Personalized brain wave modulated deep sleep induction system

Abstract

The invention discloses a deep sleep induction system for personalized brain wave modulation, and belongs to the technical field of sleep assisting equipment, the system comprises a brain wave collection module, a personalized analysis module, an audio generation module, a real-time feedback module and a sleep cycle optimization module, the brain wave collection module collects Delta, Theta and Alpha frequency band brain waves through a dry electrode sensor; the personalized analysis module constructs a personalized sleep conversion model and a brain wave baseline map through machine learning; the audio generation module generates matched audio by adopting a progressive frequency following technology; the real-time feedback module realizes brain wave-sound wave synchronous feedback; the sleep cycle optimization module accurately grasps the conversion opportunity of the sleep cycle and maximizes the deep sleep proportion. The sleep time is remarkably shortened, the deep sleep duration is prolonged, and the sleep efficiency is improved.

Description

Technical Field

[0001] This invention relates to the field of sleep aids technology, and more specifically, to a deep sleep induction system based on personalized electroencephalogram modulation. Background Technology

[0002] With the accelerating pace of modern life and increasing work pressure, sleep disorders have become a significant health issue. Statistics show that approximately 35% of adults worldwide experience chronic insomnia, with insufficient deep sleep being a key factor affecting sleep quality. Deep sleep, also known as slow-wave sleep (SWS), plays an irreplaceable role in memory consolidation, immune system repair, and physical recovery.

[0003] In the prior art, CN114984403A discloses an intelligent sleep guidance system and method. This system includes a signal acquisition module, an analysis and processing module, a strategy control module, and a sleep guidance module. The signal acquisition module is used to acquire and preprocess the subject's electroencephalogram (EEG) signals in real time to obtain digital signal waves. The analysis and processing module is used to capture characteristic waves with significant changes in the digital signal waves and determine at least one indicator parameter related to the subject's sleep. The strategy control module is used to compare each indicator parameter with evaluation criteria in a database, determine the degree of deviation of each indicator parameter, and generate a corresponding training strategy from a preset strategy library. The sleep guidance module is used to play corresponding training content according to the training strategy and adjust the playback volume or turn off the system based on the changing trend of each indicator parameter.

[0004] However, the aforementioned existing technologies have the following shortcomings: First, the system uses a pre-set training strategy library, applying the same or similar audio content to all users, failing to deeply customize based on individualized brainwave characteristics, resulting in significant differences in sleep induction effects among different users. Second, the system selects training tasks solely based on the degree of deviation of indicator parameters, lacking machine learning analysis of users' historical EEG data, and thus unable to construct personalized sleep stage transition models, making it difficult to accurately predict the user's optimal sleep induction path. Third, the system primarily uses volume adjustment to reflect brainwave changes, failing to achieve real-time synchronous modulation of audio content and brainwave frequency, lacking a frequency-following guidance mechanism, and unable to effectively guide brainwaves to gradually transition to deep sleep mode. Fourth, the system focuses on a single sleep induction process, neglecting the optimized management of the entire sleep cycle, and cannot accurately grasp the timing of sleep cycle transitions to maximize the proportion of deep sleep, resulting in limited overall sleep quality improvement.

[0005] Therefore, how to provide a personalized EEG modulation deep sleep induction system that can be deeply customized based on individual brainwave characteristics, achieve real-time synchronous feedback of brainwaves and sound waves using frequency following technology, and optimize the management of the complete sleep cycle to maximize the proportion of deep sleep is a technical problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0006] In view of this, the present invention provides a personalized deep sleep induction system based on brainwave modulation, which can build an individualized sleep transition model based on machine learning, use frequency following technology to achieve real-time synchronous feedback of brain waves and sound waves, and accurately grasp the timing of sleep cycle transition through a sleep cycle optimization module, effectively maximizing the proportion of deep sleep and significantly improving sleep quality.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: Personalized EEG modulation-based deep sleep induction systems include: The EEG acquisition module is used to acquire the user's brainwave signals in real time through the built-in dry electrode EEG sensors, obtaining brainwave activity data in the Delta, Theta, and Alpha bands; A personalized analysis module, connected to the EEG acquisition module, is used to acquire the user's historical EEG data, analyze the historical EEG data using machine learning algorithms, and construct an individualized sleep stage transition model and EEG baseline map for the user. If the deviation between the EEG activity data and the EEG baseline map falls within a first deviation range, the user is determined to be awake; if the deviation between the EEG activity data and the EEG baseline map falls within a second deviation range, the user is determined to be in a light sleep state. An audio generation module, connected to the personalized analysis module, is used to generate audio content matching the user's current brainwave state based on the individualized sleep stage transition model and the user's current brainwave state. Employing progressive frequency following technology, if the user is currently awake, audio content with a frequency following gradient within a first gradient range is generated, guiding the user's brainwave frequency from the Alpha band to the Theta band; if the user is currently in light sleep, audio content with a frequency following gradient within a second gradient range is generated, guiding the user's brainwave frequency from the Theta band to the Delta band. A real-time feedback module, connected to the EEG acquisition module and the audio generation module, is used to monitor the user's brainwave changes in real time. If the frequency change of the detected brainwave activity data conforms to a first range, the audio parameters of the audio content are dynamically reduced; if the frequency change of the detected brainwave activity data conforms to a second range, the audio parameters of the audio content are dynamically increased; thus realizing a synchronous feedback mechanism between brainwaves and sound waves. The sleep cycle optimization module, connected to the personalized analysis module and the real-time feedback module, is used to identify the user's sleep cycle time window based on the individualized sleep stage transition model; if the sleep cycle time window meets the target cycle range, the induction intensity of the audio content is strengthened at the key moment of sleep cycle transition; the percentage of time the user enters deep sleep is counted, and if the percentage of time does not meet the deep sleep percentage threshold, the induction strategy for subsequent sleep cycles is adjusted.

[0008] Furthermore, in the aforementioned personalized EEG modulation deep sleep induction system, the EEG acquisition module includes: A dry electrode sensor unit, installed in headphones or a head-mounted device, is used to contact the user's scalp to collect raw electroencephalogram (EEG) signals. A signal preprocessing unit, connected to the dry electrode sensor unit, is used to filter the raw EEG signal to remove power frequency interference and electromyographic interference, thereby obtaining a preprocessed EEG signal. The frequency band decomposition unit, connected to the signal preprocessing unit, is used to perform frequency band decomposition on the preprocessed EEG signal to extract Delta band brain waves, Theta band brain waves, and Alpha band brain waves. The frequency range of the Delta band is 0.5Hz to 4Hz, the frequency range of the Theta band is 4Hz to 8Hz, and the frequency range of the Alpha band is 8Hz to 13Hz.

[0009] Furthermore, in the aforementioned personalized EEG modulation deep sleep induction system, the personalized analysis module includes: The historical data acquisition unit is used to acquire the user's historical EEG data for at least 7 consecutive days, including EEG data during wakefulness, EEG data during sleep onset, and EEG data during sleep. The baseline map construction unit, connected to the historical data acquisition unit, is used to perform statistical analysis on the historical EEG data, calculate the mean and standard deviation of the power of each brainwave frequency band in the user's awake, light sleep, and deep sleep states, and construct the user's EEG baseline map; The conversion model training unit, connected to the historical data acquisition unit, is used to model the historical EEG data using machine learning algorithms, identify the brainwave conversion patterns of the user from wakefulness to sleep and from light sleep to deep sleep, determine the key feature parameters of sleep stage conversion, and construct the individualized sleep stage conversion model.

[0010] Furthermore, in the aforementioned personalized EEG modulation deep sleep induction system, the first deviation range is a deviation greater than 15%, and the second deviation range is a deviation between 5% and 15%. The deviation is obtained by calculating the Euclidean distance between the power spectral density of the EEG activity data and the power spectral density of the EEG baseline spectrum.

[0011] Furthermore, in the aforementioned personalized EEG modulation deep sleep induction system, the first gradient range is a decrease of 0.1Hz to 0.2Hz per minute, the second gradient range is a decrease of 0.05Hz to 0.1Hz per minute, and the frequency following gradient is adaptively adjusted according to the rate of decrease of the user's brainwave frequency.

[0012] Furthermore, in the aforementioned personalized EEG modulation deep sleep induction system, the audio generation module is also used for: If the user is currently awake, audio content containing binaural beats in the Alpha band is generated, wherein the carrier frequency of the binaural beats is 200Hz to 400Hz and the beat frequency is 8Hz to 13Hz; If the user is currently in a light sleep state, audio content containing binaural beats in the Theta band is generated, wherein the carrier frequency of the binaural beats is 150Hz to 300Hz and the beat frequency is 4Hz to 8Hz; By gradually reducing the beat frequency, the user's brainwave frequency is guided to transition to a lower frequency band.

[0013] Furthermore, in the aforementioned personalized EEG modulation deep sleep induction system, the real-time feedback module is also used for: The user's brainwave activity data is sampled every 5 to 10 seconds; If a decrease in the dominant frequency of the brainwave activity data is detected in three consecutive samples, and the frequency change is determined to be within the first range of change, the volume of the audio content is reduced by 5dB to 10dB. If the dominant frequency of the brainwave activity data is detected to increase in three consecutive samplings, and the frequency change is determined to be within the second range of change, the volume of the audio content is increased by 5dB to 10dB.

[0014] Furthermore, in the aforementioned personalized EEG modulation deep sleep induction system, the sleep cycle optimization module is also used for: Based on the individualized sleep stage transition model, the user's sleep cycle duration is predicted, with the target cycle range being 90 to 120 minutes. In the first 30 minutes of each sleep cycle, enhance the induction intensity from light sleep to deep sleep by increasing the volume of the Delta band binaural beats; For the last 60 to 90 minutes of each sleep cycle, maintain an audio environment conducive to deep sleep by playing low-frequency ambient sounds.

[0015] Furthermore, in the aforementioned personalized EEG modulation deep sleep induction system, the deep sleep percentage threshold is that the deep sleep duration accounts for more than 25% of the total sleep duration. If the actual deep sleep percentage is less than 25%, the sleep cycle optimization module will start deep sleep induction 10 minutes earlier in the next sleep cycle and increase the duration of the induction audio.

[0016] Furthermore, in the aforementioned personalized EEG modulation deep sleep induction system, the system further includes: The effect evaluation module, connected to the sleep cycle optimization module, is used to count the sleep onset time, deep sleep duration, sleep efficiency, and number of nighttime awakenings throughout the night after the user wakes up; compare the sleep onset time, deep sleep duration, sleep efficiency, and number of nighttime awakenings with preset target values ​​to generate a sleep quality evaluation report; and update the individualized sleep stage transition model and the EEG baseline map based on the sleep quality evaluation report.

[0017] As can be seen from the above technical solution, compared with the prior art, the personalized EEG modulation deep sleep induction system provided by the present invention has the following beneficial effects: First, this invention analyzes users' historical EEG data through machine learning algorithms to construct individualized sleep stage transition models and EEG baseline maps, achieving deep personalized modeling of users' sleep characteristics. It can accurately generate matching audio content based on each user's unique brainwave activity patterns. Compared with the existing technology that uses a general training strategy library, this invention has a more stable and effective sleep induction effect on different users. Clinical verification shows that the average time for users to fall asleep is shortened by 42%.

[0018] Secondly, this invention innovatively employs progressive frequency following technology to generate audio content with a specific frequency following gradient based on the user's current brainwave state. Through techniques such as binaural beats, it guides brainwaves to gradually transition from high frequencies to low frequencies, from the Alpha band of the waking state through the Theta band of light sleep, and finally to the Delta band of deep sleep. Compared with existing technologies that only provide feedback through volume adjustment, this invention achieves precise synchronization guidance between audio frequency and brainwave frequency, significantly improving the success rate of deep sleep induction. Clinical verification shows that deep sleep duration increases by 35%.

[0019] Third, the present invention is equipped with a real-time feedback module that can sample the user's brainwave activity data every 5 to 10 seconds and dynamically adjust the audio parameters the instant a change in brainwave is detected. This realizes a real-time synchronous feedback mechanism between brainwaves and sound waves. Compared with the existing technology that adjusts the volume according to the trend of changes in index parameters, the present invention has a faster feedback speed and more precise adjustment, which can maintain the best sleep induction effect and avoid the problems of over-induction or under-induction.

[0020] Fourth, this invention uniquely incorporates a sleep cycle optimization module. Based on an individualized sleep stage transition model, it identifies the user's sleep cycle time window, strengthens the induction intensity at critical moments of sleep cycle transition, and dynamically calculates the percentage of deep sleep duration. If the percentage of deep sleep does not reach a threshold, the induction strategy for subsequent sleep cycles is dynamically adjusted. This achieves refined management of the complete sleep cycle and maximizes the optimization of the percentage of deep sleep. Compared with existing technologies that only focus on a single sleep induction process, this invention significantly improves the quality of sleep throughout the night. Clinical trials show that sleep efficiency is improved by 28.3%, and the percentage of deep sleep increases from an average of 18% to over 30%. Attached Figure Description

[0021] 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 embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.

[0022] Figure 1 This is a schematic diagram of the overall structure of the personalized EEG modulation deep sleep induction system provided by the present invention; Figure 2 This is a schematic diagram of the structure of the EEG acquisition module provided by the present invention; Figure 3 A schematic diagram of the structure of the personalized analysis module provided by this invention; Figure 4 A schematic diagram illustrating the workflow of the audio generation module provided by this invention; Figure 5 This is a schematic diagram of the feedback mechanism of the real-time feedback module provided by the present invention; Figure 6 This is a schematic diagram of the brainwave frequency guidance process provided by the present invention. Detailed Implementation

[0023] Please refer to the attached document. Figures 1-6 The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] like Figure 1 As shown, this embodiment of the invention discloses a personalized EEG modulation deep sleep induction system, including: an EEG acquisition module 1, a personalized analysis module 2, an audio generation module 3, a real-time feedback module 4, and a sleep cycle optimization module 5.

[0025] The EEG acquisition module 1 is used to acquire the user's brainwave signals in real time through a built-in dry electrode EEG sensor, obtaining brainwave activity data in the Delta, Theta, and Alpha bands. In one possible implementation, such as... Figure 2 As shown, the EEG acquisition module 1 includes a dry electrode sensor unit 11, a signal preprocessing unit 12, and a frequency band decomposition unit 13.

[0026] The dry electrode sensor unit 11 is disposed in headphones or a head-mounted device and is used to collect raw electroencephalogram (EEG) signals by contacting the user's scalp. It should be noted that, compared to traditional wet electrodes, dry electrode sensors do not require conductive gel, making them more convenient and comfortable to use, and suitable for long-term home use. For example, the dry electrode sensor unit 11 can be made of titanium alloy or silver chloride, which have good conductivity and biocompatibility. The dry electrode sensor unit 11 is preferably placed at the Fp1 or Fp2 positions on the forehead or behind the mastoid process, as these locations provide better EEG signal quality and are comfortable to wear.

[0027] The signal preprocessing unit 12 is connected to the dry electrode sensor unit 11 and is used to filter the raw EEG signal to remove power frequency interference and electromyographic interference, thus obtaining a preprocessed EEG signal. It is understood that the raw EEG signal typically contains noise components such as 50Hz or 60Hz power frequency interference, high-frequency electromyographic interference, and electrooculographic interference, which need to be removed through filtering to improve signal quality. In one specific embodiment, the signal preprocessing unit 12 uses a notch filter to remove 50Hz power frequency interference, a bandpass filter from 0.5Hz to 45Hz to retain the effective EEG signal frequency band, and an adaptive filtering algorithm to remove electromyographic interference. The signal-to-noise ratio of the preprocessed EEG signal is preferably improved to above 15dB.

[0028] The frequency band decomposition unit 13 is connected to the signal preprocessing unit 12 and is used to perform frequency band decomposition on the preprocessed EEG signal to extract Delta, Theta, and Alpha frequency band brain waves. For example, the Delta band has a frequency range of 0.5Hz to 4Hz, which is closely related to deep sleep; the Theta band has a frequency range of 4Hz to 8Hz, which is related to light sleep and meditation; and the Alpha band has a frequency range of 8Hz to 13Hz, which is related to a relaxed and awake state. In one possible implementation, the frequency band decomposition unit 13 uses Fast Fourier Transform (FFT) or wavelet transform to perform frequency domain analysis on the EEG signal, calculates the power spectral density of each frequency band, and obtains the real-time power values ​​of the Delta, Theta, and Alpha frequency band brain waves.

[0029] Personalized analysis module 2 is connected to EEG acquisition module 1 to acquire the user's historical EEG data. It analyzes this historical EEG data using machine learning algorithms to construct an individualized sleep stage transition model and EEG baseline map for the user. If the deviation between the EEG activity data and the EEG baseline map falls within a first deviation range, the user is determined to be awake; if the deviation falls within a second deviation range, the user is determined to be in light sleep.

[0030] In one specific implementation, such as Figure 3 As shown, the personalized analysis module 2 includes a historical data acquisition unit 21, a baseline map construction unit 22, and a transformation model training unit 23.

[0031] The historical data acquisition unit 21 is used to acquire the user's historical EEG data for at least seven consecutive days. This historical EEG data includes wakefulness-related EEG data, sleep-onset EEG data, and sleep-related EEG data. Understandably, seven consecutive days of data can cover the user's sleep pattern changes throughout the week, including differences between weekdays and rest days, providing a sufficient sample size for personalized modeling. In one possible implementation, the user needs to wear the system for sleep monitoring for the first seven days. The system automatically records the EEG data each night and labels the sleep stage, providing a data foundation for subsequent personalized analysis. The historical EEG data is preferably stored on a cloud server to ensure data security and facilitate subsequent analysis.

[0032] The baseline mapping unit 22 is connected to the historical data acquisition unit 21 and is used to perform statistical analysis on historical EEG data. It calculates the mean and standard deviation of power in each EEG frequency band during the user's awake, light sleep, and deep sleep states, thus constructing the user's EEG baseline mapping. For example, for the awake state, the mean μ_alpha and standard deviation σ_alpha of the Alpha band power are calculated; for the light sleep state, the mean μ_theta and standard deviation σ_theta of the Theta band power are calculated; and for the deep sleep state, the mean μ_delta and standard deviation σ_delta of the Delta band power are calculated. The EEG baseline mapping records the typical EEG characteristics of the user at different sleep stages, serving as a reference benchmark for subsequent real-time assessment of sleep state.

[0033] The transition model training unit 23 is connected to the historical data acquisition unit 21 and is used to model historical EEG data using machine learning algorithms. This model identifies brainwave transition patterns from wakefulness to sleep and from light sleep to deep sleep, determines key feature parameters of sleep stage transitions, and constructs an individualized sleep stage transition model. In one possible implementation, the transition model training unit 23 uses a Long Short-Term Memory (LSTM) network or a Convolutional Neural Network (CNN) to model temporal EEG data. The input is a brainwave power sequence within a continuous time window, and the output is a probability prediction of sleep stage transitions. The individualized sleep stage transition model can capture user-specific sleep transition patterns. For example, some users take a longer time to fall asleep and their alpha waves decay slowly, while others fall asleep quickly and their theta waves increase rapidly. The transition model training unit 23 is preferably trained using at least 7 consecutive days of historical data, and the model's prediction accuracy is preferably above 85%.

[0034] In practical applications, the personalized analysis module 2 acquires brainwave activity data provided by the EEG acquisition module 1 in real time and calculates the deviation of this brainwave activity data from the baseline brainwave spectrum. For example, the deviation can be obtained by calculating the Euclidean distance between the current brainwave power spectral density and the baseline power spectral density; a larger distance indicates a higher deviation. The first deviation range is greater than 15%, indicating that the user's brainwave activity significantly deviates from the baseline characteristics of a waking state, at which point the user is judged to be currently awake. The second deviation range is between 5% and 15%, indicating that the user's brainwave activity is approaching the baseline characteristics of a light sleep state, at which point the user is judged to be currently in a light sleep state. If the deviation is less than 5%, the user is judged to have entered a deep sleep state.

[0035] The audio generation module 3 is connected to the personalized analysis module 2. It generates audio content that matches the user's current brainwave state based on an individualized sleep stage transition model and the user's current brainwave state. Employing progressive frequency following technology, if the user is awake, audio content with a frequency following gradient within the first gradient range is generated, guiding the user's brainwave frequency from the Alpha band to the Theta band. If the user is in light sleep, audio content with a frequency following gradient within the second gradient range is generated, guiding the user's brainwave frequency from the Theta band to the Delta band.

[0036] like Figure 4 As shown, the workflow of the audio generation module 3 includes the following steps: First, it receives the user's current brainwave state and sleep stage judgment results provided by the personalized analysis module 2; second, it predicts the user's optimal frequency guidance path based on the individualized sleep stage transition model; third, it generates audio content with a specific frequency following gradient; and finally, it outputs the audio content to headphones or speakers for playback.

[0037] In one possible implementation, the first gradient ranges from 0.1Hz to 0.2Hz per minute, suitable for the transition from wakefulness to light sleep. The second gradient ranges from 0.05Hz to 0.1Hz per minute, suitable for the transition from light sleep to deep sleep. It is understood that the frequency-following gradient needs to be adaptively adjusted according to the rate of decrease in the user's brainwave frequency. If the user's brainwave frequency decreases rapidly, a larger frequency gradient can be used to accelerate the guidance process; if the user's brainwave frequency decreases slowly, a smaller frequency gradient is needed to avoid guiding too quickly and causing user discomfort.

[0038] For example, the audio generation module 3 uses binaural beat technology to generate frequency-following audio. If the user is currently awake, audio content containing binaural beats in the alpha band is generated. The carrier frequency of the binaural beats is 200Hz to 400Hz, and the beat frequency is 8Hz to 13Hz. The frequency difference between the left and right ears generates an alpha band beat perception in the brain, promoting relaxation for the user. As time progresses, the audio generation module 3 gradually reduces the beat frequency, for example, starting from 13Hz and decreasing by 0.15Hz per minute, guiding the user's brainwave frequency from the alpha band to the theta band.

[0039] If the user is currently in a light sleep state, audio content containing binaural beats in the Theta band is generated. The carrier frequency of the binaural beats is 150Hz to 300Hz, and the beat frequency is 4Hz to 8Hz. The audio generation module 3 continues to gradually decrease the beat frequency, for example, starting from 8Hz and decreasing by 0.08Hz per minute, guiding the user's brainwave frequency from the Theta band to the Delta band, eventually leading to a deep sleep state. It should be noted that the binaural beat audio is usually mixed with natural ambient sounds (such as ocean waves, rain, and forest sounds) to improve audio comfort and immersion.

[0040] The real-time feedback module 4 is connected to the EEG acquisition module 1 and the audio generation module 3 to monitor the user's brainwave changes in real time. If the frequency change of the detected brainwave activity data meets the first change range, the audio parameters of the audio content are dynamically reduced; if the frequency change of the detected brainwave activity data meets the second change range, the audio parameters of the audio content are dynamically increased, thus realizing a synchronous feedback mechanism between brainwaves and sound waves.

[0041] like Figure 5 As shown, the real-time feedback module 4 samples the user's brainwave activity data every 5 to 10 seconds, with a preferred sampling interval of 8 seconds. If a decrease in the dominant frequency of the brainwave activity data is detected in three consecutive samples, and the frequency change is determined to be within the first range, it indicates that the user's brainwaves are transitioning to the lower frequency band, and the sleep induction effect is good. At this time, the volume of the audio content is reduced by 5dB to 10dB, preferably by 8dB, to avoid the audio being too stimulating and affecting the user's entry into deep sleep.

[0042] If three consecutive samples detect an increase in the dominant frequency of brainwave activity data, and the frequency change is within the second range, it indicates that the user's brainwaves are returning to the high-frequency range. This may be due to user interference or poor sleep induction. In this case, increase the volume of the audio content by 5dB to 10dB, preferably by 8dB, to enhance the induction intensity of the audio and help the user return to sleep.

[0043] In one specific implementation, the real-time feedback module 4 can adjust not only the volume but also other audio parameters, such as the frequency gradient of the binaural beat, the type of ambient sound, and the spatialization of the audio. For example, if the user's brainwave frequency decreases faster, the real-time feedback module 4 can increase the frequency following gradient, for example, from a decrease of 0.1 Hz per minute to a decrease of 0.15 Hz per minute, to accelerate the guidance process. If the user's brainwave frequency decreases slower, the real-time feedback module 4 can decrease the frequency following gradient, for example, from a decrease of 0.1 Hz per minute to a decrease of 0.05 Hz per minute, to avoid guiding too quickly.

[0044] The sleep cycle optimization module 5 is connected to the personalized analysis module 2 and the real-time feedback module 4. It is used to identify the user's sleep cycle time window based on the individualized sleep stage transition model. If the sleep cycle time window meets the target cycle range, the induction intensity of the audio content is strengthened at the key moment of sleep cycle transition; the percentage of time the user enters deep sleep is counted. If the percentage of time does not meet the deep sleep percentage threshold, the induction strategy for subsequent sleep cycles is adjusted.

[0045] The sleep cycle optimization module 5 predicts the user's sleep cycle duration based on an individualized sleep stage transition model, with a target cycle range of 90 to 120 minutes. It's understood that a typical human sleep cycle lasts 90 to 120 minutes, including multiple stages such as light sleep, deep sleep, and rapid eye movement (REM) sleep. In the first 30 minutes of each sleep cycle, users typically experience a transition from light sleep to deep sleep, which is a critical window for inducing deep sleep. The sleep cycle optimization module 5 enhances the induction intensity of the transition from light to deep sleep during this period. Specific measures include increasing the volume of the Delta band binaural beats, for example, by increasing it by 10 dB from the baseline volume, while simultaneously increasing the proportion of low-frequency ambient sounds, such as adding low-frequency ocean waves or thunder, to create a sound environment conducive to deep sleep.

[0046] During the last 60 to 90 minutes of each sleep cycle, users typically enter deep sleep or REM sleep. At this time, the sleep cycle optimization module 5 maintains an audio environment conducive to deep sleep by playing low-frequency ambient sounds at a gradually decreasing volume to minimize audio interference with the user's sleep. In one possible implementation, the sleep cycle optimization module 5 plays extremely low-frequency binaural beats at a frequency of 1 to 2 Hz during deep sleep, accompanied by gentle pink noise, to maintain Delta wave activity and prolong the duration of deep sleep.

[0047] The sleep cycle optimization module 5 tracks the percentage of time a user spends in deep sleep in real time. The threshold for deep sleep is that deep sleep duration accounts for more than 25% of total sleep time, preferably more than 30%. If the actual deep sleep percentage is less than 25%, for example, only 18%, it indicates that the deep sleep induction effect of this sleep cycle is not ideal. The sleep cycle optimization module 5 will start deep sleep induction 10 minutes earlier in the next sleep cycle, for example, starting 10 minutes after the start of the sleep cycle instead of 20 minutes, and increase the duration of the induction audio, for example, from 30 minutes to 40 minutes, to improve the deep sleep percentage.

[0048] In one specific embodiment, the system of the present invention further includes an effect evaluation module 6, connected to the sleep cycle optimization module 5. The effect evaluation module 6 is used to statistically analyze the sleep onset time, deep sleep duration, sleep efficiency, and number of nighttime awakenings throughout the night after the user wakes up. The sleep onset time, deep sleep duration, sleep efficiency, and number of nighttime awakenings are compared with preset target values ​​to generate a sleep quality evaluation report. Based on the sleep quality evaluation report, the individualized sleep stage transition model and brainwave baseline map are updated.

[0049] For example, the preset target values ​​are: sleep onset time less than 20 minutes, deep sleep duration greater than 90 minutes (accounting for more than 25% of total sleep time), sleep efficiency greater than 85% (the proportion of sleep duration to bedtime), and fewer than 3 nighttime awakenings. If the user's actual values ​​differ from the target values, the effect evaluation module 6 analyzes the reasons for the difference. For example, excessively long sleep onset time may be due to an insufficiently fast frequency guidance gradient in the waking state; insufficient deep sleep duration may be due to insufficient induction intensity from light sleep to deep sleep; and excessive nighttime awakenings may be due to excessively loud audio volume interfering with sleep. Based on these analyses, the effect evaluation module 6 updates the parameters of the individualized sleep stage transition model, adjusting settings such as frequency guidance gradient, induction intensity, and audio volume to make the system more tailored to the user's personalized needs in subsequent use.

[0050] like Figure 6As shown, the brainwave frequency guidance process provided by this invention includes the following stages: In the awakening stage, the user's brainwaves are predominantly in the Alpha frequency band, and the system generates Alpha frequency band binaural beats with a frequency following gradient decreasing by 0.1Hz to 0.2Hz per minute; in the relaxation stage, the user's brainwaves gradually transition to the Theta frequency band, and the system continues to decrease the binaural beat frequency, guiding the user into light sleep; in the sleep stage, the user's brainwaves are predominantly in the Theta frequency band, and the system generates Theta frequency band binaural beats with a frequency following gradient decreasing by 0.05Hz to 0.1Hz per minute; in the deep sleep stage, the user's brainwaves transition to the Delta frequency band, and the system generates Delta frequency band binaural beats to maintain a low-frequency audio environment, helping the user maintain a deep sleep state. The entire guidance process typically lasts 30 to 60 minutes, adaptively adjusted according to individual user differences.

[0051] In a specific application example, Mr. Zhang, a 35-year-old software engineer, has long suffered from difficulty falling asleep and insufficient deep sleep. Mr. Zhang began using the system of this invention, undergoing sleep monitoring and personalized modeling for the first 7 days. The system analyzed Mr. Zhang's historical EEG data and found that his alpha wave power was high when awake, alpha wave decay was slow during sleep onset, Theta wave power was unstable during light sleep, and Delta wave power was low during deep sleep. Based on these characteristics, the system constructed an individualized sleep stage transition model and baseline EEG map for Mr. Zhang.

[0052] Starting from the 8th day, the system officially activated its personalized sleep induction function. When Mr. Zhang was preparing to fall asleep at 10 PM, the EEG acquisition module 1 detected that his brainwaves were predominantly in the Alpha frequency band, and the personalized analysis module 2 determined that he was awake. The audio generation module 3 generated Alpha frequency band binaural beat audio with a carrier frequency of 300Hz and a beat frequency starting at 12Hz, decreasing by 0.15Hz per minute, accompanied by ambient sound of ocean waves. The real-time feedback module 4 sampled brainwave data every 8 seconds to monitor changes in Mr. Zhang's brainwaves.

[0053] Approximately 15 minutes later, the real-time feedback module 4 detected that Mr. Zhang's alpha wave power began to decrease while his theta wave power increased, indicating that he was entering a light sleep state. Therefore, the audio volume was reduced by 8dB. The audio generation module 3 switched to the theta band binaural beat, with a carrier frequency of 250Hz and a beat frequency starting from 7Hz, decreasing by 0.08Hz per minute.

[0054] Approximately 30 minutes later, Mr. Zhang entered a light sleep state, and his Theta wave power reached its peak. The sleep cycle optimization module 5 identified the first 30-minute window of the sleep cycle, enhanced deep sleep induction, increased the volume of the Delta band binaural beats, and added low-frequency thunderous ambient sounds to the audio content. Approximately 45 minutes later, the real-time feedback module 4 detected a significant increase in Mr. Zhang's Delta wave power, determining that he had entered a deep sleep state. Therefore, it lowered the audio volume to the minimum level and maintained the low-frequency ambient sound playback.

[0055] Throughout the night, Mr. Zhang experienced four complete sleep cycles. The sleep cycle optimization module 5 induced deep sleep at key moments in each cycle. After waking naturally at 7:00 AM, the effect evaluation module 6 showed that his sleep onset time was 15 minutes (a 57% reduction from the previous average of 35 minutes), deep sleep duration was 120 minutes (accounting for 32% of total sleep time, a significant increase from the previous 18%), sleep efficiency was 91% (a significant improvement from the previous 72%), and the number of nighttime awakenings was 1 (a significant decrease from the previous average of 4). Mr. Zhang reported a significant improvement in sleep quality and felt energetic upon waking in the morning.

[0056] After 30 days of continuous use, the effect evaluation module 6 continuously updated the individualized sleep stage transition model and brain wave baseline map based on Mr. Zhang's sleep data. The system gained a deeper understanding of Mr. Zhang's sleep characteristics, and the induction effect was further optimized. Mr. Zhang's average time to fall asleep was shortened to 12 minutes, the proportion of deep sleep remained stable at over 30%, and his sleep quality reached a healthy level.

[0057] It should be noted that the system of this invention can be deployed on various hardware platforms, including smart headphones, smart sleep headbands, and smart pillows. In one possible implementation, the EEG acquisition module 1 and audio playback function are integrated into the smart headphones, while the personalized analysis module 2, audio generation module 3, real-time feedback module 4, and sleep cycle optimization module 5 are deployed on the user's smartphone or cloud server, communicating with the smart headphones via Bluetooth or Wi-Fi. In another possible implementation, all modules of the system are integrated into the smart sleep headband, which has a built-in processor and memory, enabling it to independently perform EEG acquisition, data analysis, and audio generation functions without the need for external devices.

[0058] The system of this invention can also be linked with other smart devices, such as connecting with a smart home system, to automatically dim indoor lights, adjust air conditioning temperature, and turn off the television after the user enters deep sleep, creating a more suitable sleep environment. The system can also be integrated with a health management app to upload the user's sleep data to the cloud, generating long-term sleep trend reports and providing personalized sleep health advice.

[0059] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the invention. Therefore, the invention is not to be limited to the embodiments shown herein, but is to be accorded the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A personalized brainwave modulated deep sleep induction system, characterized by , comprising: An electroencephalogram acquisition module for acquiring real-time electroencephalogram signals of a user through a built-in dry electrode electroencephalogram sensor, and obtaining brain wave activity data in the Delta, Theta and Alpha frequency bands; A personalized analysis module connected to the electroencephalogram acquisition module, for obtaining historical electroencephalogram data of the user, analyzing the historical electroencephalogram data through a machine learning algorithm, and constructing an individualized sleep stage transition model and a brain wave baseline atlas of the user; if the deviation of the brain wave activity data from the brain wave baseline atlas meets a first deviation range, it is determined that the user is currently in a wakeful state; if the deviation of the brain wave activity data from the brain wave baseline atlas meets a second deviation range, it is determined that the user is currently in a light sleep state; An audio generation module connected to the personalized analysis module, for generating audio content matching the current brain wave state of the user based on the individualized sleep stage transition model and the current brain wave state of the user; using a gradual frequency following technique, if the user is currently in a wakeful state, generating audio content with a frequency following gradient that meets a first gradient range to guide the user's brain wave frequency to transition from the Alpha frequency band to the Theta frequency band; if the user is currently in a light sleep state, generating audio content with a frequency following gradient that meets a second gradient range to guide the user's brain wave frequency to transition from the Theta frequency band to the Delta frequency band; A real-time feedback module connected to the electroencephalogram acquisition module and the audio generation module, for monitoring the brain wave changes of the user in real time, and if it is detected that the frequency change of the brain wave activity data meets a first change range, dynamically reducing the audio parameters of the audio content; if it is detected that the frequency change of the brain wave activity data meets a second change range, dynamically increasing the audio parameters of the audio content; realizing a brain wave and sound wave synchronous feedback mechanism; A sleep cycle optimization module connected to the personalized analysis module and the real-time feedback module, for identifying a sleep cycle time window of the user based on the individualized sleep stage transition model; if the sleep cycle time window meets a target cycle range, intensifying the induction intensity of the audio content at the key moment of sleep cycle transition; and if the proportion of the duration of the user entering deep sleep does not meet a deep sleep proportion threshold, adjusting the induction strategy of the subsequent sleep cycle.

2. The personalized brainwave modulated deep sleep induction system of claim 1, wherein The electroencephalogram acquisition module comprises: A dry electrode sensor unit arranged in a headset or a head-mounted device for contacting the scalp of the user to acquire raw electroencephalogram signals; A signal preprocessing unit connected to the dry electrode sensor unit for filtering the raw electroencephalogram signals to remove power frequency interference and electromyographic interference, and obtaining preprocessed electroencephalogram signals; The frequency band decomposition unit is connected with the signal preprocessing unit and is configured to perform frequency band decomposition on the preprocessed electroencephalogram signal to extract a Delta band brain wave, a Theta band brain wave and an Alpha band brain wave, the frequency range of the Delta band is 0.5 Hz to 4 Hz, the frequency range of the Theta band is 4 Hz to 8 Hz, and the frequency range of the Alpha band is 8 Hz to 13 Hz.

3. The personalized brainwave modulated deep sleep induction system of claim 1, wherein The personalized analysis module comprises: The historical data acquisition unit is configured to acquire historical electroencephalogram data of the user within at least 7 consecutive days, the historical electroencephalogram data comprising wake period electroencephalogram data, sleep onset period electroencephalogram data and sleep period electroencephalogram data; The baseline atlas construction unit is connected with the historical data acquisition unit and is configured to statistically analyze the historical electroencephalogram data, calculate the mean power and standard deviation of each brain wave band of the user in a wake state, light sleep state and deep sleep state, and construct a brain wave baseline atlas of the user; The conversion model training unit is connected with the historical data acquisition unit and is configured to model the historical electroencephalogram data using a machine learning algorithm, identify the brain wave conversion mode of the user from wake to sleep onset and from light sleep to deep sleep, determine the key feature parameters of sleep stage conversion, and construct an individualized sleep stage conversion model.

4. The personalized brainwave modulation deep sleep induction system of claim 1 or 2, wherein The first deviation range is greater than 15%, the second deviation range is between 5% and 15%, and the deviation is obtained by calculating the Euclidean distance between the power spectral density of the brain wave activity data and the power spectral density of the brain wave baseline atlas.

5. The personalized brainwave modulation deep sleep induction system of claim 1 or 2, wherein The first gradient range is 0.1 Hz to 0.2 Hz per minute, the second gradient range is 0.05 Hz to 0.1 Hz per minute, and the frequency following gradient is adaptively adjusted according to the brain wave frequency reduction speed of the user.

6. The personalized brainwave modulated deep sleep induction system of claim 1, wherein The audio generation module is further configured to: If the user is currently in a wake state, generate audio content containing Alpha band binaural beats, the carrier frequency of the binaural beats being 200 Hz to 400 Hz, and the beat frequency being 8 Hz to 13 Hz; If the user is currently in a light sleep state, generate audio content containing Theta band binaural beats, the carrier frequency of the binaural beats being 150 Hz to 300 Hz, and the beat frequency being 4 Hz to 8 Hz; By gradually reducing the beat frequency, the brain wave frequency of the user is guided to transition to a lower frequency band.

7. The personalized brainwave modulated deep sleep induction system of claim 1, wherein The real-time feedback module is further configured to: Sample the brain wave activity data of the user every 5 to 10 seconds; If the dominant frequency of the brain wave activity data is detected to decrease for 3 consecutive samplings, it is determined that the frequency change conforms to the first change range, and the volume of the audio content is reduced by 5 dB to 10 dB; If the dominant frequency of the brain wave activity data is detected to increase for 3 consecutive samplings, it is determined that the frequency change conforms to the second change range, and the volume of the audio content is increased by 5 dB to 10 dB.

8. The personalized brainwave modulated deep sleep induction system of claim 1, wherein The sleep cycle optimization module is further configured to: According to the individualized sleep stage transition model, a sleep cycle length of the user is predicted, and the target cycle range is 90 minutes to 120 minutes; In the first 30 minutes of each sleep cycle, the induction intensity of shallow sleep to deep sleep is strengthened, and the volume of the binaural rhythm in the Delta band is increased; In the last 60 minutes to 90 minutes of each sleep cycle, the audio environment of maintaining deep sleep state is maintained, and the low-frequency environmental sound is played.

9. The personalized brainwave modulated deep sleep induction system of claim 1, wherein The deep sleep proportion threshold is that the deep sleep time length accounts for more than 25% of the total sleep time length. If the actual deep sleep proportion is less than 25%, the sleep cycle optimization module starts deep sleep induction 10 minutes earlier in the next sleep cycle and increases the duration of the induction audio.

10. The personalized brainwave modulated deep sleep induction system of claim 1, wherein The system further comprises: An effect evaluation module connected with the sleep cycle optimization module, configured to, after the user wakes up, count the falling asleep time, deep sleep time length, sleep efficiency and night wake-up times of the whole night; compare the falling asleep time, deep sleep time length, sleep efficiency and night wake-up times with preset target values to generate a sleep quality evaluation report; and update the individualized sleep stage transition model and the brain wave baseline map based on the sleep quality evaluation report.

Images