating or suggesting relative importance.

[0067] In the several embodiments provided in the present application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0068] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0069] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0070] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk.

[0071] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. An intelligent sleep state monitoring method based on brain waves, characterized in that: include: Collect brain wave signals of the target object through EEG equipment; Performing signal preprocessing on the brain wave signal of the target object and decomposing the signal into a set of brain wave signal components; Performing signal component feature extraction on each brain wave signal component in the set of brain wave signal components to obtain a set of brain wave signal component waveform coding features; The set of brain wave signal component waveform coding features is subjected to feature distinguishable enhancement processing to obtain a set of distinguishable enhanced brain wave signal component coding features, including: performing field mapping on each brain wave signal component waveform coding feature in the set of brain wave signal component waveform coding features to obtain a set of entry brain wave signal component waveform coding features; based on the essential features of the set of entry brain wave signal component waveform coding features as conditional features, the set of brain wave signal component waveform coding features is subjected to feature enhancement processing to obtain the set of distinguishable enhanced brain wave signal component coding features; Sleep state identification is performed based on the set of distinguishable enhanced brain wave signal component encoding features to determine a sleep state type label.

2. The method for intelligent sleep state monitoring based on brain waves according to claim 1, characterized in that: The brain wave signal of the target object is preprocessed and then decomposed into a set of brain wave signal components, including: Performing signal preprocessing on the brain wave signal of the target object to filter out high-frequency noise in the brain wave signal to obtain a noise-reduced brain wave signal; The denoised brain wave signal is subjected to signal decomposition based on CEEMDAN to obtain a set of brain wave signal components.

3. The method for intelligent sleep state monitoring based on brain waves according to claim 2, characterized in that: The brain wave signal of the target object is subjected to signal preprocessing to filter out high-frequency noise in the brain wave signal to obtain a de-noised brain wave signal, including: using a low-pass filter to filter out high-frequency noise in the brain wave signal to obtain the de-noised brain wave signal.

4. The method for intelligent sleep state monitoring based on brain waves according to claim 3, characterized in that: Signal component feature extraction is performed on each brainwave signal component in the set of brainwave signal components to obtain a set of brainwave signal component waveform coding features, including: each brainwave signal component in the set of brainwave signal components is input into a signal component feature extractor based on an RNN model to obtain a set of brainwave signal component waveform coding feature vectors as the set of brainwave signal component waveform coding features.

5. The method for intelligent sleep state monitoring based on brain waves according to claim 4, characterized in that: Field mapping is performed on each brainwave signal component waveform coding feature in the set of brainwave signal component waveform coding features to obtain a set of entry brainwave signal component waveform coding features, including: using a field mapper to field map each brainwave signal component waveform coding feature vector in the set of brainwave signal component waveform coding feature vectors to obtain a set of entry brainwave signal component waveform coding feature vectors as the set of entry brainwave signal component waveform coding features.

6. The method for intelligent sleep state monitoring based on brain waves according to claim 5, characterized in that: Based on the essential characteristics of the set of waveform coding features of the entry brain wave signal components as conditional features, the set of waveform coding features of the brain wave signal components is subjected to feature enhancement processing to obtain the set of distinguishable enhanced brain wave signal component coding features, including: Calculating the essential features of the set of waveform encoding feature vectors of the incoming brain wave signal components to obtain the brain wave signal component waveform field essential feature vectors; Taking the intrinsic characteristic vector of the brain wave signal component waveform field as the conditional characteristic vector, the set of the brain wave signal component waveform coding characteristic vectors is subjected to feature enhancement processing to obtain a set of distinguishable and enhanced brain wave signal component coding characteristic vectors as the set of distinguishable and enhanced brain wave signal component coding features.

7. The method for intelligent sleep state monitoring based on brain waves according to claim 6, characterized in that: Calculating the essential features of the set of waveform encoding feature vectors of the incoming brain wave signal components to obtain the brain wave signal component waveform field essential feature vectors, including: Calculating the field depth factor of each incoming brain wave signal component waveform encoding feature vector in the set of incoming brain wave signal component waveform encoding feature vectors to obtain a set of brain wave signal component waveform field depth factors; Based on the set of brain wave signal component waveform field depth factors, the essential characteristics of the set of incoming brain wave signal component waveform encoding feature vectors are calculated to obtain the brain wave signal component waveform field essential feature vectors.

8. The method for intelligent sleep state monitoring based on brain waves according to claim 7, characterized in that: Taking the essential characteristic vector of the brainwave signal component waveform field as the conditional characteristic vector, the set of the brainwave signal component waveform coding characteristic vectors is subjected to feature enhancement processing to obtain a set of distinguishable and enhanced brainwave signal component coding characteristic vectors as the set of distinguishable and enhanced brainwave signal component coding features, including: taking the essential characteristic vector of the brainwave signal component waveform field as the conditional characteristic vector, respectively distinguishably enhancing each brainwave signal component waveform coding characteristic vector in the set of the brainwave signal component waveform coding characteristic vector to obtain the set of distinguishable and enhanced brainwave signal component coding characteristic vectors.

9. The method for intelligent sleep state monitoring based on brain waves according to claim 8, characterized in that: Sleep state recognition is performed based on the set of distinguishable and enhanced brain wave signal component coding features to determine the sleep state type label, including: feature splicing the set of distinguishable and enhanced brain wave signal component coding feature vectors into a brain wave full-frequency domain significant aggregation coding feature vector, and then inputting the brain wave full-frequency domain significant aggregation coding feature vector into a classifier-based sleep state identifier to obtain a recognition result, and the recognition result is used to represent the sleep state type label.

10. An intelligent sleep state monitoring device based on brain waves, characterized in that: include: The brain wave signal acquisition module is used to collect the brain wave signal of the target object through the EEG device; A brain wave signal processing module, used for performing signal preprocessing on the brain wave signal of the target object and decomposing it into a set of brain wave signal components; An electroencephalogram signal component feature extraction module is used to extract signal component features of each electroencephalogram signal component in the set of electroencephalogram signal components to obtain a set of waveform coding features of the electroencephalogram signal components; The brain wave signal feature distinguishable enhancement processing module is used to perform feature distinguishable enhancement processing on the set of brain wave signal component waveform coding features to obtain a set of distinguishable enhanced brain wave signal component coding features, wherein the brain wave signal feature distinguishable enhancement processing module includes: a field mapping processing unit, used to perform field mapping on each brain wave signal component waveform coding feature in the set of brain wave signal component waveform coding features to obtain a set of entry brain wave signal component waveform coding features; a feature enhancement processing unit, used to perform feature enhancement processing on the set of brain wave signal component waveform coding features based on the essential features of the set of entry brain wave signal component waveform coding features as conditional features to obtain the set of distinguishable enhanced brain wave signal component coding features; The sleep state identification module is used to identify the sleep state based on the set of encoding features of the distinguishable enhanced brain wave signal components to determine the sleep state type label.

Images