Self-adaptive white noise insomnia intervention system based on real-time electroencephalogram feedback

Abstract

The invention provides a self-adaptive white noise insomnia intervention system based on real-time electroencephalogram feedback, and the system comprises an original signal module which collects the original electroencephalogram time sequence data of a patient through an electroencephalogram collection device; the electroencephalogram feature module is used for performing digital filtering, artifact detection and correction and feature extraction to obtain an electroencephalogram feature vector; the sleep evaluation module is used for defining a sleep stage and calculating sleep latency time, the number of times of awakening at night, total sleep time and a ratio of light sleep to deep sleep to obtain a sleep state evaluation result; the white noise control module configures individualized white noise parameters according to the sleep state evaluation result; and the intervention output module is used for outputting a final audio signal according to the individualized white noise parameter and sending the final audio signal to an earphone or a loudspeaker of the patient. The sleep stage and the insomnia degree can be accurately evaluated, white noise parameters are adaptively adjusted, individualized and dynamically optimized insomnia intervention is achieved, and objectivity, intervention effectiveness, comfort and safety of sleep evaluation are remarkably improved.

Description

Technical Field

[0001] This invention relates to the field of insomnia intervention technology, and in particular to an adaptive white noise insomnia intervention system based on real-time EEG feedback. Background Technology

[0002] In the general population, the incidence of chronic insomnia or insomnia disorder is typically around 10%–20%. Researching insomnia intervention methods aims to improve treatment effectiveness, reduce drug dependence and complications, and improve overall patient prognosis and quality of life, while ensuring safety and feasibility. Existing non-pharmacological treatments include cognitive behavioral therapy, physical therapy and instrumental therapy, traditional Chinese medicine rehabilitation techniques, etc. However, existing methods and technologies have the following limitations: 1) Cognitive behavioral therapy requires specially trained therapists, leading to a shortage of professionals; the treatment course is relatively long, requiring significant time and energy from patients, resulting in poor adherence. 2) Existing technologies do not adequately process EEG signals and have weak anti-spoofing capabilities: some sleep products rely only on peripheral indicators such as movement, heart rate, and respiration, lacking detailed analysis of EEG. Even when using EEG, there may be a lack of systematic artifact detection and correction processes, easily leading to misjudgments. 3) Existing sleep assessments mostly rely on simple stage identification or subjective scales: wearable devices on the market mostly estimate light sleep / deep sleep time through simple algorithms, lacking comprehensive quantitative indicators such as insomnia severity and sleep stability. Summary of the Invention

[0003] To overcome the shortcomings of existing technologies, the purpose of this invention is to provide an adaptive white noise insomnia intervention system based on real-time EEG feedback, which can accurately assess sleep stages and insomnia levels, adaptively adjust white noise parameters, achieve individualized and dynamically optimized insomnia intervention, and significantly improve the objectivity of sleep assessment and the effectiveness, comfort, and safety of intervention.

[0004] To achieve the above objectives, the present invention provides the following solution: an adaptive white noise insomnia intervention system based on real-time EEG feedback, comprising: The raw signal module is used to collect the patient's raw signals using an EEG acquisition device, and then preprocess the raw signals using a hardware bandpass filter to obtain raw EEG time series data. The EEG feature module is used to perform digital filtering, artifact detection and correction, and feature extraction based on the original EEG time series data to obtain EEG feature vectors. The sleep assessment module is used to define sleep stages based on the EEG feature vectors, calculate sleep latency, number of nighttime awakenings, total sleep time, and the ratio of light sleep to deep sleep, and obtain sleep state assessment results. The white noise control module is used to configure various parameters of white noise based on the sleep state assessment results to obtain individualized white noise parameters. The intervention output module is used to generate, shape, adjust and output signals based on the individualized white noise parameters to obtain the final audio signal, and then send the final audio signal to the patient's headphones or speakers through a digital-to-analog converter. The original signal module, EEG feature module, sleep assessment module, white noise control module, and intervention output module are interconnected.

[0005] Optionally, the original signal module includes: A pre-preparation unit is used to select a wearable EEG acquisition device, in which electrodes and a preamplifier circuit are configured; wherein the electrodes are deployed in sleep-related brain regions; The data acquisition unit is used to set the sampling frequency to match the sleep-related EEG rhythm. Based on the sampling frequency, the EEG acquisition device is activated to start data acquisition when the patient is preparing to fall asleep, and the raw signal is obtained. A quality monitoring unit is used to deploy a hardware bandpass filter in the EEG acquisition device, set the cutoff frequency of the hardware bandpass filter to 0.5Hz-40Hz, and then use the hardware bandpass filter to preprocess the original signal to obtain the original EEG time series data.

[0006] Optionally, the EEG feature module includes: The filtering and processing unit is used to process the original EEG time series data using a digital bandpass filter to obtain first data, filter artifacts in the first data using a preset amplitude threshold, and then use independent component analysis to suppress or delete the artifacts, thereby completing signal artifact detection and correction and obtaining standard EEG time series data. The feature extraction unit is used to divide the standard EEG time series data into multiple time windows, and perform frequency domain feature extraction and statistical and rhythm feature extraction based on each time window to obtain the EEG feature vector of each time window.

[0007] Optionally, the feature extraction unit includes: A windowing subunit is used to divide the standard EEG time-series data into multiple time windows of a fixed length; The frequency domain feature extraction subunit is used to perform a fast Fourier transform on the signal within each time window to obtain a frequency domain representation. Based on the frequency domain representation, the power spectral density within the time window is calculated, and the power spectral integral of each frequency band within the time window is calculated. Then, based on the power spectral integral, the power spectral integral ratio between each frequency band is calculated to reflect the degree of wakefulness and drowsiness, thus obtaining frequency domain features. The statistical and rhythmic feature extraction subunit is used to calculate the mean and variance of the EEG signal based on the time window, and to calculate the peak value of the autocorrelation function to evaluate the rhythmicity and the coherence index to reflect the synchronous activity of brain regions, so as to obtain statistical and rhythmic features.

[0008] Optionally, the sleep assessment module includes: The model input unit is used to normalize the EEG feature vectors within each time window, thereby obtaining the input feature vectors. The sleep classification unit is used to divide each time window into the waking stage, light sleep stage, moderate non-rapid eye movement sleep stage, deep sleep stage and rapid eye movement sleep stage using a multi-class classification model, resulting in 5 stage labels. Then, based on the principle of maximum probability, the sleep stage label of each time window is determined to obtain a sleep stage sequence in time series form. The insomnia-related unit is used to calculate the sleep latency, number of nighttime awakenings, total sleep time, and the ratio of light sleep to deep sleep based on the sleep stage sequence, so as to construct a sleep stability score and an insomnia severity score through a normalization function. The assessment results unit is used to integrate the sleep stage sequence, sleep latency, number of nighttime awakenings, total sleep time, ratio of light sleep to deep sleep, sleep stability score, and insomnia severity score to obtain sleep state assessment results.

[0009] Optionally, the white noise control module includes: The white noise template unit is used to define a white noise basic configuration template and define a basic parameter vector for each template. Then, a phase-template mapping function from sleep stage to template is introduced. According to the phase-template mapping function, the initial white noise parameters corresponding to the sleep stage label in the current time window are obtained. The parameter correction unit is used to define an insomnia severity adjustment factor based on the insomnia severity score, define a sleep stability adjustment factor based on the sleep stability score, and then adjust each variable in the initial white noise parameters according to the insomnia severity adjustment factor and the sleep stability adjustment factor to obtain individualized white noise parameters.

[0010] Optionally, the white noise basic configuration template includes a first template, a second template, a third template, and a fourth template. The first template is adapted to the wakefulness stage and the light sleep stage, the second template is adapted to the moderate non-rapid eye movement sleep stage, the third template is adapted to the deep sleep stage, and the fourth template is adapted to the rapid eye movement sleep stage.

[0011] Optionally, the basic parameter vector for each template includes noise type, low-frequency energy weight, mid-frequency energy weight, high-frequency energy weight, volume level, and rhythm modulation degree.

[0012] Optionally, the intervention output module includes: A random noise unit is used to define the sound output sampling frequency and time series according to the noise type, and to generate a basic random noise signal through a standard normal distribution; The spectrum shaping unit is used to perform frequency band energy allocation on the basic random noise signal according to the low-frequency band energy weight, the mid-frequency band energy weight and the high-frequency band energy weight to obtain a spectrum-shaped noise signal; The adjustment and output unit is used to perform global volume control and rhythm modulation on the spectrum-shaping noise signal according to the volume level and the rhythm modulation degree to obtain the final audio signal, and then send the final audio signal to the patient's headphones or speaker through a digital-to-analog converter.

[0013] Optionally, the spectrum shaping unit includes: The frequency domain representation subunit is used to perform a fast Fourier transform on the basic random noise signal to obtain the basic noise spectrum, and to design a piecewise gain function that assigns different weights to different frequency bands to output the frequency gain function. The weighted and reconstructed subunit is used to multiply the basic noise spectrum and the frequency gain function to obtain the frequency domain signal after spectrum shaping, and then perform an inverse fast Fourier transform on the frequency domain signal after spectrum shaping to return to the time domain and obtain the spectrum-shaped noise signal, thus completing the adjustment of the spectrum energy distribution.

[0014] This invention discloses the following technical effects by providing an adaptive white noise insomnia intervention system based on real-time EEG feedback: 1. Upgrading from a single static intervention to individualized and dynamic adaptive intervention: By mapping sleep stages to templates, different spectrum and volume configurations are used for different sleep stages. By superimposing insomnia severity and sleep stability adjustments, differentiated treatment is achieved for different patients, different nights, and even different times of the same night.

[0015] 3. Improve the objectivity and precision of sleep assessment: It does not rely on subjective questionnaires but is based on objective EEG signals. It can distinguish five stages: wakefulness, light sleep, moderate NREM, deep sleep, and REM sleep. The temporal resolution is determined by the window length, allowing for relatively precise assessment. Quantitative indicators such as insomnia severity scores and sleep stability scores support efficacy evaluation and long-term follow-up.

[0016] 4. Improved safety and comfort: Precise control of volume and spectral distribution avoids excessively high volumes or uncomfortable frequency bands. Rhythm modulation allows for smooth and non-abrupt effects, enabling long-term action without disrupting sleep.

[0017] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of the system architecture provided in an embodiment of the present invention; Figure 2 A schematic diagram of the sleep state assessment process provided in an embodiment of the present invention; Figure 3 This is a schematic diagram of the white noise intervention process provided in an embodiment of the present invention. Detailed Implementation

[0020] 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.

[0021] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0022] like Figure 1 As shown, this invention provides an adaptive white noise insomnia intervention system based on real-time EEG feedback, comprising: 1. Original signal module like Figure 2 As shown, this is used to acquire the patient's raw signals using an EEG acquisition device, and then preprocess the raw signals using a hardware bandpass filter to obtain raw EEG time-series data; the raw signal module includes: 1.1 Preparatory Unit This is used to select a wearable EEG acquisition device, in which electrodes and a preamplifier circuit are configured; wherein the wearable EEG acquisition device is, for example, a headband-mounted EEG device or a simplified polysomnography device. The electrodes are deployed in sleep-related brain regions, such as frontal, parietal, and occipital regions.

[0023] 1.2 Data Acquisition Unit The sampling frequency is set to match the sleep-related EEG rhythms, preferably within the range of 128 Hz to 512 Hz, to satisfy the sleep-related EEG rhythms. Based on the sampling frequency, the EEG acquisition device is activated to begin data acquisition when the patient is preparing to fall asleep, obtaining the raw signal.

[0024] 1.3 Quality Control Unit A hardware bandpass filter is deployed in the EEG acquisition device, and the cutoff frequency of the hardware bandpass filter is set to 0.5Hz-40Hz. The original signal is then preprocessed using the hardware bandpass filter to obtain the original EEG time series data.

[0025] 2. EEG Feature Module like Figure 2 As shown, the system is used to perform digital filtering, artifact detection and correction, and feature extraction based on the original EEG time-series data to obtain EEG feature vectors; the EEG feature module includes: 2.1 Filtering and Processing Unit This method is used to process the original EEG time series data using a digital bandpass filter to obtain first data. Artifacts in the first data are filtered using a preset amplitude threshold, and then the artifacts are suppressed or deleted using independent component analysis to complete signal artifact detection and correction, thereby obtaining standard EEG time series data.

[0026] Artifacts mainly include blink artifacts (EOG) and electromyography (EMG) interference. These can be addressed by combining amplitude thresholding and independent component analysis. When the instantaneous amplitude exceeds the amplitude threshold, the point is marked as a suspected artifact.

[0027] 2.2 Feature Extraction Unit This unit is used to divide the standard EEG time-series data into multiple time windows, and to extract frequency domain features and statistical and rhythmic features based on each time window to obtain an EEG feature vector for each time window. The feature extraction unit includes: 2.2.1 Windowed Subunit This is used to divide the standard EEG time-series data into multiple time windows of a fixed length; each time window includes the number of sampling points, the time window length, and the sampling frequency. The starting point of each time window is the point following the ending point of the previous time window.

[0028] 2.2.2 Frequency Domain Feature Extraction Subunit The signal within each time window is subjected to a Fast Fourier Transform to obtain a frequency domain representation. Based on the frequency domain representation, the power spectral density within the time window is calculated, and the power spectral integral of each frequency band within the time window is calculated. Then, based on the power spectral integral, the power spectral integral ratio between each frequency band is calculated to reflect the degree of wakefulness and drowsiness, thus obtaining the frequency domain characteristics.

[0029] Conventional frequency band allocation, for example: Delta wave frequency band: 0.5-4 Hz; Theta wave frequency band: 4-8 Hz; Alpha band: 8–13 Hz; Beta band: 13-30 Hz.

[0030] 2.2.3 Statistical and Rhythmic Feature Extraction Subunit Based on the time window, the average value and variance of the EEG signal are calculated, and the peak value of the autocorrelation function for assessing rhythmicity and the coherence index reflecting the synchronous activity of brain regions are calculated to obtain statistical and rhythmic characteristics.

[0031] 3. Sleep assessment module like Figure 2 As shown, this is used to define sleep stages based on the EEG feature vectors, calculate sleep latency, number of nighttime awakenings, total sleep time, and the ratio of light sleep to deep sleep, to obtain sleep state assessment results; the sleep assessment module includes: 3.1 Model Input Unit This is used to normalize the EEG feature vectors within each time window, resulting in the input feature vector; 3.2 Sleep Classification Unit This method employs a multi-class classification model to divide each time window into five stages: wakefulness, light sleep, moderate non-rapid eye movement (NREM) sleep, deep sleep, and rapid eye movement (REM) sleep. This results in five stage labels and probability vectors for each stage. Based on the maximum probability principle, the sleep stage label for each time window is determined, resulting in a sleep stage sequence in time series form.

[0032] 3.3 Insomnia-related units Based on the sleep stage sequence, the sleep latency, number of nighttime awakenings, total sleep time, and the ratio of light sleep to deep sleep are calculated to construct a sleep stability score and an insomnia severity score through a normalization function.

[0033] 3.4 Evaluation Result Unit This is used to integrate the sleep stage sequence, sleep latency, number of nighttime awakenings, total sleep time, ratio of light sleep to deep sleep, sleep stability score, and insomnia severity score to obtain sleep state assessment results.

[0034] 4. White noise control module like Figure 3 As shown, the white noise control module is used to configure various parameters of white noise based on the sleep state assessment results to obtain individualized white noise parameters; the white noise control module includes: 4.1 White Noise Template Unit This is used to define a basic white noise configuration template and define a basic parameter vector for each template. Then, a phase-to-template mapping function is introduced to map sleep stages to templates. Based on the phase-to-template mapping function, the initial white noise parameters corresponding to the sleep stage label within the current time window are obtained.

[0035] The white noise basic configuration template includes a first template, a second template, a third template, and a fourth template. The first template is adapted to the wakefulness stage and the light sleep stage, the second template is adapted to the moderate non-rapid eye movement sleep stage, the third template is adapted to the deep sleep stage, and the fourth template is adapted to the rapid eye movement sleep stage.

[0036] Each template's basic parameter vector includes noise type, low-frequency energy weight, mid-frequency energy weight, high-frequency energy weight, volume level, and rhythm modulation degree. Noise types include, for example, white noise, pink noise, brown noise, or combinations thereof.

[0037] 4.2 Parameter Correction Unit The system is used to define an insomnia severity adjustment factor based on the insomnia severity score, a sleep stability adjustment factor based on the sleep stability score, and then adjust each variable in the initial white noise parameters according to the insomnia severity adjustment factor and the sleep stability adjustment factor to obtain individualized white noise parameters.

[0038] For example, the insomnia severity modulator factor, with a value close to 1, indicates severe insomnia and requires more "active" white noise intervention; the sleep stability modulator factor, with a value close to 1, indicates relatively stable sleep and a tendency to reduce volume interference during deep sleep.

[0039] 5. Intervention Output Module like Figure 3 As shown, the system is used to generate, spectral shape, adjust, and output a signal based on the individualized white noise parameters to obtain a final audio signal, which is then sent to the patient's headphones or speaker via a digital-to-analog converter; the intervention output module includes: 5.1 Random Noise Unit This is used to define the sound output sampling frequency and time series according to the noise type, and to generate a basic random noise signal through a standard normal distribution; 5.2 Spectrum Shaping Unit The frequency band shaping unit is used to perform frequency band energy allocation on the basic random noise signal according to the low-frequency band energy weight, the mid-frequency band energy weight, and the high-frequency band energy weight to obtain a spectrum-shaped noise signal; the spectrum shaping unit includes: 5.2.1 Frequency Domain Representation Subunit This is used to perform a fast Fourier transform on the basic random noise signal to obtain the basic noise spectrum, and to design a piecewise gain function that assigns different weights to different frequency bands, and outputs a frequency gain function.

[0040] 5.2.2 Weighted and Reconstructed Sub-units The signal is used to multiply the basic noise spectrum and the frequency gain function to obtain the frequency domain signal after spectrum shaping. Then, the frequency domain signal after spectrum shaping is subjected to inverse fast Fourier transform to return to the time domain and obtain the spectrum-shaped noise signal, thus completing the adjustment of the spectrum energy distribution.

[0041] 5.3 Adjustment and Output Unit The system is used to perform global volume control and rhythm modulation on the spectrum-shaping noise signal according to the volume level and the rhythm modulation degree to obtain a final audio signal, which is then sent to the patient's headphones or speaker via a digital-to-analog converter. The rhythm modulation envelope determines the relative change in instantaneous volume.

[0042] Therefore, this invention provides an adaptive white noise insomnia intervention system based on real-time EEG feedback, which can accurately assess sleep stages and insomnia levels, adaptively adjust white noise parameters, and achieve individualized and dynamically optimized insomnia intervention, significantly improving the objectivity of sleep assessment and the effectiveness, comfort, and safety of intervention.

[0043] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. The same or similar parts between the various embodiments can be referred to each other.

[0044] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. An adaptive white noise insomnia intervention system based on real-time EEG feedback, characterized in that, include: The raw signal module is used to collect the patient's raw signals using an EEG acquisition device, and then preprocess the raw signals using a hardware bandpass filter to obtain raw EEG time series data. The EEG feature module is used to perform digital filtering, artifact detection and correction, and feature extraction based on the original EEG time series data to obtain EEG feature vectors. The sleep assessment module is used to define sleep stages based on the EEG feature vectors, calculate sleep latency, number of nighttime awakenings, total sleep time, and the ratio of light sleep to deep sleep, and obtain sleep state assessment results. The white noise control module is used to configure various parameters of white noise based on the sleep state assessment results to obtain individualized white noise parameters. The intervention output module is used to generate, shape, adjust and output signals based on the individualized white noise parameters to obtain the final audio signal, and then send the final audio signal to the patient's headphones or speakers through a digital-to-analog converter. The original signal module, EEG feature module, sleep assessment module, white noise control module, and intervention output module are interconnected.

2. The adaptive white noise insomnia intervention system based on real-time EEG feedback according to claim 1, characterized in that, The original signal module includes: A pre-preparation unit is used to select a wearable EEG acquisition device, in which electrodes and a preamplifier circuit are configured; wherein the electrodes are deployed in sleep-related brain regions; The data acquisition unit is used to set the sampling frequency to match the sleep-related EEG rhythm. Based on the sampling frequency, the EEG acquisition device is activated to start data acquisition when the patient is preparing to fall asleep, and the raw signal is obtained. A quality monitoring unit is used to deploy a hardware bandpass filter in the EEG acquisition device, set the cutoff frequency of the hardware bandpass filter to 0.5Hz-40Hz, and then use the hardware bandpass filter to preprocess the original signal to obtain the original EEG time series data.

3. The adaptive white noise insomnia intervention system based on real-time EEG feedback according to claim 2, characterized in that, The EEG feature module includes: The filtering and processing unit is used to process the original EEG time series data using a digital bandpass filter to obtain first data, filter artifacts in the first data using a preset amplitude threshold, and then use independent component analysis to suppress or delete the artifacts, thereby completing signal artifact detection and correction and obtaining standard EEG time series data. The feature extraction unit is used to divide the standard EEG time series data into multiple time windows, and perform frequency domain feature extraction and statistical and rhythm feature extraction based on each time window to obtain the EEG feature vector of each time window.

4. The adaptive white noise insomnia intervention system based on real-time EEG feedback according to claim 3, characterized in that, The feature extraction unit includes: A windowing subunit is used to divide the standard EEG time-series data into multiple time windows of a fixed length; The frequency domain feature extraction subunit is used to perform a fast Fourier transform on the signal within each time window to obtain a frequency domain representation. Based on the frequency domain representation, the power spectral density within the time window is calculated, and the power spectral integral of each frequency band within the time window is calculated. Then, based on the power spectral integral, the power spectral integral ratio between each frequency band is calculated to reflect the degree of wakefulness and drowsiness, thus obtaining frequency domain features. The statistical and rhythmic feature extraction subunit is used to calculate the mean and variance of the EEG signal based on the time window, and to calculate the peak value of the autocorrelation function to evaluate the rhythmicity and the coherence index to reflect the synchronous activity of brain regions, so as to obtain statistical and rhythmic features.

5. The adaptive white noise insomnia intervention system based on real-time EEG feedback according to claim 4, characterized in that, The sleep assessment module includes: The model input unit is used to normalize the EEG feature vectors within each time window, thereby obtaining the input feature vectors. The sleep classification unit is used to divide each time window into the waking stage, light sleep stage, moderate non-rapid eye movement sleep stage, deep sleep stage and rapid eye movement sleep stage using a multi-class classification model, resulting in 5 stage labels. Then, based on the principle of maximum probability, the sleep stage label of each time window is determined to obtain a sleep stage sequence in time series form. The insomnia-related unit is used to calculate the sleep latency, number of nighttime awakenings, total sleep time, and the ratio of light sleep to deep sleep based on the sleep stage sequence, so as to construct a sleep stability score and an insomnia severity score through a normalization function. The assessment results unit is used to integrate the sleep stage sequence, sleep latency, number of nighttime awakenings, total sleep time, ratio of light sleep to deep sleep, sleep stability score, and insomnia severity score to obtain sleep state assessment results.

6. The adaptive white noise insomnia intervention system based on real-time EEG feedback according to claim 5, characterized in that, The white noise control module includes: The white noise template unit is used to define a white noise basic configuration template and define a basic parameter vector for each template. Then, a phase-template mapping function from sleep stage to template is introduced. According to the phase-template mapping function, the initial white noise parameters corresponding to the sleep stage label in the current time window are obtained. The parameter correction unit is used to define an insomnia severity adjustment factor based on the insomnia severity score, define a sleep stability adjustment factor based on the sleep stability score, and then adjust each variable in the initial white noise parameters according to the insomnia severity adjustment factor and the sleep stability adjustment factor to obtain individualized white noise parameters.

7. The adaptive white noise insomnia intervention system based on real-time EEG feedback according to claim 6, characterized in that, The white noise basic configuration template includes a first template, a second template, a third template, and a fourth template. The first template is adapted to the wakefulness stage and the light sleep stage, the second template is adapted to the moderate non-rapid eye movement sleep stage, the third template is adapted to the deep sleep stage, and the fourth template is adapted to the rapid eye movement sleep stage.

8. The adaptive white noise insomnia intervention system based on real-time EEG feedback according to claim 7, characterized in that, The basic parameter vector for each template includes noise type, low-frequency energy weight, mid-frequency energy weight, high-frequency energy weight, volume level, and rhythm modulation degree.

9. The adaptive white noise insomnia intervention system based on real-time EEG feedback according to claim 8, characterized in that, The intervention output module includes: A random noise unit is used to define the sound output sampling frequency and time series according to the noise type, and to generate a basic random noise signal through a standard normal distribution; The spectrum shaping unit is used to perform frequency band energy allocation on the basic random noise signal according to the low-frequency band energy weight, the mid-frequency band energy weight and the high-frequency band energy weight to obtain a spectrum-shaped noise signal; The adjustment and output unit is used to perform global volume control and rhythm modulation on the spectrum-shaping noise signal according to the volume level and the rhythm modulation degree to obtain the final audio signal, and then send the final audio signal to the patient's headphones or speaker through a digital-to-analog converter.

10. The adaptive white noise insomnia intervention system based on real-time EEG feedback according to claim 9, characterized in that, The spectrum shaping unit includes: The frequency domain representation subunit is used to perform a fast Fourier transform on the basic random noise signal to obtain the basic noise spectrum, and to design a piecewise gain function that assigns different weights to different frequency bands to output the frequency gain function. The weighted and reconstructed subunit is used to multiply the basic noise spectrum and the frequency gain function to obtain the frequency domain signal after spectrum shaping, and then perform an inverse fast Fourier transform on the frequency domain signal after spectrum shaping to return to the time domain and obtain the spectrum-shaped noise signal, thus completing the adjustment of the spectrum energy distribution.

Images