Electroencephalogram micro-state recognition and sleep-aiding guide method for transitional period of falling asleep

Abstract

The invention provides an electroencephalogram micro-state recognition and sleep-aiding guiding method for a sleep transition period, and relates to the technical field of intelligent sleep aiding. The method comprises the following steps: synchronously acquiring environmental disturbance and electroencephalogram physiological signals in a sleep transition period, calculating two interference indexes of noise-electroencephalogram coupling and illumination-melatonin regulation, and carrying out normalized coupling pretreatment; a stable environment interference coupling strength index is obtained through iterative feedback, a nonlinear evolution function is matched according to interference grade branches, and a dimensionless sleep state index is generated; and dynamically adjusting an electroencephalogram micro-state matching threshold value, adapting to sleep-aiding guide parameters such as audio and light, and circularly checking the effect until the standard is reached. According to the method, the problems of inaccurate electroencephalogram micro-state recognition and poor sleep-aiding guide adaptation under bedroom low-frequency noise and illumination coupling interference are solved, the recognition precision and the sleep-aiding adaptation are improved, the sleep transition period is shortened, and the method is suitable for indoor complex interference sleep scenes.

Description

Technical Field

[0001] This invention relates to the field of intelligent sleep aid technology, and in particular to a method for recognizing and guiding sleep through the brainwave microstate during the sleep transition period. Background Technology

[0002] With the integration of smart home and sleep health technologies, intelligent sleep aid systems based on electroencephalogram (EEG) signal monitoring have become an important means of improving users' sleep quality. The sleep onset transition period, a crucial stage transitioning from wakefulness to light sleep, typically lasts 10-30 minutes. Changes in EEG signal characteristics during this stage are the core basis for determining a user's sleep state and represent the optimal window for initiating sleep-aiding guidance. In practical applications within the home bedroom, dedicated equipment monitors the microstate characteristics of the user's EEG rhythms during the sleep onset transition period in real time, simultaneously activating sleep-aiding techniques such as audio guidance and lighting adjustments. This effectively helps users fall asleep quickly, shortens the sleep onset transition period, and improves sleep efficiency.

[0003] However, significant environmental disturbances exist in home bedrooms, primarily manifested as low-frequency environmental noise and fluctuations in ambient light. These two types of interference often coexist and create a coupling effect. Low-frequency environmental noise mainly includes traffic noise outside the window and air conditioner operation, with frequencies concentrated in the 20-200Hz range. This frequency band overlaps with the basic rhythm band of EEG signals, masking the EEG signals. This causes subtle changes in key features of EEG rhythm microstates, such as EEG rhythm, microstate duration, and transition frequency, to be obscured by noise, making them difficult to extract accurately. Fluctuations in ambient light mainly include direct streetlights at night and light leakage through curtains, with light intensity fluctuations ranging from 0.1-50 lux. Even weak light fluctuations can inhibit the secretion of melatonin in the human brain. As a core hormone regulating the human sleep-wake cycle, reduced melatonin secretion directly prolongs the user's sleep transition period, increasing the difficulty of falling asleep.

[0004] Under the coupled effect of the two types of environmental interference mentioned above, existing EEG microstate matching and sleep-aid guidance technologies face deep-seated technical contradictions and bottlenecks that are difficult to overcome. Specifically, existing technologies for recognizing EEG rhythm microstates are mostly based on fixed threshold algorithms, without considering the dynamic impact of environmental coupling interference on EEG physiological signal characteristics. This results in low accuracy of EEG microstate matching and an inability to accurately reflect the user's real-time sleep state. In addition, the sleep-aid guidance strategy is linearly bound to the EEG rhythm recognition results. Guidance parameters such as audio frequency and light intensity cannot be adjusted in real time according to the dynamic changes of environmental coupling interference, resulting in guidance adaptation lag. This not only fails to help users fall asleep but may also cause users to become irritable due to mismatched guidance parameters, further prolonging the sleep transition period.

[0005] To address these issues, existing technologies attempt to optimize from a single dimension. For example, they may use filtering algorithms to reduce the interference of environmental noise on EEG signals or use light-blocking devices to reduce the impact of light fluctuations on melatonin secretion. However, these methods can only alleviate the impact of a single environmental disturbance and cannot handle the coupled interference problem of low-frequency noise and light fluctuations. Other technologies attempt to improve the sampling accuracy of dedicated equipment or optimize the parameter combination of sleep aid guidance. However, they have not established a dynamic correlation mechanism between the intensity of environmental interference and EEG rhythm recognition and sleep aid guidance, resulting in limited improvement in recognition accuracy and guidance adaptability. They still cannot meet the practical application needs in the non-steady-state coupled interference scenario of a home bedroom. Summary of the Invention

[0006] The main objective of this invention is to overcome the shortcomings of the prior art and provide a method for EEG microstate recognition and sleep aid guidance for the sleep transition period, solving the problems of low accuracy of EEG microstate matching and poor adaptability of sleep aid guidance under the coupling interference of low-frequency environmental noise and light fluctuations during the sleep transition period in home bedrooms.

[0007] To achieve the above objectives, the present invention proposes a method for EEG microstate recognition and sleep-aid guidance during the sleep transition period, comprising the following steps: S1. Simultaneously acquire environmental disturbance signals and EEG physiological signals during the sleep transition period, and output the raw input dataset containing environmental feature data and raw EEG data. S2. Based on the original input dataset, calculate the multi-dimensional environmental interference index and perform normalized coupling preprocessing on it, and calculate the feature difference between the indices. S3. Based on the preprocessed multi-dimensional environmental interference index and the characteristic difference between the indices, calculate the environmental interference coupling strength index, correct the coupling calculation parameters through an iterative feedback mechanism, and output a stable coupling strength index and the corrected coupling calculation parameters. S4. Based on a stable coupling strength index, the corresponding nonlinear evolution function is matched through a branch decision mechanism to transform the coupling strength index into a dimensionless sleep state index that is monotonically correlated with the coupling strength. S5. Based on the sleep state index and the original EEG data, dynamically adjust the EEG microstate matching threshold and complete the EEG microstate matching. At the same time, dynamically adapt the sleep aid guidance parameters according to the sleep state index and the EEG microstate matching result, and output the EEG microstate matching result and the adapted sleep aid guidance parameters. S6. Based on the EEG microstate matching results and the adapted sleep-aid guidance parameters, verify the sleep-aid guidance adaptation effect: If the expected results are not achieved, repeat steps S1 to S5 to adjust the calculation parameters; If the desired effect is achieved, the optimized process parameters, the final sleep state index, and the sleep-aid guidance plan will be output.

[0008] A further improvement of the present invention is that, in step S1, dual parallel links are used to acquire the environmental disturbance signal and the electroencephalogram (EEG) signal respectively, and the acquisition duration is synchronized with the duration of the sleep transition period. The environmental disturbance signal includes environmental noise signal and real-time illumination signal; The environmental characteristic data includes environmental noise amplitude, noise power spectral density, and real-time light intensity data; the raw EEG data is the EEG rhythm characteristic data of the sleep transition period.

[0009] A further improvement of the present invention is that, in step S2, the multi-dimensional environmental interference index includes a noise-EEG coupling interference index characterizing the interference of noise on EEG physiological signals and a light-melatonin regulation abnormality index characterizing the effect of light on melatonin secretion regulation; the noise-EEG coupling interference index is calculated based on the amplitude ratio and power spectral density ratio of noise and EEG rhythm, as well as preset weighting coefficients and correction coefficients; the light-melatonin regulation abnormality index is calculated based on real-time light intensity, the inhibition coefficient of light on melatonin secretion, and preset correction coefficients. The normalization coupling preprocessing maps each environmental disturbance index to the same value range, eliminating the differences in value ranges between indices; the characteristic difference between indices is the difference in absolute values ​​of each environmental disturbance index after normalization.

[0010] A further improvement of the present invention is that, in step S3, the environmental interference coupling strength index is the product of the coupling correction coefficient and the sum of the preprocessed multidimensional environmental interference index and the characteristic differences between the indices. The iterative feedback mechanism includes: calculating an initial value of the coupling strength index, comparing it with a preset scene adaptation threshold, and if the initial value exceeds a reasonable range or the fluctuation exceeds a preset threshold, iteratively correcting the coupling calculation parameters and recalculating until the coupling strength index stabilizes within a preset reasonable range.

[0011] A further improvement of the present invention is that the preset scene adaptation threshold divides environmental coupling interference into three interference levels; The coupling calculation parameters include coupling correction coefficients, which have preset correction ranges and correction step sizes.

[0012] A further improvement of the present invention is that, in step S4, the branch decision mechanism selects a matching nonlinear evolution function based on the environmental coupling interference level corresponding to the coupling strength index; different environmental coupling interference levels correspond to different nonlinear evolution paths, and the nonlinear evolution function makes the sleep state index and the coupling strength index monotonically increasing.

[0013] A further improvement of the present invention is that the environmental coupling interference levels include weak coupling, medium coupling, and strong coupling interference. Weak coupling interference corresponds to a logarithmic nonlinear evolution function, medium coupling interference corresponds to a square root nonlinear evolution function, and strong coupling interference corresponds to a linear scalar multiplication nonlinear evolution function.

[0014] A further improvement of the present invention is that, in step S5, the dynamic adjustment rule of the EEG microstate matching threshold is: the smaller the sleep state index value, the more lenient the recognition threshold; the larger the sleep state index value, the more stringent the recognition threshold. The EEG microstate matching results include microstate type, duration, and transition frequency; The dynamic adaptation rule of the sleep aid guidance parameters is as follows: the smaller the sleep state index value, the gentler the guidance parameters; the larger the sleep state index value, the stronger the anti-interference ability of the guidance parameters. The sleep aid guidance parameters include audio parameters and light parameters. The audio parameters include audio frequency and audio volume, and the light parameters include light intensity and light color temperature.

[0015] A further improvement of the present invention is that the expected effect in step S6 includes: the accuracy rate of EEG microstate matching is not lower than a preset accuracy threshold, and the shortening ratio of the sleep transition period is not lower than a preset duration threshold. The effectiveness of sleep aid guidance can be verified by obtaining feedback on the user's sleep state, which includes the proportion of EEG rhythm characteristics and the frequency of body movements.

[0016] A further improvement of the present invention is that the method is applicable to indoor sleep scenarios where there is environmental coupling interference; the sleep transition period is the time period from wakefulness to light sleep; and the environmental disturbance is the coupling interference formed by low-frequency environmental noise and ambient light fluctuations.

[0017] A further improvement of the present invention is that when the expected effect is not achieved, the calculation parameters adjusted include the normalization parameters in step S2, the correction range of the coupling calculation parameters in step S3, and the nonlinear evolution function parameters in step S4, and are adjusted sequentially from the calculation of the bottom index to the evaluation of the top state.

[0018] Compared with the prior art, the present invention has at least the following beneficial effects: (1) Using dual parallel links to synchronously acquire environmental disturbance signals and EEG physiological signals, ensuring the synchronization and integrity of the data, avoiding the distortion of coupling interference assessment caused by acquisition lag, and reducing the impact of non-steady-state interference such as sudden noise enhancement and sudden change in illumination on data acquisition, so as to provide accurate raw data for subsequent coupling interference intensity calculation.

[0019] (2) Based on the coupling interference level, a branch decision mechanism is constructed to match the corresponding nonlinear evolution function for different interference intensities. The coupling intensity index is transformed into a dimensionless monotonic sleep state index, which not only ensures the monotonic correlation between the evaluation index and the coupling interference intensity, but also adapts to the evaluation requirements of different interference intensity scenarios. It ensures the evaluation accuracy when the interference is weak and improves the response speed when the interference is strong, thus solving the problem that the traditional linear evolution path cannot adapt to the dynamic changes of coupling interference.

[0020] (3) The sleep state index is matched with the EEG microstate and deeply bound to the sleep aid guidance. The EEG rhythm recognition threshold and sleep aid guidance parameters are dynamically adjusted according to the evaluation index, which solves the problem of low EEG rhythm recognition accuracy under non-steady-state interference. At the same time, it achieves accurate matching between sleep aid guidance and the user's real-time sleep state and the intensity of environmental coupling interference.

[0021] (4) Through the synergistic effect of the above technical solutions, the accurate identification of brainwave rhythm microstates and the dynamic adaptation of sleep-aid guidance can be achieved, which can effectively offset the influence of low-frequency noise and light fluctuation coupling interference on the user's sleep, inhibit the problem of reduced melatonin secretion, and help the user quickly enter a light sleep state. Attached Figure Description

[0022] Figure 1 This is a flowchart of the EEG microstate recognition and sleep-aid guidance method for the sleep transition period of the present invention. Detailed Implementation

[0023] In the following description, numerous specific details are set forth in order to provide a more thorough understanding of the invention. However, it will be apparent to those skilled in the art that the invention can be practiced without one or more of these details. In other instances, certain technical features well-known in the art have not been described in order to avoid obscuring the invention. Example 1

[0024] This embodiment provides a method for EEG microstate recognition and sleep-aid guidance during the sleep transition period. (See...) Figure 1 Specifically, it includes the following steps: S1. Parallel acquisition of environmental disturbance signals and electroencephalographic signals during the sleep transition period. The system synchronously acquires environmental disturbance signals and electroencephalogram (EEG) signals during the sleep transition period in a home bedroom, and outputs a raw input data set. The environmental disturbance signals include environmental noise signals and real-time illumination signals, and the raw input data set includes raw EEG signal data, environmental noise amplitude data, noise power spectral density data, and real-time illumination intensity data. The system uses dual parallel links to acquire environmental disturbance signals and EEG signals separately, with the sampling frequency set differently according to the signal type, and the acquisition duration synchronized with the sleep transition period.

[0025] S2. Based on the original input data set, calculate the two coupling interference indices in parallel and perform coupling preprocessing. Based on the original input data set output by S1, the noise-EEG coupling interference index and the light-melatonin regulation abnormality index are calculated in parallel. After coupling preprocessing of the two indices, the preprocessed indices and their real-time difference are output. The coupling preprocessing is a normalization process, which normalizes the two indices to the same value range, eliminating the difference in their value ranges. The calculation models of the two indices run in parallel to improve the efficiency of feature calculation.

[0026] S3. Based on the preprocessed index data, calculate the coupling strength index and perform iterative correction. Based on the two preprocessed indices output by S2 and their real-time difference, the coupling strength index of environmental interference during the sleep transition period is calculated. The coupling function parameters are corrected through an iterative feedback mechanism, and a stable coupling strength index value and the corrected coupling function parameters are output. At the same time, the coupling strength index value is fed back to S2 to optimize the normalization preprocessing accuracy of the two indices. The iterative feedback mechanism is as follows: first, the initial value of the coupling strength index is calculated and compared with a preset scene adaptation threshold. If the initial value exceeds a reasonable range or fluctuates too much, the coupling function parameters are iteratively corrected and recalculated until the coupling strength index value stabilizes within a reasonable range.

[0027] S4. Based on a stable coupling strength index value, the sleep state index is obtained through nonlinear evolution of branch decision-making. Based on the stable coupling strength index value output by S3, a branch decision mechanism selects the corresponding nonlinear evolution path, transforming the coupling strength index into a dimensionless monotonic sleep state index, and outputting the sleep state index. At the same time, the real-time change trend of the sleep state index is fed back to S3 to optimize the iterative correction logic of the coupling strength index. The branch decision mechanism selects a nonlinear evolution function according to the coupling interference level corresponding to the coupling strength index value. The nonlinear evolution function makes the sleep state index and the coupling strength index monotonically increasing, and different coupling interference levels correspond to different nonlinear evolution paths.

[0028] S5. Based on the sleep state index and raw EEG data, dynamically identify the microstates of brainwave rhythms and adapt sleep-aid guidance parameters. Based on the sleep state index output by S4 and the raw EEG data in the original input data set output by S1, the threshold of the EEG microstate matching algorithm is dynamically adjusted to achieve accurate identification of EEG rhythm microstates. At the same time, sleep aid guidance parameters are dynamically adapted according to the sleep state index and EEG microstate matching results, and the EEG microstate matching results and dynamically adapted sleep aid guidance parameters are output. The identification results and guidance effects are fed back to S1 to optimize the sampling frequency of signal acquisition. The EEG microstate matching results include microstate type, duration, and transition frequency, and the sleep aid guidance parameters include audio frequency and light intensity.

[0029] S6. Verify the effect based on the recognition results and guidance parameters. If the preset effect is not achieved, optimize the process parameters and execute the process repeatedly. Based on the EEG microstate matching results output by S5 and the sleep aid guidance parameters, the accuracy of EEG microstate matching and the adaptation effect of sleep aid guidance are verified. If the preset effect is not achieved, the process parameters are adjusted and S1-S5 are re-executed. If the preset effect is achieved, the optimized process parameters, the EEG microstate matching accuracy report, the sleep aid guidance effect report, and the final sleep state index and sleep aid guidance scheme are output. The preset effect includes an EEG microstate matching accuracy rate ≥90% and a sleep transition period shortening ≥30%. The effect to be verified is obtained by obtaining user sleep state feedback. If the effect is not achieved, the process parameters adjusted include the normalization parameters of S2, the correction range of the coupling function parameters of S3, and the evolution function parameters of S4. Example 2

[0030] Based on Example 1, this example discloses some details of each step.

[0031] In this embodiment, the dual parallel links in step S1 include an environmental signal link and a physiological signal link. The environmental signal link acquires environmental noise signals through an electret microphone and real-time light signals through a light sensor. The physiological signal link acquires electroencephalogram (EEG) physiological signals through an EEG rhythm acquisition electrode. The sampling frequency of the EEG physiological signals is 250Hz, the sampling frequency of the environmental noise signals is 100Hz, and the sampling frequency of the real-time light signals is 10Hz. The acquisition time is 10-30 minutes, which is synchronized with the duration of the sleep transition period. The acquired raw data is stored in real time to the data processing module and simultaneously transmitted to S2 and S3 to achieve bidirectional data splitting and multiplexing.

[0032] In this embodiment, the calculation models of the noise-EEG coupling interference index and the light-melatonin regulation abnormality index in step S2 are constructed based on preset weight coefficients and correction coefficients. The weight coefficients include α = 0.6 and β = 0.4, and the correction coefficient k = 100. The normalization process normalizes the noise-EEG coupling interference index to the value range of [0, 998], which is unified with the value range of the light-melatonin regulation abnormality index. At the same time, the real-time difference |NECI - LMRI| between the two indexes is calculated in real time. The two preprocessed indexes and their real-time difference are synchronously transmitted to S3.

[0033] In this embodiment, the preset scenario adaptation threshold in step S3 is calibrated according to the non-steady interference characteristics of the home bedroom during the sleep transition period. The environmental coupling interference is divided into three levels: the coupling intensity index CI ≤ 100 is weak coupling interference, 100 < CI ≤ 300 is medium coupling interference, and CI > 300 is strong coupling interference. The coupling function parameter is the coupling intensity correction coefficient kc, with an initial value of kc = 0.002 and a correction range of 0.001 - 0.003. By iteratively correcting the kc parameter, the coupling intensity index value is stabilized within a reasonable range that matches the actual coupling interference level.

[0034] In this embodiment, the non-linear evolution path in step S4 is set according to the coupling interference level. Specifically: when the coupling interference is weak (CI ≤ 100), the non-linear evolution function EI = ln(1 + CI) is used to achieve a slow evolution of the sleep state index, adapting to the precise evaluation when the interference is small; when the coupling interference is medium (100 < CI ≤ 300), the non-linear evolution function EI = √CI is used to achieve a moderate evolution of the sleep state index, balancing the evaluation accuracy and calculation efficiency; when the coupling interference is strong (CI > 300), the non-linear evolution function EI = 0.01CI is used to achieve a rapid evolution of the sleep state index, adapting to the real-time response when the interference is strong. The sleep state index EI is a dimensionless number, with a value range of [0, +∞), and EI and CI show a strictly monotonic increasing relationship. The larger the EI value, the greater the difficulty of matching the EEG microstate and the worse the adaptability of the sleep aid guidance.

[0035] In this embodiment, the dynamic adjustment rule for the EEG microstate matching threshold in step S5 is as follows: the smaller the sleep state index (EI) value, the more lenient the recognition threshold, thus improving the recognition speed of EEG rhythm microstates; the larger the EI value, the more stringent the recognition threshold, thus improving the recognition accuracy of EEG rhythm microstates and achieving accurate recognition of A and B type EEG rhythm microstates. The dynamic adaptation rule for the sleep aid guidance parameters is as follows: the smaller the EI value, the gentler the guidance parameters, using low-frequency, low-volume audio guidance and low-intensity, warm-toned lighting guidance; the larger the EI value, the stronger the anti-interference ability of the guidance parameters, increasing the volume and frequency of audio guidance and reducing the intensity of lighting guidance to counteract the influence of environmental coupling interference, thus achieving accurate matching between the sleep aid guidance parameters and the intensity of environmental coupling interference and the user's real-time sleep state.

[0036] In this embodiment, the user's sleep state feedback in step S6 includes the proportion of the EEG rhythm theta waves and the user's turning frequency. The proportion of the EEG rhythm theta waves is acquired in real time through a dedicated device to reflect the true state of the user's EEG rhythm microstate. The user's turning frequency is acquired through a motion sensor to reflect the user's physical sleep state. The two types of feedback are combined to comprehensively verify the accuracy of EEG microstate matching and the effectiveness of sleep-aid guidance. If the preset effect is not achieved, the normalization parameter of S2, the kc correction range of S3, and the evolution function parameter of S4 are adjusted in sequence, and the process of S1-S5 is restarted until the preset effect is achieved. Example 3

[0037] Based on Examples 1 and 2, this embodiment discloses some of the feasible calculation processes.

[0038] In this embodiment, the calculation model for the Noise-Brain Coupling Interference Index (NECI) in step S2 is as follows: In the formula: The ambient noise amplitude is expressed in decibels (dB). The amplitude of the electroencephalogram (EEG) signal is expressed in microvolts (μV). Noise power spectral density, in units of ; Power spectral density of brain electrophysiological signals, in units of ; These are weighting coefficients, dimensionless, with values ​​of 0.6 and 0.4 respectively. This is a correction factor, dimensionless, with a value of 100.

[0039] The calculation model for the Light-Melatonin Regulation Abnormality Index (LMRI) is as follows: In the formula: The actual amount of melatonin secreted is expressed in pmol / L. Normal melatonin secretion under light-free conditions, expressed in pmol / L, is denoted as . ; Real-time light intensity, in lux. The coefficient of inhibition of melatonin secretion by light exposure is expressed in units of... , which are constants for experimental calibration; This is a correction factor, dimensionless, with a value of 100.

[0040] In this embodiment, the original NECI value is normalized and mapped to the interval [0, 998], as shown in the formula: In the formula: is the normalized noise-EEG coupling interference index, dimensionless, with a value range of [0, 998]; To obtain the minimum measured value of NECI in the scenario, it is of the same dimension as the original NECI; To obtain the maximum measured value of NECI in the scenario, it is of the same dimension as the original NECI.

[0041] The real-time difference is calculated as follows: In the formula: The real-time difference between the two normalized exponents is dimensionless.

[0042] In this embodiment, the coupling strength index (CI) calculation model in step S3 is as follows: In the formula: This is a coupling strength correction coefficient, dimensionless, with an initial value of 0.002 and a correction range of [0.001, 0.003]. The normalized noise-EEG coupling interference index is dimensionless. The light-melatonin regulation abnormality index is dimensionless. This represents the real-time difference between the two indices, which is dimensionless.

[0043] The iterative correction formula is as follows: like (Reasonable range for scenario adaptation) or fluctuation value (Preset fluctuation threshold), then for Perform iterative corrections: In the formula: This is the coupling strength correction coefficient for the nth iteration, which is dimensionless. This is the coupling strength correction coefficient for the (n+1)th iteration, which is dimensionless. To correct the step size, it is dimensionless and takes a value of 0.0001; The direction of correction is: If the step size is too large, decrease the step size; if it is too small, increase the step size.

[0044] Iteration termination condition: and At this point, CI is the stable coupling strength index value.

[0045] In this embodiment, the nonlinear evolution function of the branch decision in step S4 is as follows: Weak coupling interference ( ): Medium coupling interference ( ): Strong coupling interference ( ): EI stands for Sleep Attention Index.

[0046] In this embodiment, the EEG microstate matching in step S5 adopts a threshold method, and the core identification threshold is the judgment threshold of the EEG rhythm feature value. Its dynamic adjustment formula is: In the formula: The threshold for EEG rhythm recognition is dynamically adjusted and has the same dimensions as the EEG rhythm characteristic value (e.g., μV, Hz). The basic identification threshold is a calibration value under interference-free conditions, and has the same dimension as the EEG rhythm feature value. is the threshold adjustment coefficient, dimensionless, and is the experimental calibration value (range [0.2, 0.8]). This is a real-time sleep state index, dimensionless; This is the maximum calibration value of EI in the given scenario, and it is dimensionless.

[0047] Adjustment logic: The smaller, The smaller, The closer The more lenient the recognition threshold; The larger, The larger the value, the stricter the recognition threshold.

[0048] The formulas for dynamically adapting sleep-inducing parameters are as follows: (1) Audio frequency adaptation: In the formula, is The audio frequency is dynamically adapted and is measured in Hertz (Hz). Based on the basic audio frequency, the soothing and sleep-inducing frequency under dimensionless interference (calibrated as 432Hz). This is the audio frequency adjustment factor, in units of... (Dimensionless), calibrated to 0.2~0.5; This is a dimensionless index representing the state of falling asleep.

[0049] (2) Audio volume adaptation: In the formula: Dynamically adapted audio volume, in decibels (dB). The base audio volume is calibrated to 30dB. Audio volume adjustment factor, unit: (Dimensionless), calibrated to 0.5~1.0; Sleep state index, dimensionless.

[0050] (3) Light intensity matching: In the formula: Dynamically adapted light intensity, measured in lux. The base light intensity is calibrated to 8 lux; Light intensity adjustment factor, unit: (Dimensionless), calibrated to 0.1~0.3; Sleep state index, dimensionless.

[0051] (4) Lighting color tone (color temperature) matching: In the formula: Dynamically adapted light color temperature, measured in Kelvin (K). The base light color temperature is calibrated to 3000K; Light color temperature adjustment coefficient, unit: (Dimensionless), calibrated to 10~20; Sleep state index, dimensionless.

[0052] In this embodiment, the accuracy of EEG microstate matching is calculated in step S6 as follows: In the formula: EEG microstate matching accuracy, dimensionless, expressed as a percentage; The number of misidentified EEG rhythm microstate samples, dimensionless; The total number of EEG rhythm microstate samples involved in the identification is dimensionless.

[0053] Judgment criteria: To achieve the desired effect.

[0054] The percentage reduction in the sleep transition period is calculated as follows: In the formula: The percentage reduction in the sleep transition period, dimensionless, expressed as a percentage. The actual sleep transition period is measured in minutes (min). The preset sleep transition period duration is in minutes (min), consistent with the standard sleep transition period duration (10-30 min).

[0055] Judgment criteria: To achieve the desired effect.

[0056] This invention forms a multi-iterative feedback closed loop through the above technical means. Specifically, the first-level iterative closed loop, which feeds back the coupling strength index value from step S3 to step S2, is used to optimize the normalization preprocessing accuracy; the second-level iterative closed loop, which feeds back the changing trend of the sleep state evaluation index from step S4 to step S3, is used to optimize the iterative correction logic of the coupling strength index; the third-level iterative closed loop, which feeds back the identification result and guidance effect from step S5 to S1, is used to optimize the sampling frequency of signal acquisition; and the full-process optimization closed loop from step S6 to step S1 is used to dynamically adjust the core parameters of the entire process, ensuring the robustness and stability of the method under non-steady-state coupling interference scenarios.

[0057] The method of this invention is applied to the sleep transition period in a home bedroom, which is a 10-30 minute period of transition from a waking state to light sleep. The environmental disturbance is the coupling interference formed by low-frequency environmental noise with a frequency of 20-200Hz and ambient light fluctuations with an intensity of 0.1-50 lux. It can effectively solve the technical problems of low accuracy of EEG rhythm recognition and delayed guidance adaptation caused by environmental interference coupling in the home bedroom scenario, and break the vicious cycle of "interference-recognition lag-difficulty in falling asleep". Example 4

[0058] Based on any one of the foregoing embodiments 1 to 3, this embodiment continues to disclose implementation details, aiming to reproduce the process and effect of this technical solution.

[0059] S1: Parallel Environmental and Physiological Signal Acquisition The purpose of this step is to simultaneously acquire environmental disturbance signals and electroencephalographic signals during the sleep transition period, solve the signal lag problem caused by traditional serial acquisition, and provide raw data for subsequent feature calculation and interference assessment.

[0060] Data Acquisition Preparation: Based on the parameters of the sleep transition period in a home bedroom, the acquisition duration was set to 20 minutes, synchronized with the sleep transition period; the sampling frequencies of each signal were set as follows: 250Hz for EEG physiological signals, 100Hz for environmental noise signals, and 10Hz for real-time light signals; dual parallel acquisition links were activated, including an electret microphone and a light sensor for the environmental signal acquisition link, and a portable dry electrode EEG acquisition device for the physiological signal acquisition link. All sensors and the data processing module were connected wirelessly to ensure real-time data transmission.

[0061] Signal Acquisition: After the user enters the sleep preparation stage, the acquisition process is initiated. The environmental signal acquisition link collects low-frequency noise signals in the bedroom environment in real time through an electret microphone, and outputs environmental noise amplitude data and noise power spectral density data. The light sensor collects real-time light intensity data in the bedroom, with a light intensity monitoring range of 0.1-50 lux. The physiological signal acquisition link uses a dry electrode EEG acquisition device attached to the user's forehead and temporal lobe to collect the user's EEG physiological signals in real time, focusing on extracting core rhythm signals such as theta waves, and outputting raw EEG physiological signal data.

[0062] Data output and storage: The raw data of the collected electroencephalogram (EEG) physiological signals, ambient noise amplitude data, noise power spectral density data, and real-time light intensity data are integrated into a raw data set and stored in real time in the local storage unit of the data processing module. At the same time, the raw data set is synchronously transmitted to steps S2 and S3 to realize bidirectional data splitting and reuse, providing a data foundation for the subsequent steps of the present invention.

[0063] The innovation of this step lies in the use of dual parallel acquisition links to ensure the synchronous acquisition of environmental interference signals and EEG physiological signals, avoiding coupling interference and evaluation distortion caused by acquisition lag. At the same time, the sampling frequency is set differently to reduce the amount of data processing while ensuring data accuracy and adapting to the computing power of embedded microprocessors.

[0064] S2: Parallel calculation and coupling preprocessing of noise-EEG coupling interference index and light-melatonin regulation abnormality index The purpose of this step is to calculate two coupling interference indices in parallel based on the original acquired data of S1 and perform normalization preprocessing to eliminate the difference in dimensions, thus laying the foundation for the calculation of the core feature coupling strength index.

[0065] Index Calculation Model Loading: The data processing module loads the preset Noise-Electroencephalogram Coupling Interference Index (NECI) and Light-Melatonin Regulation Abnormality Index (LMRI) calculation models. The core parameters of the models include weighting coefficients α=0.6, β=0.4, and correction coefficient k=100. The NECI model is calculated based on raw EEG physiological signal data and environmental noise-related data, characterizing the interference intensity of low-frequency noise on EEG microstate characteristics. The LMRI model is calculated based on real-time light intensity data, characterizing the degree of inhibition of melatonin secretion by light fluctuations, and thus reflecting the impact on the sleep transition period.

[0066] Dual-exponential parallel computation: The data processing module calls two independent computation threads to run the NECI and LMRI computation models in parallel. The raw acquisition data set output by S1 is substituted into the two models respectively to calculate the raw values ​​of NECI and LMRI in real time. The raw value of LMRI is in the range of [0, 998], while the range of raw value of NECI fluctuates due to different noise intensities and is inconsistent with the dimensions of raw value of LMRI.

[0067] Coupling preprocessing: The original NECI values ​​are normalized using a linear normalization algorithm to a range of [0, 998], ensuring consistency with the range of LMRI values ​​and eliminating the difference in the ranges of the two indices, thus resolving the problem of unbalanced dimensional weights. Simultaneously, the data processing module calculates the real-time difference between the normalized NECI and LMRI values, |NECI-LMRI|, in real time, reflecting the degree of synchronous change between the two indices.

[0068] Data output: The preprocessed NECI, LMRI, and real-time difference |NECI-LMRI| are integrated into an index preprocessed data set and synchronously transmitted to step S3 to provide accurate index data for the calculation of the coupling strength index.

[0069] The innovation of this step lies in adopting a dual-exponential parallel computing mode to improve the efficiency of feature calculation and avoid the latency caused by traditional serial calculation. At the same time, through normalized coupling preprocessing, the dimensionality problem in the construction of core features is solved in advance, which provides a guarantee for the accurate calculation of the coupling strength index in the future.

[0070] S3: Dynamic Calculation and Iterative Correction of Environmental Disturbance Coupling Strength Index during Sleep Transition Period The purpose of this step is to calculate the coupling strength index (CI) of the core feature of the sleep transition period based on the exponential preprocessing data of S2, and to correct the coupling function parameters through an iterative feedback mechanism to ensure that CI can accurately characterize the coupling interference strength of low-frequency noise and light fluctuations, while forming an iterative feedback closed loop to S2.

[0071] CI Model Loading and Initial Calculation: The data processing module loads the preset CI mathematical definition formula, substitutes the preprocessed NECI, LMRI, and |NECI - LMRI| output from S2 into the formula, and at the same time loads the coupling function parameter, that is, the initial value of the coupling strength correction coefficient kc, kc = 0.002, and calculates the initial value of CI.

[0072] Scene Adaptation Threshold Verification: The data processing module calls the preset scene adaptation threshold, which is calibrated according to the characteristics of non - steady - state interference in a home bedroom. Specifically: CI ≤ 100 is weak - coupling interference, 100 < CI ≤ 300 is medium - coupling interference, CI > 300 is strong - coupling interference; The initial value of CI is compared with the scene adaptation threshold to determine whether it is within a reasonable range that matches the actual environmental interference. At the same time, the fluctuation amplitude of the initial value of CI is monitored to determine whether there is a large fluctuation caused by non - steady - state interference.

[0073] Iterative Correction of Coupling Function Parameters: If the initial value of CI exceeds the reasonable range or fluctuates too much, the data processing module performs dynamic correction within the kc correction range of 0.001 - 0.003, with a step size of 0.0001 for each correction. After correction, it is substituted back into the CI formula to calculate the new CI value, and the threshold verification is performed again. The above process is repeated until the CI value is stable within the reasonable range, and the stable CI value and the corrected kc parameter are obtained.

[0074] Data Output and Feedback: The stable CI value and the corrected kc parameter are transmitted to step S4 to provide a basis for calculating the sleep state evaluation index; at the same time, the stable CI value is fed back to step S2. Step S2 optimizes the parameters of the normalization algorithm according to the size of the CI value, improves the normalization pre - processing accuracy of NECI and LMRI, and forms a first - level iterative feedback closed - loop from S3 to S2.

[0075] The innovation of this step lies in introducing a dynamic iterative feedback correction mechanism, breaking the traditional fixed mode of "calculating once and outputting", and iteratively adjusting the kc parameter by real - time verifying the rationality of the CI value. This not only solves the accuracy problem of coupling function design but also improves the robustness of CI in non - steady - state interference scenarios, ensuring that CI can stably and accurately represent the coupling interference state.

[0076] S4: Branch Decision - Making and Non - linear Evolution of CI → EI [[ID=!7]]The purpose of this step is to select a non - linear evolution path through a branch decision - making mechanism based on the stable CI value of S3, convert CI into a dimensionless and monotonic sleep state evaluation index (EI), and at the same time form an iterative feedback closed - loop to S3.

[0077] Coupling Interference Level Determination: The data processing module determines the current environmental coupling interference level based on the stable CI value output by S3 and the scene adaptation threshold. In this embodiment, the stable CI value calculated by S3 is 250, which is determined to be moderate coupling interference (100). <CI≤300)。

[0078] Branch decision selection evolution path: The data processing module loads the preset branch decision mechanism and selects the corresponding nonlinear evolution function according to the coupling interference level. In this embodiment, for medium coupling interference, the nonlinear evolution function with EI=√CI is selected; for weak coupling interference (CI≤100), the evolution function with EI=ln(1+CI) is selected; for strong coupling interference (CI>300), the evolution function with EI=0.01CI is selected.

[0079] EI Calculation and Characteristic Verification: A stable CI value is substituted into a selected nonlinear evolution function to calculate the Sleep Attention Assessment Index (EI) value. In this embodiment... EI is a dimensionless number with a value range of [0, +∞). The data processing module verifies the monotonically increasing relationship between EI and CI to ensure that the larger CI is, the larger EI is, and that the EI value can truly reflect the difficulty of EEG microstate recognition and the suitability of sleep-aid guidance.

[0080] Data output and feedback: The calculated EI value is transmitted to step S5 to provide a basis for EEG microstate recognition and sleep aid guidance adaptation; at the same time, the real-time change trend of the EI value is fed back to step S3. Step S3 optimizes the logic of CI iterative correction based on the change trend of EI. For example, when the change of EI is large, the correction step size of kc is reduced to improve the stability of CI, forming a two-level iterative feedback closed loop from S4 to S3.

[0081] The innovation of this step lies in the adoption of a branch decision mechanism to match the corresponding nonlinear evolution path for different coupling interference intensities. This ensures the monotonicity of EI and adapts to the evaluation requirements of different interference intensities. It guarantees evaluation accuracy under weak interference and improves response speed under strong interference, thus solving the problem that traditional linear evolution paths cannot adapt to the dynamic changes of coupling interference.

[0082] S5: EEG Microstate Recognition and Dynamic Adaptation for Sleep Aid Guidance The purpose of this step is to dynamically adjust the EEG microstate matching threshold and sleep aid guidance parameters based on the EI value of S4 and the original EEG rhythm data of S1, so as to achieve accurate identification of EEG microstate and dynamic adaptation of sleep aid guidance, and at the same time form an iterative feedback closed loop to S1. This is a key step in achieving the core technical objectives of this invention.

[0083] Dynamic adjustment of EEG microstate matching threshold: The data processing module loads an EEG microstate recognition algorithm. The core of the algorithm is a threshold recognition method based on EEG rhythm characteristics, which can identify two core EEG microstates: type A and type B. The threshold of the recognition algorithm is dynamically adjusted according to the EI value output by S4. The adjustment rule is: the smaller the EI value, the more lenient the recognition threshold, improving the recognition speed; the larger the EI value, the more stringent the recognition threshold, improving the recognition accuracy. In this embodiment, EI≈15.81 is the EI value corresponding to moderate coupling interference. The data processing module adjusts the recognition threshold to a moderately stringent level, balancing recognition accuracy and speed.

[0084] Precise identification of EEG microstates: The raw EEG physiological signal data output by S1 is substituted into the EEG microstate identification algorithm after the threshold is adjusted, and the core features of the EEG microstate are extracted in real time, including microstate type, duration, and transition frequency. In this embodiment, it is identified that the user is currently in a type A EEG microstate, the microstate duration is 8s, and the transition frequency is 0.125Hz. This identification result accurately reflects the user's real-time sleep state. The EEG microstate identification result is transmitted to step S6 and stored in the data processing module.

[0085] Dynamic Adaptation of Sleep-Aid Guidance Parameters: The data processing module dynamically adjusts the sleep-aid guidance parameters based on the EI value and EEG microstate recognition results. The sleep-aid guidance devices include a smart speaker and a smart ceiling light. The guidance parameters include audio frequency, audio volume, light intensity, and light hue. The adjustment rule is: the smaller the EI value, the gentler the guidance parameters; the larger the EI value, the stronger the anti-interference ability of the guidance parameters. In this embodiment, the interference is moderate, with EI≈15.81, and the user is in a type A EEG microstate. The data processing module adjusts the guidance parameters to: audio frequency 432Hz, audio volume 35dB, light intensity 5lux, and light hue warm yellow (2700K). These parameters have a certain degree of anti-interference ability and can gently guide the user to sleep, avoiding user agitation caused by excessively strong parameters.

[0086] Sleep Aid Guidance Execution: The data processing module transmits the dynamically adapted sleep aid guidance parameters to the smart speaker and smart ceiling light via wireless communication. The two devices execute sleep aid guidance operations synchronously according to the parameters. The smart speaker plays soothing audio at 432Hz, and the smart ceiling light is adjusted to a 5 lux, 2700K warm yellow light, realizing coordinated guidance of audio and light.

[0087] Data output and feedback: The EEG microstate recognition results and dynamically adapted sleep-aid guidance parameters are transmitted to step S6 to provide a basis for effect verification; at the same time, the EEG microstate recognition results and sleep-aid guidance effects (such as the changing trend of the user's EEG physiological signals) are fed back to step S1. Step S1 optimizes the sampling frequency of signal acquisition based on the feedback results. For example, when the recognition accuracy is low, the sampling frequency of EEG physiological signals is increased from 250Hz to 300Hz, forming a three-level iterative feedback closed loop from S5 to S1.

[0088] The innovation of this step lies in deeply binding EI values ​​with EEG rhythm recognition and sleep aid guidance, achieving dynamic adaptation of "recognition-guidance-interference assessment". This breaks the traditional "separation of recognition and guidance" model. The recognition threshold and guidance parameters are dynamically adjusted according to the EI value, which not only solves the problem of low accuracy of EEG rhythm recognition under non-steady-state interference, but also achieves precise adaptation of sleep aid guidance, directly resolving the technical bottleneck of the transition period to sleep in the home bedroom.

[0089] S6: Results Verification and Process Optimization The purpose of this step is to verify the accuracy of EEG microstate recognition and the effectiveness of sleep-aid guidance, dynamically optimize process parameters, form a closed loop throughout the process, and ensure the robustness and stability of the method of this invention under non-steady-state coupling interference scenarios. This is a guarantee step for achieving the preset technical goals.

[0090] Effect verification index loading: The data processing module loads preset effect verification indexes, with the core objectives being: EEG microstate recognition accuracy ≥90% and sleep transition period shortening ≥30%; at the same time, it loads the collected indicators of user sleep state feedback, including the proportion of EEG theta waves and the frequency of user turning over. The proportion of EEG theta waves is collected in real time by the EEG acquisition device to reflect the real situation of the user's EEG microstate, and the frequency of user turning over is collected by the body motion sensor to reflect the user's physical sleep state.

[0091] User sleep state feedback collection: During the sleep aid guidance process, the data processing module collects the user's theta wave ratio and turning frequency in real time. In this embodiment, after a 20-minute sleep transition period, the user's theta wave ratio increased from the initial 15% to 45%, and the turning frequency decreased from the initial 5 times / 10 minutes to 1 time / 10 minutes, indicating that the user's sleep state continued to improve and that the user had successfully entered a light sleep state.

[0092] Recognition and guidance effect verification: The data processing module calculates the EEG microstate recognition accuracy rate based on the collected user sleep state feedback and the EEG microstate recognition results output by S5. In this embodiment, the recognition accuracy rate is 94.5%, ≥ the preset target of 90%. Simultaneously, based on the user's actual sleep time, the shortening ratio of the sleep transition period is calculated. In this embodiment, the user's actual sleep time is 13 minutes, which is 7 minutes shorter than the preset 20-minute sleep transition period, representing a shortening ratio of [missing percentage]. If the accuracy is ≥30% of the preset target, the recognition and guidance effect of this embodiment is determined to have reached the preset target.

[0093] Process parameter output and scheme generation: Since the effect achieves the preset goal, the data processing module does not need to adjust the process parameters. It directly stores the core parameters of the entire process in this embodiment, including the normalization parameters of S2, the kc correction range of S3 (0.001-0.003), and the evolution function parameters of S4. At the same time, it generates an EEG microstate recognition accuracy report and a sleep aid guidance effect report. The report includes CI value, EI value, EEG microstate recognition results, sleep aid guidance parameters, and user sleep state feedback data.

[0094] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to any specific implementation. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A method for EEG microstate recognition and sleep-aid guidance during the sleep transition period, characterized in that, Includes the following steps: S1. Simultaneously acquire environmental disturbance signals and EEG physiological signals during the sleep transition period, and output the raw input dataset containing environmental feature data and raw EEG data. S2. Based on the original input dataset, calculate the multi-dimensional environmental interference index and perform normalized coupling preprocessing on it, and calculate the feature difference between the indices. S3. Based on the preprocessed multi-dimensional environmental interference index and the characteristic difference between the indices, calculate the environmental interference coupling strength index, correct the coupling calculation parameters through an iterative feedback mechanism, and output a stable coupling strength index and the corrected coupling calculation parameters. S4. Based on a stable coupling strength index, the corresponding nonlinear evolution function is matched through a branch decision mechanism to transform the coupling strength index into a dimensionless sleep state index that is monotonically correlated with the coupling strength. S5. Based on the sleep state index and the original EEG data, dynamically adjust the EEG microstate matching threshold and complete the EEG microstate matching. At the same time, dynamically adapt the sleep aid guidance parameters according to the sleep state index and the EEG microstate matching result, and output the EEG microstate matching result and the adapted sleep aid guidance parameters. S6. Based on the EEG microstate matching results and the adapted sleep-aid guidance parameters, verify the sleep-aid guidance adaptation effect: If the expected results are not achieved, repeat steps S1 to S5. If the desired effect is achieved, the optimized process parameters, the final sleep state index, and the sleep-aid guidance plan will be output.

2. The method for EEG microstate recognition and sleep-aid guidance during the sleep transition period according to claim 1, characterized in that, In step S1, the environmental disturbance signal and the electroencephalogram (EEG) signal are acquired using dual parallel links, and the acquisition duration is synchronized with the duration of the sleep transition period. The environmental disturbance signal includes environmental noise signal and real-time illumination signal; The environmental characteristic data includes environmental noise amplitude, noise power spectral density, and real-time light intensity data; the raw EEG data is the EEG rhythm characteristic data of the sleep transition period.

3. The method for EEG microstate recognition and sleep-aid guidance during the sleep transition period according to claim 1, characterized in that, In step S2, the multidimensional environmental interference index includes the noise-brain-electrophysiological coupling interference index, which characterizes the interference of noise on brain electrophysiological signals, and the light-melatonin regulation abnormality index, which characterizes the effect of light on the regulation of melatonin secretion. The noise-brain electrical coupling interference index is calculated based on the amplitude ratio of noise to brain electrical rhythm, the power spectral density ratio, and preset weighting coefficients and correction coefficients. The light-melatonin regulation abnormality index is calculated based on real-time light intensity, the inhibition coefficient of light on melatonin secretion, and a preset correction coefficient. The normalization coupling preprocessing maps each environmental disturbance index to the same value range, eliminating the differences in value ranges between indices; the characteristic difference between indices is the difference in absolute values ​​of each environmental disturbance index after normalization.

4. The method for EEG microstate recognition and sleep-aid guidance during the sleep transition period according to claim 1, characterized in that, In step S3, the environmental interference coupling strength index is the product of the coupling correction coefficient and the sum of the preprocessed multidimensional environmental interference index and the characteristic differences between the indices. The iterative feedback mechanism includes: calculating an initial value of the coupling strength index, comparing it with a preset scene adaptation threshold, and if the initial value exceeds a reasonable range or the fluctuation exceeds a preset threshold, iteratively correcting the coupling calculation parameters and recalculating until the coupling strength index stabilizes within a preset reasonable range.

5. The method for EEG microstate recognition and sleep-aid guidance during the sleep transition period according to claim 4, characterized in that, The preset scene adaptation threshold divides environmental coupling interference into three interference levels; The coupling calculation parameters include coupling correction coefficients, which have preset correction ranges and correction step sizes.

6. The method for EEG microstate recognition and sleep-aid guidance during the sleep transition period according to claim 5, characterized in that, In step S4, the branch decision mechanism selects a matching nonlinear evolution function based on the environmental coupling interference level corresponding to the coupling strength index; different environmental coupling interference levels correspond to different nonlinear evolution paths, and the nonlinear evolution function makes the sleep state index and the coupling strength index monotonically increasing.

7. The method for EEG microstate recognition and sleep-aid guidance during the sleep transition period according to claim 6, characterized in that, The environmental coupling interference levels include weak coupling, medium coupling, and strong coupling interference. Weak coupling interference corresponds to logarithmic nonlinear evolution functions, medium coupling interference corresponds to square root nonlinear evolution functions, and strong coupling interference corresponds to linear scalar multiplication nonlinear evolution functions.

8. The method for EEG microstate recognition and sleep-aid guidance for the sleep transition period according to claim 1, characterized in that, In step S5, the dynamic adjustment rule for the EEG microstate matching threshold is: the smaller the sleep state index value, the more lenient the recognition threshold; the larger the sleep state index value, the more stringent the recognition threshold. The EEG microstate matching results include microstate type, duration, and transition frequency; The dynamic adaptation rule of the sleep aid guidance parameters is as follows: the smaller the sleep state index value, the gentler the guidance parameters; the larger the sleep state index value, the stronger the anti-interference ability of the guidance parameters. The sleep aid guidance parameters include audio parameters and light parameters. The audio parameters include audio frequency and audio volume, and the light parameters include light intensity and light color temperature.

9. The method for EEG microstate recognition and sleep-aid guidance for the sleep transition period according to claim 1, characterized in that, The expected effects mentioned in step S6 include: the accuracy rate of EEG microstate matching is not lower than the preset accuracy threshold, and the proportion of shortened sleep transition period is not lower than the preset duration threshold. The effectiveness of sleep aid guidance can be verified by obtaining feedback on the user's sleep state, which includes the proportion of EEG rhythm characteristics and the frequency of body movements.

10. The method for EEG microstate recognition and sleep-aid guidance for the sleep transition period according to any one of claims 1 to 9, characterized in that, This method is applicable to indoor sleep scenarios where there is environmental coupling interference; the sleep transition period is the time period from wakefulness to light sleep; the environmental disturbance is the coupling interference formed by low-frequency environmental noise and ambient light fluctuations.

Images