A deep sleep staging management method and system based on electroencephalogram feedback

Abstract

The present application relates to the technical field of deep sleep management, and particularly relates to a deep sleep staging management method and system based on electroencephalogram feedback. The method comprises the following steps: collecting electroencephalogram feedback of a target user, extracting an actual deep sleep waveform and determining an expected spectrum domain of an ideal state transition stage, and locking an unachieved stage; when the unachieved stage is achieved, constructing a signal mode prediction model, combining an intervention entry point migration analysis to determine an optimal guide window period; obtaining an excellent wave band and a clutter band of the optimal guide window period through wavelet basis decomposition; when the signal-to-noise ratio is high, performing crosstalk coupling analysis and generating white noise by using a rhythm coding library to perform guide management; when the signal-to-noise ratio is low, planning a light environment and a temperature response control, and combining a color coding chain and a signal excitation result to manage deep sleep. The present application can realize guide management of different decisions for the target user to enter the staging stage of deep sleep, and improve the quality, stability and persistence of deep sleep.

Description

Technical Field

[0001] This invention relates to the field of deep sleep management technology, and in particular to a method and system for deep sleep stage management based on electroencephalogram (EEG) feedback. Background Technology

[0002] Deep sleep is a crucial stage in the human sleep cycle, playing a significant role in physiological processes such as memory consolidation, immune function recovery, and metabolic regulation. Studies have shown that insufficient deep sleep quality may lead to cognitive decline, mood disorders, and an increased risk of various chronic diseases. In recent years, with the development of wearable devices and smart home technology, sleep monitoring and staging methods based on electroencephalogram (EEG) signals have been increasingly applied to home health management scenarios. However, traditional deep sleep staging management methods typically rely on polysomnography (PSG) to analyze the frequency domain characteristics of EEG waveforms to determine sleep stages. This process is cumbersome, and the staging results are prone to errors, making it difficult to determine the deep sleep stage under the premise of different users' actual sleep states.

[0003] Secondly, existing management methods typically control smart home sleep devices to alter the sleep environment from the moment a user begins to sleep, aiming to facilitate the transition from light sleep and accelerate the entry into deep sleep. However, creating a sleep environment throughout the entire process may actually make it difficult for users with poor sleep patterns to adapt to their current sleep rhythm. For example, continuous exposure to blue light during sleep may lead to decreased or disordered melatonin levels, significantly reducing the probability of entering deep sleep. This can result in mistimed intervention and management irregularities, making it even more difficult for users to stably enter deep sleep mode. Furthermore, existing methods generally have limitations in managing wave guidance to achieve deep sleep modes under varying signal-to-noise ratios and clutter noise, exhibiting poor adaptability and robustness, leading to unstable deep sleep staging results. Summary of the Invention

[0004] This invention overcomes the shortcomings of the prior art and provides a method and system for deep sleep stage management based on electroencephalogram (EEG) feedback.

[0005] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0006] The first aspect of this invention provides a method for managing deep sleep stages based on electroencephalographic feedback, comprising the following steps:

[0007] S102: Collect EEG feedback to extract the target user's actual deep sleep waveform, and plan the expected deep sleep spectrum domain of the target user's standard deep sleep waveform based on the ideal state transition phase of sleep health quality data. Determine whether the actual deep sleep waveform is in the expected deep sleep spectrum domain and lock the unreached stage.

[0008] S104: When the target user is in the unreached stage, a signal pattern prediction model is constructed to predict the current signal pattern that is about to enter actual deep sleep. Based on the current signal pattern, the optimal guidance window for deep sleep intervention of the target user is planned according to the intervention entry point migration analysis.

[0009] S106: The low-frequency and high-frequency components of the EEG signal within the optimal guidance window period are filtered by wavelet basis decomposition to obtain the excellent and cluttered bands attached to the real-time sleep waveform within the optimal guidance window period.

[0010] S108: If the signal-to-noise ratio of the excellent band and the clutter band is greater than the preset signal-to-noise ratio, then the crosstalk coupling analysis will apply the clutter band to the easily disturbed waveform segment of the excellent band. By checking the mask tampering of the easily disturbed waveform segment through the rhythm coding library of the standard waveform, white noise audio will be generated to guide and manage the deep sleep stage, and the first stage management strategy will be obtained.

[0011] S110: If the signal-to-noise ratio of the excellent band and the clutter band is less than the preset signal-to-noise ratio, then the color gamut binning interval of the light environment-temperature response control is planned to obtain the color coding chain of the clutter band that deviates from the standard, the signal excitation trend is analyzed to obtain the signal excitation result, and the deep sleep stage is managed by combining the color coding chain and the signal excitation result to obtain the second stage management strategy.

[0012] Preferably, step S102 specifically includes the following steps:

[0013] Acquire the deep sleep staging strategy of the smart home sleep device for the target user and the pre-designed deep sleep management plan when the smart home sleep device executes the deep sleep staging strategy;

[0014] By using smart home sleep devices, a predetermined deep sleep management plan is run on the target user and the EEG response is monitored in real time to obtain the EEG feedback signal during the target user's sleep process;

[0015] A short-time Fourier transform algorithm is introduced to transform the EEG feedback signal in the time domain and frequency domain to generate the EEG feedback time spectrum. The actual deep sleep waveform of the target user is extracted from the EEG feedback time spectrum.

[0016] By acquiring the potential sleep state chain of a standardized sleep staging system to execute a predetermined deep sleep management plan, a hierarchical nested tree of sleep states is constructed. The hierarchical nested tree of sleep states is used to perform state transition triggering inference on sleep health quality data to obtain the ideal state transition path for the target user's deep sleep staging.

[0017] Based on the time-domain and frequency-domain characteristics of standard deep sleep waveforms, sleep state knowledge graph identification is performed to obtain the oscillation rhythm of standard deep sleep waveforms under different sleep state conditions.

[0018] Based on the set of observed parameters of the oscillatory rhythm, a Naive Bayes inference model is constructed by calculating the transient prior probability generated by the standard deep sleep waveform. The Naive Bayes inference model is then used to infer the oscillation sequence points of the EEG feedback spectrum of the ideal state transition path and output the steady-state conditional probability.

[0019] By combining transient prior probability and steady-state conditional probability analysis, the ideal stage landing point for the target user to generate a standard deep sleep waveform in time and frequency is determined. Based on the ideal stage landing point, the deep sleep stage planning is carried out in the EEG feedback time spectrum to obtain the target user's expected deep sleep spectrum domain.

[0020] If the actual deep sleep waveform does not exist within the expected deep sleep spectrum at this moment, the target user will be locked as not having reached the stage.

[0021] Preferably, the step of constructing a hierarchical nested tree of sleep states by acquiring the potential sleep state chain of a standardized sleep staging system to execute a predetermined deep sleep management plan, and using the hierarchical nested tree of sleep states to perform state transition triggering inference on sleep health quality data to obtain the ideal state transition path for the target user's deep sleep staging, specifically includes the following steps:

[0022] Based on big data, a sleep state knowledge graph is obtained, along with a standardized sleep staging system that matches the target user group. The sleep state knowledge graph is used to simultaneously search and identify the sleep states that occur when a predetermined deep sleep management plan is executed in accordance with the standardized sleep staging system, resulting in several potential sleep state chains.

[0023] Establish a state stack from the top-level parent state to the bottom-level leaf states, determine the contextual logic structure of the state stack based on several potential sleep state chains, and construct a hierarchical nested tree of sleep states based on the contextual logic structure.

[0024] Obtain the sleep health logs of the target user, preset multiple sets of evenly continuous health monitoring timestamps, and extract the sleep health quality data of the target user at each health monitoring timestamp when the predetermined deep sleep management plan is implemented;

[0025] Import sleep health quality data into the nested tree of sleep state hierarchy and search layer by layer from the current state level to the parent state. Calculate the current trigger matching value between the current trigger constraint premise of each stage in the standardized sleep stage system and the proposed state trigger constraint premise.

[0026] By determining whether the trigger matching threshold is greater than the trigger matching threshold of each state trigger constraint premise, the transition entropy weight of the top-level parent state corresponding to the trigger constraint premise of that state is weighted in the strategy trade-off algorithm to generate the entry channel and the exit channel.

[0027] Based on the logic pattern of entering and leaving the exit channel, the ideal state transition path of the target user's deep sleep stage is obtained by calculating the transition clues of sleep state under the premise of sleep health quality.

[0028] Preferably, step S104 specifically includes the following steps:

[0029] When the target user is locked and marked as not reached, the EEG management log of the smart home sleep device is obtained. The historical response inflection points of the target user's deep sleep waveform during EEG feedback and the timing of the emergence of each historical response inflection point are extracted from the EEG management log.

[0030] A multilayer perceptron model is introduced to construct a variational recognition encoder and a variational recognition decoder based on dynamic time series, and the historical response inflection points and corresponding emergence timing data are injected into the variational recognition encoder.

[0031] After injection, the variational recognition encoder uses a recurrent neural network to analyze the pattern characteristics of the historical response inflection point at the emergence time, and uses a gated recurrent unit to track the pattern temporal change characteristics of the emergence time with respect to the occurrence of the historical response inflection point. The pattern characteristics and pattern temporal change characteristics are combined and input into the variational recognition decoder to reconstruct the propagation, so as to build a signal pattern prediction model for the deep sleep waveform.

[0032] The signal pattern prediction model is used to predict the actual deep sleep waveform of the EEG feedback spectrum, and outputs the current signal pattern that is about to enter the actual deep sleep. Based on the mode switching rhythm of the current signal pattern, the intervention point of the actual deep sleep waveform on the future timeline is determined and defined as the first intervention entry point.

[0033] The intervention point in the desired deep sleep spectrum is obtained and defined as the second intervention point. The transfer function between the first intervention point and the second intervention point is calculated.

[0034] If the transfer function is greater than the preset transfer function, the jurisdiction of the first intervention entry point and the second intervention entry point is marked as the long-term window margin; if it is less than the preset transfer function, it is marked as the short-term window margin. The optimal guidance window period for the deep sleep intervention of the target user is generated by co-planning based on the long-term window margin and the short-term window margin.

[0035] Preferably, step S106 specifically includes the following steps:

[0036] Preset wavelet basis functions, and use wavelet basis functions to perform spectral decomposition of EEG during the optimal guidance window period to obtain low-frequency and high-frequency sub-signals of different wavelet basis decomposition layers;

[0037] Filtering is performed based on the wavelet coefficients of low-frequency and high-frequency sub-signals to extract the excellent and cluttered bands attached to the real-time sleep wave during the optimal guidance window.

[0038] Preferably, step S108 specifically includes the following steps:

[0039] If the signal-to-noise ratio between the excellent band and the clutter band is greater than the preset signal-to-noise ratio, then the rhythm representation template of the output EEG signal corresponding to the standard deep sleep waveform is obtained based on big data, and the rhythm representation template is encoded using the EEG signal encoding rules to obtain the rhythm encoding library of the standard deep sleep waveform.

[0040] The real-time sleep wave during the optimal guidance window is highlighted, and the positional pattern of the excellent wave band relative to the real-time sleep wave is extracted and defined as the first accessory wave distribution. The positional pattern of the clutter wave band relative to the real-time sleep wave is extracted and defined as the second accessory wave distribution.

[0041] The crosstalk coupling strength of the clutter band to the excellent band is calculated based on the same frequency feedback data of amplitude and frequency between the excellent band and the clutter band. Based on the second auxiliary wave distribution, only the sub-coupled waveform segments in the first auxiliary wave distribution with crosstalk coupling strength greater than the preset crosstalk coupling strength are marked and defined as the easily disturbed waveform segments of the real-time sleep wave.

[0042] A registration domain is constructed, and the real-time sleep waveform overlay mask is registered onto the standard deep sleep waveform in the registration domain. At this time, the mask segment where the easily disturbed waveform segment is located is extracted from the rhythm coding library to obtain the alignment mask sequence, and the actual coding sequence of the easily disturbed coupled waveform segment is obtained.

[0043] Obtain the association management mapping table of sleep wave frequency band and white noise frequency band for smart home sleep devices, check the tampered mask characters missing in the waveform frequency band of the actual encoded sequence compared with the normalized mask sequence, find the corresponding white noise compensation frequency band in the association management mapping table according to the tampered mask characters, and generate white noise audio control decision.

[0044] Based on the white noise audio modulation decision-making guidance for the target user's deep sleep stage, the first phase management strategy is obtained.

[0045] Preferably, step S110 specifically includes the following steps:

[0046] If the signal-to-noise ratio between the excellent band and the clutter band is less than the preset signal-to-noise ratio, then a real-time sleep wave signal spike model diagram is constructed using signal oscilloscope modeling software.

[0047] The Wigner-Ville algorithm is introduced to calculate the energy distribution of the clutter band and output the power spectral density of the clutter band. At the same time, the peak slope of the clutter band is obtained through scalar data analysis of the signal peak model.

[0048] Obtain the control rules for creating the light environment and temperature, perform differential integral estimation of the probability density distribution of each clutter band based on the power spectral density located below the signal peak model diagram, and render according to the encoding of the control rules to generate color gamut binning intervals for different clutter bands in the light environment and temperature response control.

[0049] By combining the peak slope with the amplitude and frequency of the clutter band, the signal excitation state trend of the clutter band is calculated, and the excitation steepness is obtained.

[0050] If the excitation steepness is greater than the preset excitation steepness, then the enhancement state is marked as 1 in the color gamut binning interval corresponding to the clutter band in the signal peak model diagram; if the excitation steepness is less than the preset excitation steepness, then the attenuation state is marked as 0 in the color gamut binning interval corresponding to the clutter band, thus obtaining the signal excitation result.

[0051] The registration calculation calculates the phase deviation of the excellent bands compared to the standard deep sleep waveform to obtain the phase drift value. At the same time, it extracts the deviation area between the clutter bands and the standard deep sleep waveform and obtains the color saturation scale of the deviation area in the color gamut binning interval.

[0052] Based on the signal excitation results, a guiding control law is designed. The guiding control law is executed on the smart home sleep device according to the color saturation scale to manage the light environment and temperature creation during the deep sleep stage of the target user until the phase drift value is equal to 0, thus obtaining the second phase management strategy.

[0053] A second aspect of this invention provides a deep sleep staging management system based on electroencephalogram (EEG) feedback, applicable to any of the deep sleep staging management methods based on EEG feedback described in any one of the claims. The system specifically includes:

[0054] An electroencephalogram (EEG) monitoring module is used to monitor the EEG signals of a target user during sleep in real time.

[0055] The EEG feedback module is responsible for feeding back the EEG signals of the target user during sleep and converting the EEG signals into an EEG feedback time spectrum using a short-time Fourier transform algorithm.

[0056] The data processing module is used to perform data planning and processing based on the optimal guidance window period for deep sleep intervention of the target user in coordination with the long-term window margin and the short-term window margin.

[0057] The intelligent analysis module is responsible for determining whether the signal-to-noise ratio between the excellent band and the clutter band attached to the real-time sleep waveform during the optimal guidance window is greater than the preset signal-to-noise ratio, thereby performing strategy analysis for the guidance management of the target user who is about to enter the deep sleep stage.

[0058] The phased management module is used to manage the parameter control of smart home sleep devices and provide corresponding deep sleep phased guidance to target users.

[0059] This invention addresses the technical deficiencies in the prior art, and its beneficial technical effects are as follows:

[0060] This invention acquires real-time EEG feedback signals from target users and constructs an ideal state transition staging model based on sleep health quality data. This achieves precise matching and identification of unreached stages between the user's actual deep sleep waveform and the desired deep sleep spectral domain. When an unreached stage is detected, a signal pattern prediction model and intervention entry point migration analysis are used to implement targeted interventions during the optimal guidance window. Through wavelet basis decomposition and filtering, superior and cluttered bands are separated, and different staging management strategies are selectively implemented based on signal-to-noise ratio (SNR) differences: when the SNR is high, white noise intervention is generated through crosstalk coupling analysis and comparison with a rhythm coding library to guide the rhythm of deep sleep waveforms; when the SNR is low, the deep sleep environment is stabilized through light and temperature control combined with color coding chains and signal excitation trends. This invention provides decision-making and management methods for target users entering the window stage of deep sleep under actual sleep conditions, significantly improving deep sleep quality, extending deep sleep duration, and reducing external interference that disrupts the deep sleep rhythm, resulting in more stable deep sleep. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other embodiments can be obtained from these drawings without creative effort.

[0062] Figure 1 A flowchart of the first method for deep sleep staging management based on EEG feedback is shown.

[0063] Figure 2 A flowchart of a second method for deep sleep staging management based on EEG feedback is shown.

[0064] Figure 3 A system architecture diagram of a deep sleep staging management system based on EEG feedback is shown. Detailed Implementation

[0065] To better understand the above-mentioned objectives, features, and advantages of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in these embodiments can be combined with each other.

[0066] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and therefore the scope of protection of the invention is not limited to the specific embodiments disclosed below.

[0067] The first aspect of this invention provides a method for deep sleep staging management based on electroencephalogram (EEG) feedback, such as... Figure 1 As shown, it includes the following steps:

[0068] S102: Collect EEG feedback to extract the target user's actual deep sleep waveform, and plan the expected deep sleep spectrum domain of the target user's standard deep sleep waveform based on the ideal state transition phase of sleep health quality data. Determine whether the actual deep sleep waveform is in the expected deep sleep spectrum domain and lock the unreached stage.

[0069] S104: When the target user is in the unreached stage, a signal pattern prediction model is constructed to predict the current signal pattern that is about to enter actual deep sleep. Based on the current signal pattern, the optimal guidance window for deep sleep intervention of the target user is planned according to the intervention entry point migration analysis.

[0070] S106: The low-frequency and high-frequency components of the EEG signal within the optimal guidance window period are filtered by wavelet basis decomposition to obtain the excellent and cluttered bands attached to the real-time sleep waveform within the optimal guidance window period.

[0071] S108: If the signal-to-noise ratio of the excellent band and the clutter band is greater than the preset signal-to-noise ratio, then the crosstalk coupling analysis will apply the clutter band to the easily disturbed waveform segment of the excellent band. By checking the mask tampering of the easily disturbed waveform segment through the rhythm coding library of the standard waveform, white noise audio will be generated to guide and manage the deep sleep stage, and the first stage management strategy will be obtained.

[0072] S110: If the signal-to-noise ratio of the excellent band and the clutter band is less than the preset signal-to-noise ratio, then the color gamut binning interval of the light environment-temperature response control is planned to obtain the color coding chain of the clutter band that deviates from the standard, the signal excitation trend is analyzed to obtain the signal excitation result, and the deep sleep stage is managed by combining the color coding chain and the signal excitation result to obtain the second stage management strategy.

[0073] Preferably, step S102 specifically includes the following steps:

[0074] Acquire the deep sleep staging strategy of the smart home sleep device for the target user and the pre-designed deep sleep management plan when the smart home sleep device executes the deep sleep staging strategy;

[0075] By using smart home sleep devices, a predetermined deep sleep management plan is run on the target user and the EEG response is monitored in real time to obtain the EEG feedback signal during the target user's sleep process;

[0076] A short-time Fourier transform algorithm is introduced to transform the EEG feedback signal in the time domain and frequency domain to generate the EEG feedback time spectrum. The actual deep sleep waveform of the target user is extracted from the EEG feedback time spectrum.

[0077] By acquiring the potential sleep state chain of a standardized sleep staging system to execute a predetermined deep sleep management plan, a hierarchical nested tree of sleep states is constructed. The hierarchical nested tree of sleep states is used to perform state transition triggering inference on sleep health quality data to obtain the ideal state transition path for the target user's deep sleep staging.

[0078] Based on the time-domain and frequency-domain characteristics of standard deep sleep waveforms, sleep state knowledge graph identification is performed to obtain the oscillation rhythm of standard deep sleep waveforms under different sleep state conditions.

[0079] Based on the set of observed parameters of the oscillatory rhythm, a Naive Bayes inference model is constructed by calculating the transient prior probability generated by the standard deep sleep waveform. The Naive Bayes inference model is then used to infer the oscillation sequence points of the EEG feedback spectrum of the ideal state transition path and output the steady-state conditional probability.

[0080] By combining transient prior probability and steady-state conditional probability analysis, the ideal stage landing point for the target user to generate a standard deep sleep waveform in time and frequency is determined. Based on the ideal stage landing point, the deep sleep stage planning is carried out in the EEG feedback time spectrum to obtain the target user's expected deep sleep spectrum domain.

[0081] If the actual deep sleep waveform does not exist within the expected deep sleep spectrum at this moment, the target user will be locked as not having reached the stage.

[0082] It should be noted that smart home sleep devices include smart mattresses, smart bracelets, sleep monitoring pillows, sleep aid lamps, energy-saving air conditioners, and automatic curtains. The standardized sleep staging system is the AASM staging standard system. In this system, deep sleep belongs to stage N3 of non-rapid eye movement (NREM) sleep, typically transitioning after the N2 light sleep stage. The staging of deep and light sleep usually depends on the target user's actual sleep state. For example, if the target user has recently experienced insomnia and poor sleep quality, the delta waves of deep sleep may be delayed or difficult to appear. However, current methods cannot ideally segment the target user's deep sleep stage in advance based on different sleep states, making it difficult to determine whether the actual sleep waves in the current EEG feedback have reached the deep sleep stage. This can easily cause smart home sleep devices to miss the optimal staging management opportunity for the entry of deep sleep. To address this, this method uses a sleep state knowledge graph to identify potential sleep states that are dependent on the entire process of a standardized sleep staging system and a predetermined deep sleep management plan. For example, if the predetermined deep sleep management plan decides to dim the room lights, the target user's melatonin will surge, leading to a transition from light sleep to deep sleep. This creates potential transitions between different sleep states. Therefore, this method constructs a nested tree of sleep states guided by comprehensive clues to further traverse and search the target user's sleep health quality data, thereby exploring possible actual sleep transition states and providing an accurate and highly reliable planning and reasoning basis for the target user's optimal deep sleep staging.

[0083] It should be noted that the basic flash probability of the standard deep sleep waveform when no feedback on the actual sleep state of the target user is detected is calculated by using the set of observed parameters of the oscillating rhythm. This is the transient prior probability, which defines the initial probability that the target user can enter the deep sleep stage without knowing the sleep quality. This allows the construction of a Naive Bayes inference network that can capture the characteristics of deep sleep rhythm in terms of state, thus avoiding inference bias caused by relying solely on state feature information. Then, the Naive Bayes inference model is used to calculate the fluctuations of the oscillation sequence points at each time-series sleep state transition in the ideal state transition path. This allows for the statistical calculation of the probability of reaching the responding deep sleep rhythm features under the target user's actual sleep conditions, i.e., the steady-state conditional probability. In other words, this steady-state conditional probability indicates the feature distribution captured by the Naive Bayes inference model for deep sleep waveforms along the ideal state transition path. It reflects the regular fitting nodes of the progression from one ideal sleep state to the next ideal sleep state under actual sleep conditions, leading to deep sleep. Therefore, by combining the analysis of transient prior probabilities and steady-state conditional probabilities, the ideal time-frequency landing points where the target user can generate standard deep sleep waveforms can be further determined. Based on these ideal landing points, sleep stages can be planned in stages to obtain the expected time-frequency spectral position of the waveform when the target user enters deep sleep, i.e., the expected deep sleep spectral domain. If the actual deep sleep waveform does not exist within the expected deep sleep spectral domain at this moment, it indicates that the target user has not yet entered a deep sleep stage. This method can infer the ideal stage of deep sleep for users based on their actual sleep patterns, thereby analyzing whether the actual waveform of EEG feedback has reached the ideal deep sleep stage, improving the timing accuracy of deep sleep stage management, and ensuring a smooth transition from light sleep stage to deep sleep.

[0084] Preferably, the step of constructing a hierarchical nested tree of sleep states by acquiring the potential sleep state chain of a standardized sleep staging system to execute a predetermined deep sleep management plan, and using the hierarchical nested tree of sleep states to perform state transition triggering inference on sleep health quality data to obtain the ideal state transition path for the target user's deep sleep staging, specifically includes the following steps:

[0085] Based on big data, a sleep state knowledge graph is obtained, along with a standardized sleep staging system that matches the target user group. The sleep state knowledge graph is used to simultaneously search and identify the sleep states that occur when a predetermined deep sleep management plan is executed in accordance with the standardized sleep staging system, resulting in several potential sleep state chains.

[0086] Establish a state stack from the top-level parent state to the bottom-level leaf states, determine the contextual logic structure of the state stack based on several potential sleep state chains, and construct a hierarchical nested tree of sleep states based on the contextual logic structure.

[0087] Obtain the sleep health logs of the target user, preset multiple sets of evenly continuous health monitoring timestamps, and extract the sleep health quality data of the target user at each health monitoring timestamp when the predetermined deep sleep management plan is implemented;

[0088] Import sleep health quality data into the nested tree of sleep state hierarchy and search layer by layer from the current state level to the parent state. Calculate the current trigger matching value between the current trigger constraint premise of each stage in the standardized sleep stage system and the proposed state trigger constraint premise.

[0089] By determining whether the trigger matching value is greater than the trigger matching threshold of each state trigger constraint premise, the transition entropy weight of the top-level parent state corresponding to the state trigger constraint premise is calculated in the strategy trade-off algorithm to generate the entry channel and the exit channel.

[0090] Based on the logic pattern of entering and leaving the exit channel, the ideal state transition path of the target user's deep sleep stage is obtained by calculating the transition clues of sleep state under the premise of sleep health quality.

[0091] It's important to note that by organizing all states in the potential sleep state chain hierarchically according to the development trajectory from the top-level parent state (a certain sleep state) to the bottom-level leaf state (the next possible sleep state), the contextual logic information of the structured transition of potential states is clearly defined. This allows the constructed hierarchical nested sleep state tree to share common behaviors and conditional judgments for similar potential sleep states, reducing the reuse of state definitions and making the exploration of ideal states based on actual sleep conditions and following the standardized system of deep sleep staging more accurate. Subsequently, by importing sleep health quality data and searching layer by layer along the state hierarchy of the hierarchical nested sleep state tree towards the parent state, the bottom-level leaf state that can be triggered by different sleep health quality data is found. This ensures that the reasoning for state transitions is based on the state logic conditions that can be satisfied by actual sleep conditions. The state triggering constraint premise is the triggering premise for each stage of the sleep staging criterion in the standardized sleep staging system. The current trigger matching value defines the degree of fit between sleep health quality data and the corresponding state logic of the deep sleep stage staging criterion. If the trigger matching value is greater than the trigger matching threshold of the state triggering constraint premise, it indicates that the probability of the target user experiencing the corresponding sleep state under that sleep stage when a certain sleep health quality data is present is extremely high. Therefore, the transition priority, i.e., the transition entropy weight, is further weighted in the top-level parent state to generate a sleep state transition channel that continuously fits the target user's sleep health quality data from the beginning (entry channel) to the end (exit channel). Finally, based on the exit logic pattern from the entry channel to the exit channel, the possible sleep transition progression clues that may occur under the user's actual sleep conditions according to the standardized sleep staging system can be determined, improving the interpretability of the optimal deep sleep staging for different target users.

[0092] Preferably, S104, as Figure 2 As shown, the specific steps include:

[0093] When the target user is locked and marked as not reached, the EEG management log of the smart home sleep device is obtained. The historical response inflection points of the target user's deep sleep waveform during EEG feedback and the timing of the emergence of each historical response inflection point are extracted from the EEG management log.

[0094] A multilayer perceptron model is introduced to construct a variational recognition encoder and a variational recognition decoder based on dynamic time series, and the historical response inflection points and corresponding emergence timing data are injected into the variational recognition encoder.

[0095] After injection, the variational recognition encoder uses a recurrent neural network to analyze the pattern characteristics of the historical response inflection point at the emergence time, and uses a gated recurrent unit to track the pattern temporal change characteristics of the emergence time with respect to the occurrence of the historical response inflection point. The pattern characteristics and pattern temporal change characteristics are combined and input into the variational recognition decoder to reconstruct the propagation, so as to build a signal pattern prediction model for the deep sleep waveform.

[0096] The signal pattern prediction model is used to predict the actual deep sleep waveform of the EEG feedback spectrum, and outputs the current signal pattern that is about to enter the actual deep sleep. Based on the mode switching rhythm of the current signal pattern, the intervention point of the actual deep sleep waveform on the future timeline is determined and defined as the first intervention entry point.

[0097] The intervention point in the desired deep sleep spectrum is obtained and defined as the second intervention point. The transfer function between the first intervention point and the second intervention point is calculated.

[0098] If the transfer function is greater than the preset transfer function, the jurisdiction of the first intervention entry point and the second intervention entry point is marked as the long-term window margin; if it is less than the preset transfer function, it is marked as the short-term window margin. The optimal guidance window period for the deep sleep intervention of the target user is generated by co-planning based on the long-term window margin and the short-term window margin.

[0099] It should be noted that when the target user's EEG feedback signal is not in the deep sleep stage, the historical response inflection points and corresponding emergence timing data of the deep sleep waveform feedback are obtained. The historical response inflection point is the abrupt turning point in the past time period where the target user's deep sleep waveform starts from the light sleep stage and its waveform curve significantly rises and persists for a period of time. The emergence timing data represents the moment when these abrupt turning points begin to appear. A variational recognition encoder and a variational recognition decoder are constructed using a multilayer perceptron model. The variational recognition encoder integrates a recurrent neural network and a gated recurrent unit, which can capture the complex and nonlinear pattern relationships in the deep sleep waveform, improving the abstraction ability of feature representation. The recurrent neural network is responsible for analyzing the regularity of the historical response inflection points at the emergence timing, revealing the temporal dependency chain and repetitive patterns between inflection points, thereby exploring the characteristic change patterns of signal patterns when the target user is about to enter the deep sleep stage, providing a training basis for the predictive model to capture long-range dependencies. The gated recurrent unit is responsible for tracking the temporal pattern of the emergence timing in relation to the occurrence of historical response inflection points, accurately capturing and characterizing the dynamic feature changes of signal patterns as the historical response inflection points progress with the emergence timing. Finally, the fusion of regularity features (pattern regularity features) and temporal features (pattern temporal change features) is input into the variational recognition decoder for reconstruction training. This allows the signal pattern prediction model to retain the global pattern trend while reflecting subtle temporal changes, thereby outputting high-fidelity prediction results. This makes the signal pattern prediction model more robust in predicting the trend of signal patterns when the actual waveform enters the deep sleep stage.

[0100] It should be noted that the first intervention point represents the predicted time-frequency point where the actual deep sleep waveform is about to enter the deep sleep stage under the current signal pattern, while the second intervention point represents the ideal time-frequency point where it enters the desired deep sleep spectral domain. The transfer function reflects the signal wave span from the first intervention point to the second intervention point, representing the management buffer period for the actual sleep state to enter the ideal deep sleep stage. If the transfer function is greater than the preset transfer function, it indicates that the signal wave span between the first and second intervention points is large, requiring a relatively longer management buffer period to reach the second intervention point. This suggests that the intervention guidance timing for deep sleep within this time-frequency region should be earlier, and this region is marked as a long-term window margin. Conversely, if the transfer function is smaller, it indicates a smaller signal wave span and a shorter management buffer period, and this region is marked as a short-term window margin. This method predicts the signal patterns indicating the approach to deep sleep based on the target user's actual sleep status, thereby planning the optimal guidance window for deep sleep intervention. The optimal guidance window is an idealized deep sleep guidance stage that is at an excellent and appropriate time after the user begins to fall asleep. The emphasis is on the best objective selection and highly suitable timing. This achieves early management of deep sleep entry, ensuring that phased management provides sufficient adjustment buffer while avoiding missing the optimal intervention time, guaranteeing the achievement rate and stability of deep sleep, and improving the comfort of deep sleep.

[0101] Preferably, step S106 specifically includes the following steps:

[0102] Preset wavelet basis functions, and use wavelet basis functions to perform spectral decomposition of EEG during the optimal guidance window period to obtain low-frequency and high-frequency sub-signals of different wavelet basis decomposition layers;

[0103] Filtering is performed based on the wavelet coefficients of low-frequency and high-frequency sub-signals to extract the excellent and clutter bands attached to the real-time sleep waveform within the optimal guidance window.

[0104] It should be noted that the excellent band is the effective focus band that enables the target user to concentrate highly and enter the deep sleep stage; the noisy band is the background noise band that easily distracts the target user's sleep focus and causes them to deviate from the deep sleep stage.

[0105] Preferably, step S108 specifically includes the following steps:

[0106] If the signal-to-noise ratio between the excellent band and the clutter band is greater than the preset signal-to-noise ratio, then the rhythm representation template of the output EEG signal corresponding to the standard deep sleep waveform is obtained based on big data, and the rhythm representation template is encoded using the EEG signal encoding rules to obtain the rhythm encoding library of the standard deep sleep waveform.

[0107] The real-time sleep wave during the optimal guidance window is highlighted, and the positional pattern of the excellent wave band relative to the real-time sleep wave is extracted and defined as the first accessory wave distribution. The positional pattern of the clutter wave band relative to the real-time sleep wave is extracted and defined as the second accessory wave distribution.

[0108] The crosstalk coupling strength of the clutter band to the excellent band is calculated based on the same frequency feedback data of amplitude and frequency between the excellent band and the clutter band. Based on the second auxiliary wave distribution, only the sub-coupled waveform segments in the first auxiliary wave distribution with crosstalk coupling strength greater than the preset crosstalk coupling strength are marked and defined as the easily disturbed waveform segments of the real-time sleep wave.

[0109] A registration domain is constructed, and the real-time sleep waveform overlay mask is registered onto the standard deep sleep waveform in the registration domain. At this time, the mask segment where the easily disturbed waveform segment is located is extracted from the rhythm coding library to obtain the alignment mask sequence, and the actual coding sequence of the easily disturbed coupled waveform segment is obtained.

[0110] Obtain the association management mapping table of sleep wave frequency band and white noise frequency band for smart home sleep devices, check the tampered mask characters missing in the waveform frequency band of the actual encoded sequence compared with the normalized mask sequence, find the corresponding white noise compensation frequency band in the association management mapping table according to the tampered mask characters, and generate white noise audio control decision.

[0111] Based on the white noise audio modulation decision-making guidance for the target user's deep sleep stage, the first phase management strategy is obtained.

[0112] It should be noted that if the signal-to-noise ratio (SNR) between the superior band and the clutter band is greater than the preset SNR, it indicates that the background clutter energy is low, the delta wave activity in deep sleep is relatively high, and the target user shows a clear tendency to rely on this real-time sleep waveform to enter the deep sleep stage, with only localized occasional interference. To further improve the activity stability of the real-time sleep waveform before entering the deep sleep stage, this method obtains a rhythmic representation template of the standard deep sleep waveform. This template represents the regularity of the electrical activity patterns of the standard deep sleep waveform within a specific range. Further EEG signal encoding is used to construct a rhythmic coding library of standardized templates for subsequent mask lookup. Then, using the real-time sleep waveform as a reference, the relative distribution patterns of the superior band and the clutter band on the real-time sleep waveform are extracted, namely the first and second accessory wave distributions, which clearly defines the actual rhythmic distribution relationship between the superior band and the clutter band. Since the same frequency across bands indicates a high coupling effect between the clutter band and the superior band, the crosstalk coupling strength of the clutter band on the superior band is calculated using the same frequency feedback data of amplitude and frequency. The crosstalk coupling strength reflects the ability of the clutter band to suppress the signal expression of the superior band. If the crosstalk coupling strength is greater than the preset crosstalk coupling strength, it indicates that the interference generated by the clutter band on the superior band is of a large magnitude. Combining the distribution patterns of the second and first patterns, the corresponding real-time sleep waveform segment, i.e., the easily disturbed waveform segment, is identified where the clutter band is reduced to the superior band.

[0113] It should be noted that these easily disturbed waveform segments are sensitive waveform segments that are prone to distortion in the real-time sleep waveform, which tends to maintain the trend of entering the deep sleep stage, under the interference of clutter noise. This makes it difficult for the target user to enter the deep sleep stage, so it is necessary to further eliminate clutter to guide the correct extension of the excellent waveform. To this end, this method registers the real-time sleep waveform onto the standard deep sleep waveform in the registration domain. Then, using the standard deep sleep waveform as the target constraint mask, it extracts the standardized coding sequence that the easily disturbed waveform segments should maintain, i.e., the normalization mask sequence. This normalization mask sequence reveals the waveform correction answer that ensures the real-time sleep waveform can enter the deep sleep mode when it is disturbed by clutter. The difference between the actual encoded sequence and the corrected mask sequence is fed back through the missing tampered mask symbol. The tampered mask symbol is the compensation basis for eliminating clutter bands and guiding the excellent bands to maintain the correct trend. Therefore, based on the tampered mask, a white noise audio modulation decision is established to generate sleep inertia and promote a stable transition to deep sleep mode. This generates sound stimulation management. Before entering deep sleep, the white noise audio is played through a smart home sleep device to stabilize the target user's delta wave towards the standard deep sleep waveform activity, stimulating and guiding them to enter the deep sleep stage faster and more smoothly, significantly improving the stability and continuity of deep sleep stage management.

[0114] Preferably, step S110 specifically includes the following steps:

[0115] If the signal-to-noise ratio between the excellent band and the clutter band is less than the preset signal-to-noise ratio, then a real-time sleep wave signal spike model diagram is constructed using signal oscilloscope modeling software.

[0116] The Wigner-Ville algorithm is introduced to calculate the energy distribution of the clutter band and output the power spectral density of the clutter band. At the same time, the peak slope of the clutter band is obtained through scalar data analysis of the signal peak model.

[0117] Obtain the control rules for creating the light environment and temperature, perform differential integral estimation of the probability density distribution of each clutter band based on the power spectral density located below the signal peak model diagram, and render according to the encoding of the control rules to generate color gamut binning intervals for different clutter bands in the light environment and temperature response control.

[0118] By combining the peak slope with the amplitude and frequency of the clutter band, the signal excitation state trend of the clutter band is calculated, and the excitation steepness is obtained.

[0119] If the excitation steepness is greater than the preset excitation steepness, then the enhancement state is marked as 1 in the color gamut binning interval corresponding to the clutter band in the signal peak model diagram; if the excitation steepness is less than the preset excitation steepness, then the attenuation state is marked as 0 in the color gamut binning interval corresponding to the clutter band, thus obtaining the signal excitation result.

[0120] The registration calculation calculates the phase deviation of the excellent bands compared to the standard deep sleep waveform to obtain the phase drift value. At the same time, it extracts the deviation area between the clutter bands and the standard deep sleep waveform and obtains the color saturation scale of the deviation area in the color gamut binning interval.

[0121] Based on the signal excitation results, a guiding control law is designed. The guiding control law is executed on the smart home sleep device according to the color saturation scale to manage the light environment and temperature creation during the deep sleep stage of the target user until the phase drift value is equal to 0, thus obtaining the second phase management strategy.

[0122] It should be noted that if the signal-to-noise ratio (SNR) between the optimal band and the clutter band is less than the preset SNR, it indicates that the real-time sleep waveform contains a large amount of interference noise. The signal characteristics of the optimal band may be hidden or blurred, significantly reducing the user's focus as they follow the optimal band into the N3 deep sleep mode, making it easy to remain in the N2 light sleep stage. Therefore, to minimize the large proportion of clutter interference and further improve the deep sleep focus quality of the optimal band, this method establishes a signal peak model diagram of the real-time sleep waveform. This diagram can express the excitation rate and sharpness of the clutter band signal change with the peak slope, thereby tracking the excitation trend of the rising and falling edges of the clutter band. This provides a quantifiable basis for the enhancement and attenuation of clutter interference and obtains the excitation steepness of the clutter band. Prior to this, the power spectral density of the clutter band energy distribution is calculated using the Wigner-Ville algorithm. The Wigner-Ville algorithm is used to simultaneously analyze the energy distribution of a signal in the time and frequency domains. The power spectral density visually illustrates the distribution of energy carried by the EEG signal in the frequency components corresponding to the clutter bands, reflecting the density variation of energy magnitude. Therefore, this method uses calculus to estimate the probability density distribution of the energy variation of the power spectral density corresponding to each clutter band, thus forming a binned thermodynamic density architecture, providing a visualization platform for the magnitude variation of energy density. Furthermore, the control rules for creating a light environment and temperature can suppress the interference of clutter bands on excellent bands to the greatest extent by regulating and managing the sleep light environment and temperature. Therefore, the RGB mapping gradient encoding of the clutter bands is further rendered according to the control rules to eliminate the thermal density architecture. This results in the projection display of the corresponding power energy density elimination color code in the bin below each clutter band on the peak model diagram, forming a dedicated color gamut bin area for each clutter band. The color gamut bin area not only shows the energy density change of the power spectrum of the clutter band in the form of thermal distribution, but also dynamically formulates the corresponding RGB color control scale control plan for eliminating the clutter band with power energy density changes at each point. For example, when the power spectrum of the clutter band in the 35-55 Hz frequency band occurs in the color gamut bin area, the wavelength color value of sleep blue light in the 215-400nm range will be displayed to suppress it, making the sleep light environment and temperature created by subsequent regulation more suitable for entering deep sleep mode and effectively suppressing the interference of clutter.

[0123] It should be noted that if the excitation steepness is greater than the preset excitation steepness, it indicates that the signal excitation trend of the clutter band is in an upward state, indicating that the clutter band activity intensity is high. Therefore, in the signal peak model diagram, the corresponding color gamut binning interval of this clutter band is marked as an enhanced state of 1. Conversely, it indicates that the signal excitation trend is in a downward state, with a low clutter band activity coefficient and intensity. Therefore, in the corresponding color gamut binning interval of the clutter band, it is marked as a decaying state of 0. The signal excitation results can provide instruction specifications for the management and control of the light environment and temperature, and precisely constrain the adjustment of the clutter band elimination. The color saturation scale provides the range of adjustment parameter values. The phase drift value reflects the phase difference amplitude between the current excellent wave and the standardized waveform signal transmission. Therefore, only by eliminating this phase drift value can the excellent wave smoothly guide the target user into the N3 deep sleep stage along the standard deep sleep waveform. Therefore, the light environment and temperature creation for the target user's deep sleep stage managed by the guidance control law and color coding chain must meet the condition that the phase drift value is equal to 0. This method can precisely create and manage the light environment and temperature during sleep, enabling users to quickly and smoothly enter deep sleep mode, reducing sleep alertness and misdirection, and improving the quality of deep sleep.

[0124] Furthermore, the aforementioned method for managing deep sleep stages based on EEG feedback also includes the following steps:

[0125] Obtain the management plan associated with the first phase management strategy and the second phase management strategy, and at the same time obtain the candidate control parameter solutions for the first phase management strategy and the second phase management strategy;

[0126] Obtain the actual number of faults in the real-time sleep waves that continuously produce excellent wavebands within the optimal guidance window period when the target user is in the unreached stage, and construct the actual fault Bezier curve of the EEG feedback response based on the actual number of faults.

[0127] The tolerance threshold for excellent band tomographic feedback is obtained through a predetermined deep sleep management plan, and a tolerance tomographic Bezier curve is constructed based on the tolerance threshold.

[0128] A discrete optimization algorithm is introduced to calculate the misalignment function between the actual fault Bézier curve and the tolerance fault Bézier curve. Based on the misalignment function, a backtracking scale is preset. According to the backtracking scale, the discrete optimization algorithm performs a globally optimal discrete simultaneous solution for each candidate control parameter solution to obtain the fit of each candidate control parameter solution.

[0129] If the fit is greater than the preset fit, the candidate control parameter solution corresponding to the fit is used to replace the current control parameter, and several feasible control parameter examples are obtained to optimize the actual fault without exceeding the tolerance fault limit.

[0130] The simulation runs and records the coordination index between each set of feasible control parameters and the management scheme. The management scheme with the largest coordination index is selected as the best management scheme and uploaded to the smart home sleep device terminal to manage the white noise, light environment and temperature parameters during the deep sleep stage of the target user.

[0131] It should be noted that since the first and second phased management strategies are different sleep environment management decisions for the target user's actual sleep conditions during deep sleep stages, there are numerous suitable control parameters. By analyzing the difference between the actual fault lines of the excellent wave and the idealized fault lines tolerated by the management plan, a misalignment function is obtained. This misalignment function expresses the degree of irrationality of the current control parameters. Based on the misalignment function, a backtracking scale for combined optimization is preset, and a discrete combination analysis of the global optimal trend is performed on all candidate control parameter solutions to form a control parameter combination that can optimize the current fault line phenomenon of the excellent wave. If the fit is greater than the preset fit, it indicates that the candidate control parameter solution can better maintain the continuity of the excellent wave transmission and improve the inertia quality of the N3 deep sleep wave. Therefore, this candidate control parameter solution is given priority to replace the current control parameters. Finally, the operation of feasible control parameter examples is simulated to analyze whether the management scheme can coordinate the replaced control parameters. This method can screen the optimal management scheme with suitable and feasible control parameters for the first and second stage management strategies, reduce the probability of errors in the control of white noise, light environment and temperature parameters during deep sleep, and improve the accuracy and guidance performance of deep sleep stage management.

[0132] A second aspect of this invention provides a deep sleep staging management system based on electroencephalogram (EEG) feedback, such as... Figure 3 As shown, the system applied to any of the aforementioned deep sleep staging management methods based on EEG feedback specifically includes:

[0133] An electroencephalogram (EEG) monitoring module is used to monitor the EEG signals of a target user during sleep in real time.

[0134] The EEG feedback module is responsible for feeding back the EEG signals of the target user during sleep and converting the EEG signals into an EEG feedback time spectrum using a short-time Fourier transform algorithm.

[0135] The data processing module is used to perform data planning and processing based on the optimal guidance window period for deep sleep intervention of the target user in coordination with the long-term window margin and the short-term window margin.

[0136] The intelligent analysis module is responsible for determining whether the signal-to-noise ratio between the excellent band and the clutter band attached to the real-time sleep waveform during the optimal guidance window is greater than the preset signal-to-noise ratio, thereby performing strategy analysis for the guidance management of the target user who is about to enter the deep sleep stage.

[0137] The phased management module is used to manage the parameter control of smart home sleep devices and provide corresponding deep sleep phased guidance to target users.

[0138] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for managing deep sleep stages based on electroencephalogram (EEG) feedback, characterized in that, Includes the following steps: S102: Collect EEG feedback to extract the target user's actual deep sleep waveform, and plan the expected deep sleep spectrum domain for the target user to generate a standard deep sleep waveform based on the ideal state transition stage of sleep health quality data. Determine whether the actual deep sleep waveform is located in the expected deep sleep spectrum domain. If not, lock the unreached stage. S104: When the target user is in the unreached stage, a signal pattern prediction model is constructed to predict the current signal pattern that is about to enter actual deep sleep. Based on the current signal pattern, the optimal guidance window for deep sleep intervention of the target user is planned according to the intervention entry point migration analysis. S106: The low-frequency and high-frequency components of the EEG signal within the optimal guidance window period are filtered by wavelet basis decomposition to obtain the excellent and cluttered bands attached to the real-time sleep waveform within the optimal guidance window period. S108: If the signal-to-noise ratio of the excellent band and the clutter band is greater than the preset signal-to-noise ratio, then the crosstalk coupling analysis will apply the clutter band to the easily disturbed waveform segment of the excellent band. By checking the mask tampering of the easily disturbed waveform segment through the rhythm coding library of the standard waveform, white noise audio will be generated to guide and manage the deep sleep stage, and the first stage management strategy will be obtained. S110: If the signal-to-noise ratio of the excellent band and the clutter band is less than the preset signal-to-noise ratio, then the color gamut binning interval of the light environment-temperature response control is planned to obtain the color coding chain of the clutter band that deviates from the standard, the signal excitation trend is analyzed to obtain the signal excitation result, and the deep sleep stage is managed by combining the color coding chain and the signal excitation result to obtain the second stage management strategy. Specifically, S104 includes the following steps: When the target user is locked and marked as not reached, the EEG management log of the smart home sleep device is obtained. The historical response inflection points of the target user's deep sleep waveform during EEG feedback and the timing of the emergence of each historical response inflection point are extracted from the EEG management log. A multilayer perceptron model is introduced to construct a variational recognition encoder and a variational recognition decoder based on dynamic time series, and the historical response inflection points and corresponding emergence timing data are injected into the variational recognition encoder. After injection, the variational recognition encoder uses a recurrent neural network to analyze the pattern characteristics of the historical response inflection point at the emergence time, and uses a gated recurrent unit to track the pattern temporal change characteristics of the emergence time with respect to the occurrence of the historical response inflection point. The pattern characteristics and pattern temporal change characteristics are combined and input into the variational recognition decoder to reconstruct the propagation, so as to build a signal pattern prediction model for the deep sleep waveform. The signal pattern prediction model is used to predict the actual deep sleep waveform of the EEG feedback spectrum, and outputs the current signal pattern that is about to enter the actual deep sleep. Based on the mode switching rhythm of the current signal pattern, the intervention point of the actual deep sleep waveform on the future timeline is determined and defined as the first intervention entry point. The intervention point in the desired deep sleep spectrum is obtained and defined as the second intervention point. The transfer function between the first intervention point and the second intervention point is calculated. If the transfer function is greater than the preset transfer function, the jurisdiction of the first intervention entry point and the second intervention entry point is marked as the long-term window margin; if it is less than the preset transfer function, it is marked as the short-term window margin. The optimal guidance window period for the deep sleep intervention of the target user is generated by co-planning based on the long-term window margin and the short-term window margin.

2. The method for deep sleep staging management based on EEG feedback according to claim 1, characterized in that, S102 specifically includes the following steps: Acquire the deep sleep staging strategy of the smart home sleep device for the target user and the pre-designed deep sleep management plan when the smart home sleep device executes the deep sleep staging strategy; By using smart home sleep devices, a predetermined deep sleep management plan is run on the target user and the EEG response is monitored in real time to obtain the EEG feedback signal during the target user's sleep process; A short-time Fourier transform algorithm is introduced to transform the EEG feedback signal in the time domain and frequency domain to generate the EEG feedback time spectrum. The actual deep sleep waveform of the target user is extracted from the EEG feedback time spectrum. By acquiring the potential sleep state chain of a standardized sleep staging system to execute a predetermined deep sleep management plan, a hierarchical nested tree of sleep states is constructed. The hierarchical nested tree of sleep states is used to perform state transition triggering inference on sleep health quality data to obtain the ideal state transition path for the target user's deep sleep staging. Based on the time-domain and frequency-domain characteristics of standard deep sleep waveforms, sleep state knowledge graph identification is performed to obtain the oscillation rhythm of standard deep sleep waveforms under different sleep state conditions. Based on the set of observed parameters of the oscillatory rhythm, a Naive Bayes inference model is constructed by calculating the transient prior probability generated by the standard deep sleep waveform. The Naive Bayes inference model is then used to infer the oscillation sequence points of the EEG feedback spectrum of the ideal state transition path and output the steady-state conditional probability. By combining transient prior probability and steady-state conditional probability analysis, the ideal stage landing point for the target user to generate a standard deep sleep waveform in time and frequency is determined. Based on the ideal stage landing point, the deep sleep stage planning is carried out in the EEG feedback time spectrum to obtain the target user's expected deep sleep spectrum domain. If the actual deep sleep waveform does not exist within the expected deep sleep spectrum at this moment, the target user will be locked as not having reached the stage.

3. The method for deep sleep staging management based on EEG feedback according to claim 2, characterized in that, The process involves constructing a hierarchical nested tree of sleep states by acquiring the potential sleep state chain of a standardized sleep staging system to execute a predetermined deep sleep management plan. This tree is then used to perform state transition triggering inference on sleep health quality data to obtain the ideal state transition path for the target user's deep sleep staging. Specifically, this includes the following steps: Based on big data, a sleep state knowledge graph is obtained, along with a standardized sleep staging system that matches the target user group. The sleep state knowledge graph is used to simultaneously search and identify the sleep states that occur when a predetermined deep sleep management plan is executed in accordance with the standardized sleep staging system, resulting in several potential sleep state chains. Establish a state stack from the top-level parent state to the bottom-level leaf states, determine the contextual logic structure of the state stack based on several potential sleep state chains, and construct a hierarchical nested tree of sleep states based on the contextual logic structure. Obtain the sleep health logs of the target user, preset multiple sets of evenly continuous health monitoring timestamps, and extract the sleep health quality data of the target user at each health monitoring timestamp when the predetermined deep sleep management plan is implemented; Import sleep health quality data into the nested tree of sleep state hierarchy and search layer by layer from the current state level to the parent state. Calculate the current trigger matching value between the current trigger constraint premise of each stage in the standardized sleep stage system and the proposed state trigger constraint premise. By determining whether the trigger matching threshold is greater than the trigger matching threshold of each state trigger constraint premise, the transition entropy weight of the top-level parent state corresponding to the trigger constraint premise of that state is weighted in the strategy trade-off algorithm to generate the entry channel and the exit channel. Based on the logic pattern of entering and leaving the exit channel, the ideal state transition path of the target user's deep sleep stage is obtained by calculating the transition clues of sleep state under the premise of sleep health quality.

4. The method for deep sleep staging management based on EEG feedback according to claim 1, characterized in that, S106 specifically includes the following steps: Preset wavelet basis functions, and use wavelet basis functions to perform spectral decomposition of EEG during the optimal guidance window period to obtain low-frequency and high-frequency sub-signals of different wavelet basis decomposition layers; Filtering is performed based on the wavelet coefficients of low-frequency and high-frequency sub-signals to extract the excellent and cluttered bands attached to the real-time sleep wave during the optimal guidance window.

5. The method for deep sleep staging management based on EEG feedback according to claim 1, characterized in that, S108 specifically includes the following steps: If the signal-to-noise ratio between the excellent band and the clutter band is greater than the preset signal-to-noise ratio, then the rhythm representation template of the output EEG signal corresponding to the standard deep sleep waveform is obtained based on big data, and the rhythm representation template is encoded using the EEG signal encoding rules to obtain the rhythm encoding library of the standard deep sleep waveform. The real-time sleep wave during the optimal guidance window is highlighted, and the positional pattern of the excellent wave band relative to the real-time sleep wave is extracted and defined as the first accessory wave distribution. The positional pattern of the clutter wave band relative to the real-time sleep wave is extracted and defined as the second accessory wave distribution. The crosstalk coupling strength of the clutter band to the excellent band is calculated based on the same frequency feedback data of amplitude and frequency between the excellent band and the clutter band. Based on the second auxiliary wave distribution, only the sub-coupled waveform segments in the first auxiliary wave distribution with crosstalk coupling strength greater than the preset crosstalk coupling strength are marked and defined as the easily disturbed waveform segments of the real-time sleep wave. A registration domain is constructed, and the real-time sleep waveform overlay mask is registered onto the standard deep sleep waveform in the registration domain. At this time, the mask segment where the disturbed waveform segment is located is extracted from the rhythm coding library to obtain the alignment mask sequence, and the actual coding sequence of the disturbed waveform segment is obtained. Obtain the association management mapping table of sleep wave frequency band and white noise frequency band for smart home sleep devices, check the tampered mask characters missing in the waveform frequency band of the actual encoded sequence compared with the normalized mask sequence, find the corresponding white noise compensation frequency band in the association management mapping table according to the tampered mask characters, and generate white noise audio control decision. Based on the white noise audio modulation decision-making guidance for the target user's deep sleep stage, the first phase management strategy is obtained.

6. The method for deep sleep staging management based on EEG feedback according to claim 1, characterized in that, S110 specifically includes the following steps: If the signal-to-noise ratio between the excellent band and the clutter band is less than the preset signal-to-noise ratio, then a real-time sleep wave signal spike model diagram is constructed using signal oscilloscope modeling software. The Wigner-Ville algorithm is introduced to calculate the energy distribution of the clutter band and output the power spectral density of the clutter band. At the same time, the peak slope of the clutter band is obtained through scalar data analysis of the signal peak model. Obtain the control rules for creating the light environment and temperature, perform differential integral estimation of the probability density distribution of each clutter band based on the power spectral density located below the signal peak model diagram, and render according to the encoding of the control rules to generate color gamut binning intervals for different clutter bands in the light environment and temperature response control. By combining the peak slope with the amplitude and frequency of the clutter band, the signal excitation state trend of the clutter band is calculated, and the excitation steepness is obtained. If the excitation steepness is greater than the preset excitation steepness, then the enhancement state is marked as 1 in the color gamut binning interval corresponding to the clutter band in the signal peak model diagram; if the excitation steepness is less than the preset excitation steepness, then the attenuation state is marked as 0 in the color gamut binning interval corresponding to the clutter band, thus obtaining the signal excitation result. The registration calculation calculates the phase deviation of the excellent bands compared to the standard deep sleep waveform to obtain the phase drift value. At the same time, it extracts the deviation area between the clutter bands and the standard deep sleep waveform and obtains the color saturation scale of the deviation area in the color gamut binning interval. Based on the signal excitation results, a guiding control law is designed. The guiding control law is executed on the smart home sleep device according to the color saturation scale to manage the light environment and temperature creation during the deep sleep stage of the target user until the phase drift value is equal to 0, thus obtaining the second phase management strategy.

7. A deep sleep staging management system based on electroencephalogram (EEG) feedback, characterized in that, Applied to implement the deep sleep staging management method based on EEG feedback as described in any one of claims 1-6, the system specifically includes: An electroencephalogram (EEG) monitoring module is used to monitor the EEG signals of a target user during sleep in real time. The EEG feedback module is responsible for feeding back the EEG signals of the target user during sleep and converting the EEG signals into an EEG feedback time spectrum using a short-time Fourier transform algorithm. The data processing module is used to perform data planning and processing based on the optimal guidance window period for deep sleep intervention of the target user in coordination with the long-term window margin and the short-term window margin. The intelligent analysis module is responsible for determining whether the signal-to-noise ratio between the excellent band and the clutter band attached to the real-time sleep waveform during the optimal guidance window is greater than the preset signal-to-noise ratio, thereby performing strategy analysis for the guidance management of the target user who is about to enter the deep sleep stage. The phased management module is used to manage the parameter control of smart home sleep devices and provide corresponding deep sleep phased guidance to target users.

Images