Feature information extraction method and system of sleep state monitoring model

Abstract

The invention relates to the technical field of sleep aiding, in particular to a method and a system for extracting feature information of a sleep state monitoring model. The method comprises the following steps: carrying out multi-mode physiological signal acquisition on a detected object to obtain original sleep monitoring data; performing noise removal and signal enhancement processing on the original sleep monitoring data, and performing signal segmentation to obtain multi-channel sleep signal fragment data; performing feature extraction based on electroencephalogram, eye movement diagram, electromyogram, electrocardiogram and accelerometer signals on the multi-channel sleep signal fragment data to obtain multi-dimensional feature data; and performing ultrasonic feature and acoustic feature extraction on the multichannel sleep signal fragment data to generate acoustic feature data. According to the method, dynamic modeling and trend prediction are carried out on the conversion process of the sleep state, potential sleep disorder events can be accurately recognized, influence evaluation is carried out on the potential sleep disorder events through deep learning, and support is provided for personalized sleep diagnosis.

Description

Technical Field

[0001] The present invention relates to the technical field of sleep assistance, and in particular to a method and system for extracting characteristic information of a sleep state monitoring model. Background Art

[0002] Feature information extraction in the sleep state monitoring model is a key step to achieve accurate monitoring. By extracting rich time domain, frequency domain, time-frequency domain and nonlinear features from a variety of sensor data, the model can comprehensively describe and distinguish different sleep stages and evaluate sleep quality. With the continuous advancement of sensor technology and data processing methods, the accuracy and diversity of feature extraction will be further improved, providing stronger support for personalized health management and precision medicine.

[0003] However, the feature information extraction methods of traditional sleep status monitoring models often have the following problems: During the sleep status monitoring process, physiological signals are easily affected by noise (such as electromagnetic interference, breathing and body movement, etc.), which leads to a decrease in the quality of monitoring data, and then affects the accuracy of feature extraction. How to remove noise and extract effective features in multimodal signals has always been a technical problem. Sleep disorders (such as insomnia, sleep apnea, etc.) often occur during sleep, and different disorders manifest themselves differently on different physiological signals. Therefore, it is difficult for traditional models to achieve real-time identification and monitoring of these events. In addition, the accuracy and efficiency of sleep monitoring cannot meet the needs of real-time health management. Summary of the invention

[0004] Based on this, it is necessary for the present invention to provide a method and system for extracting feature information of a sleep state monitoring model to solve at least one of the above technical problems.

[0005] To achieve the above purpose, a method for extracting feature information of a sleep state monitoring model comprises the following steps:

[0006] Step S1: collecting multimodal physiological signals of the subject to obtain original sleep monitoring data; removing noise and performing signal enhancement processing on the original sleep monitoring data, and performing signal segmentation to obtain multi-channel sleep signal segment data;

[0007] Step S2: extracting features of the multi-channel sleep signal segment data based on electroencephalogram, oculogram, electromyogram, electrocardiogram and accelerometer signals to obtain multi-dimensional feature data; extracting ultrasonic features and acoustic features of the multi-channel sleep signal segment data to generate sound wave feature data;

[0008] Step S3: extracting nonlinear dynamic features according to the multidimensional feature data to generate nonlinear feature data; performing complementary fusion processing on the sound wave feature data through the nonlinear feature data to generate mixed sleep feature data;

[0009] Step S4: performing spatiotemporal evolution processing and event impact assessment on the mixed sleep feature data to generate sleep event assessment data; screening potential sleep disorder events based on the sleep event assessment data, and collecting real-time physiological indicators to generate real-time sleep monitoring indicator data; performing sleep stage fluctuation analysis and physiological disturbance association processing on the multi-channel sleep signal segment data through the real-time sleep monitoring indicator data to generate physiological sleep disturbance association data;

[0010] Step S5: Identify the sleep environment impact area of ​​the sleep stage fluctuation data to generate sleep environment impact area data; perform physiological-environmental impact pattern coupling on the sleep environment impact area data to obtain physiological-environmental coupling pattern data; merge the mixed sleep feature data, physiological sleep disturbance association data and physiological-environmental coupling pattern data into feature information extraction data.

[0011] The present invention comprehensively covers the sleep state of the subject through the collection of multimodal physiological signals (EEG, EOG, EMG, ECG, accelerometer), and enhances the perception of sleep state changes from multiple dimensions. The preprocessing of the original sleep monitoring data eliminates noise and enhances key signals, ensuring the accuracy and robustness of subsequent processing. The segmentation of multi-channel signals enables feature extraction and analysis to be performed on a more fine-grained time period, accurately capturing short-term dynamic changes. The multidimensional feature extraction of EEG, eye movement, electromyography, electrocardiography and body movement can fully reveal the changes in the brain, muscles, heart and body posture during sleep, while acoustic features and ultrasonic features further enhance the detection of abnormal breathing and body movements. Nonlinear dynamic feature extraction can capture complex physiological changes in sleep, and through fusion with acoustic wave features, more accurately identify abnormal events such as sleep apnea. The system reveals the dynamic changes of sleep state through spatiotemporal evolution analysis, evaluates and identifies potential sleep disorder events in real time, and provides dynamic monitoring of important events in sleep. The fluctuation analysis of multi-channel sleep signals is combined with the physiological disturbance association to deeply analyze the association between physiological signal changes and sleep stages. Through physiological-environmental coupling analysis, the impact of external environmental factors such as light, noise, and temperature on sleep quality is revealed, thereby building a more comprehensive sleep state assessment framework. Ultimately, based on the fusion of multi-source signals and multi-dimensional features, the system can achieve a full range of dynamic assessments of individual sleep quality, provide personalized intervention recommendations, and help to more accurately identify and improve potential sleep problems.

[0012] The present invention also provides a feature information extraction system for a sleep state monitoring model, which is used to execute the feature information extraction method for the sleep state monitoring model. The feature information extraction system for the sleep state monitoring model comprises:

[0013] The signal preprocessing module is used to collect multimodal physiological signals of the subject to obtain original sleep monitoring data; remove noise and enhance the signal of the original sleep monitoring data, and perform signal segmentation to obtain multi-channel sleep signal segment data;

[0014] A multi-dimensional feature extraction module is used to extract features of multi-channel sleep signal fragment data based on electroencephalogram, oculogram, electromyogram, electrocardiogram and accelerometer signals to obtain multi-dimensional feature data; and to extract ultrasonic features and acoustic features of multi-channel sleep signal fragment data to generate sound wave feature data;

[0015] A feature fusion module is used to extract nonlinear dynamic features based on multidimensional feature data to generate nonlinear feature data; and to perform complementary fusion processing on the sound wave feature data through the nonlinear feature data to generate mixed sleep feature data;

[0016] The sleep dynamic analysis module is used to perform spatiotemporal evolution processing and event impact assessment on mixed sleep feature data to generate sleep event assessment data; screen potential sleep disorder events based on the sleep event assessment data, collect real-time physiological indicators, and generate real-time sleep monitoring indicator data; perform sleep stage fluctuation analysis and physiological disturbance correlation processing on multi-channel sleep signal fragment data through real-time sleep monitoring indicator data to generate physiological sleep disturbance correlation data;

[0017] The environmental-physiological coupling module is used to identify the sleep environment impact area of ​​the sleep stage fluctuation data and generate sleep environment impact area data; to couple the sleep environment impact area data with the physiological-environment impact pattern to obtain the physiological-environment coupling pattern data; and to merge the mixed sleep feature data, the physiological sleep disturbance association data and the physiological-environment coupling pattern data into feature information extraction data.

[0018] The present invention improves the accuracy of the original signal by noise removal and signal enhancement, and ensures the reliability of feature extraction. Signal segmentation makes the processing of long-term monitoring data more efficient, and multi-channel synchronous processing ensures the time consistency of physiological signals, which is convenient for the fusion of multidimensional features. The extracted multidimensional physiological features comprehensively reflect the sleep state, including information such as EEG, heart rate, eye movement, electromyography and body movement. Combined with ultrasonic and acoustic features, external influences such as breathing and snoring can be detected, enriching the dimensions of sleep assessment. Through the extraction of nonlinear dynamic features, the system improves the sensitivity to complex physiological changes and abnormal events. The fusion of multidimensional features and acoustic wave features generates comprehensive mixed sleep feature data, which improves the comprehensiveness of sleep assessment. Spatiotemporal evolution analysis dynamically monitors changes in sleep state and identifies potential sleep disorders or abnormal events. Real-time physiological monitoring combined with physiological perturbation analysis enables the system to detect physiological abnormalities in a timely manner and adjust monitoring strategies. In addition, the system identifies the impact of environmental factors on sleep, optimizes the sleep environment and personalized intervention strategies through physiological-environmental coupling analysis, and ultimately improves the accuracy and reliability of sleep quality assessment. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Other features, objects and advantages of the present invention will become more apparent from the detailed description of non-limiting embodiments thereof made with reference to the following drawings:

[0020] Figure 1 A schematic diagram of the steps of the feature information extraction method of the sleep state monitoring model of the present invention;

[0021] Figure 2 for Figure 1 Detailed step flow diagram of step S1;

[0022] Figure 3 for Figure 1 Detailed step flow chart of step S2 in FIG. DETAILED DESCRIPTION

[0023] The technical method of the present invention is described clearly and completely below in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0024] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0025] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0026] To achieve this, please refer to Figures 1 to 3 The present invention provides a method for extracting characteristic information of a sleep state monitoring model, the method comprising the following steps:

[0027] Step S1: collecting multimodal physiological signals of the subject to obtain original sleep monitoring data; removing noise and performing signal enhancement processing on the original sleep monitoring data, and performing signal segmentation to obtain multi-channel sleep signal segment data;

[0028] Step S2: extracting features of the multi-channel sleep signal segment data based on electroencephalogram, oculogram, electromyogram, electrocardiogram and accelerometer signals to obtain multi-dimensional feature data; extracting ultrasonic features and acoustic features of the multi-channel sleep signal segment data to generate sound wave feature data;

[0029] Step S3: extracting nonlinear dynamic features according to the multidimensional feature data to generate nonlinear feature data; performing complementary fusion processing on the sound wave feature data through the nonlinear feature data to generate mixed sleep feature data;

[0030] Step S4: performing spatiotemporal evolution processing and event impact assessment on the mixed sleep feature data to generate sleep event assessment data; screening potential sleep disorder events based on the sleep event assessment data, and collecting real-time physiological indicators to generate real-time sleep monitoring indicator data; performing sleep stage fluctuation analysis and physiological disturbance association processing on the multi-channel sleep signal segment data through the real-time sleep monitoring indicator data to generate physiological sleep disturbance association data;

[0031] Step S5: Identify the sleep environment impact area of ​​the sleep stage fluctuation data to generate sleep environment impact area data; perform physiological-environmental impact pattern coupling on the sleep environment impact area data to obtain physiological-environmental coupling pattern data; merge the mixed sleep feature data, physiological sleep disturbance association data and physiological-environmental coupling pattern data into feature information extraction data.

[0032] In the embodiment of the present invention, reference Figure 1 The above is a schematic flow chart of the steps of a method for extracting feature information of a sleep state monitoring model according to the present invention. In this example, the method for extracting feature information of a sleep state monitoring model comprises the following steps:

[0033] Step S1: collecting multimodal physiological signals of the subject to obtain original sleep monitoring data; removing noise and performing signal enhancement processing on the original sleep monitoring data, and performing signal segmentation to obtain multi-channel sleep signal segment data;

[0034] The embodiment of the present invention utilizes a multimodal physiological signal acquisition device, such as a polysomnography monitor, to continuously acquire physiological signals from the subject. These signals include electroencephalogram (EEG), oculogram (EOG), electromyogram (EMG), electrocardiogram (ECG) and accelerometer signals to comprehensively record the physiological state of the subject during sleep. During the acquisition process, the sampling rate of the device is set to 500Hz for EEG and EOG, 1000Hz for EMG and ECG, and 100Hz for accelerometer to ensure high precision and high resolution of the signal. The collected raw sleep monitoring data will be de-noised by a digital filter, and bandpass filtering (e.g., 0.5-50Hz range of EEG signal) will be used to remove environmental noise and power supply interference. Subsequently, the empirical mode decomposition (EMD) method is applied to perform signal enhancement processing to improve the signal-to-noise ratio of the signal. The processed signal is segmented according to a preset time window (e.g., 30 seconds) to generate multi-channel sleep signal segment data for subsequent feature extraction and analysis.

[0035] Step S2: extracting features of the multi-channel sleep signal segment data based on electroencephalogram, oculogram, electromyogram, electrocardiogram and accelerometer signals to obtain multi-dimensional feature data; extracting ultrasonic features and acoustic features of the multi-channel sleep signal segment data to generate sound wave feature data;

[0036] The embodiment of the present invention extracts features of various physiological signals based on the multi-channel sleep signal segment data after signal segmentation. For the electroencephalogram signal, the energy characteristics of the δ (0.5-4Hz), θ (4-8Hz), α (8-13Hz), β (13-30Hz) and γ (30-50Hz) bands are extracted, and the waveform characteristics of the K complex and sleep spindles are detected. The eye movement signal extracts its frequency and amplitude characteristics by identifying fast and slow eye movements. The electromyogram signal evaluates muscle tension by analyzing the intensity and frequency characteristics of myoelectric activity. The electrocardiogram signal uses heart rate variability analysis to extract time domain and frequency domain indicators. The accelerometer signal is used for body movement analysis to extract body movement frequency, intensity and duration and other features. After these multi-dimensional feature data are integrated, they are combined with the tiny body movement features collected by the ultrasonic sensor and the environmental sound features collected by the high-sensitivity microphone to generate comprehensive sound wave feature data. In specific applications, feature extraction algorithms such as wavelet transform and fast Fourier transform (FFT) are applied to signal processing to ensure the accuracy and effectiveness of feature extraction.

[0037] Step S3: extracting nonlinear dynamic features according to the multidimensional feature data to generate nonlinear feature data; performing complementary fusion processing on the sound wave feature data through the nonlinear feature data to generate mixed sleep feature data;

[0038] After obtaining the multidimensional feature data, the embodiment of the present invention uses a nonlinear dynamics method to further extract features. For example, the multidimensional feature data is processed using phase space reconstruction technology, and the maximum Lyapunov exponent is calculated to evaluate the stability of the system; recursive quantitative analysis (RQA) is used to extract the recursive rate and deterministic features; sample entropy and approximate entropy are used to measure the complexity of the signal. In addition, the fluctuation characteristics are extracted by detrending fluctuation analysis and long-range correlation analysis. These nonlinear feature data are complementarily fused with the acoustic feature data, and a feature fusion method based on a deep neural network is usually used to enhance the expressive power and discrimination performance of the features. Finally, mixed sleep feature data is generated, which plays a key role in subsequent sleep event evaluation and disturbance association analysis. In specific applications, the structure of the deep neural network can use a multilayer perceptron (MLP) or a convolutional neural network (CNN), and the network parameters are optimized by cross-validation to achieve the best feature fusion effect.

[0039] Step S4: performing spatiotemporal evolution processing and event impact assessment on the mixed sleep feature data to generate sleep event assessment data; screening potential sleep disorder events based on the sleep event assessment data, and collecting real-time physiological indicators to generate real-time sleep monitoring indicator data; performing sleep stage fluctuation analysis and physiological disturbance association processing on the multi-channel sleep signal segment data through the real-time sleep monitoring indicator data to generate physiological sleep disturbance association data;

[0040] The mixed sleep feature data of the embodiment of the present invention is processed through spatiotemporal evolution, and the spatiotemporal evolution event data of sleep state is generated by using methods such as time series analysis and hidden Markov model (HMM). These data are used to identify potential sleep interruptions or abnormal events, and the impact of events is evaluated by feature importance evaluation methods (such as gradient-based importance scoring) to generate sleep event evaluation data. Then, based on the evaluation data, potential sleep disorder events are screened out, and the acquisition parameters of the multimodal sensor are configured to collect key physiological indicators such as heart rate and respiratory rate in real time to generate sleep real-time monitoring indicator data. By performing sleep stage fluctuation analysis on these real-time monitoring indicator data and multi-channel sleep signal fragment data, the volatility of sleep stages is analyzed by using methods such as dynamic time warping (DTW) algorithm, and physiological disturbances are associated to generate physiological sleep disturbance association data. In a specific application scenario, the acquisition frequency of real-time monitoring indicators is set to once per second for heart rate and respiratory rate to ensure the real-time and accuracy of the data.

[0041] Step S5: Identify the sleep environment impact area of ​​the sleep stage fluctuation data to generate sleep environment impact area data; perform physiological-environmental impact pattern coupling on the sleep environment impact area data to obtain physiological-environmental coupling pattern data; merge the mixed sleep feature data, physiological sleep disturbance association data and physiological-environmental coupling pattern data into feature information extraction data.

[0042] The physiological sleep disturbance associated data of the embodiment of the present invention is used to identify the impact area of ​​the sleep environment, and an algorithm based on spatial clustering (such as DBSCAN) is used to identify the area and generate sleep environment impact area data. Then, the multivariate coupling analysis method is used to couple the physiological data with the environmental data (such as temperature and noise level), establish a physiological-environmental impact model, and generate physiological-environmental coupling model data. In practical applications, the collection of environmental parameters is achieved through smart home devices, the temperature is set between 18-22°C, and the noise level is controlled at 30-40 decibels to optimize sleep quality. Finally, the mixed sleep feature data, physiological sleep disturbance associated data, and physiological-environmental coupling model data are integrated to form comprehensive feature information extraction data. These comprehensive data can be used for further sleep quality assessment, sleep disorder diagnosis, and the formulation of personalized sleep intervention plans. The application scenarios cover fields such as home sleep monitoring, medical diagnosis, and sleep research.

[0043] Preferably, step S1 comprises the following steps:

[0044] Step S11: continuously collecting sleep data of the subject using a polysomnography monitor to obtain sleep physiological data, wherein the sleep physiological data includes electroencephalogram, oculogram, electromyogram, electrocardiogram and accelerometer signals;

[0045] Step S12: collecting body motion signals through an ultrasonic sensor and collecting environmental sounds through a high-sensitivity microphone, thereby obtaining body motion environmental acoustic data;

[0046] Step S13: merging the sleep physiological data and the body motion environment acoustic data into original sleep monitoring data;

[0047] Step S14: performing signal quality assessment on the original sleep monitoring data to obtain signal quality assessment data, wherein the signal quality assessment includes signal-to-noise ratio calculation and electrode detachment detection; performing electrooculography and electromyography artifact removal on the original sleep monitoring data, and removing baseline drift using wavelet transform, thereby obtaining sleep monitoring data;

[0048] Step S15: performing signal enhancement based on empirical mode decomposition on the sleep monitoring data according to the signal quality evaluation data, thereby obtaining enhanced sleep monitoring data;

[0049] Step S16: performing signal segmentation on the enhanced sleep monitoring data to obtain multi-channel sleep signal segment data.

[0050] As an embodiment of the present invention, refer to Figure 2 As shown, Figure 1 Detailed step flow diagram of step S1 in the embodiment of the present invention, step S1 includes the following steps:

[0051] Step S11: continuously collecting sleep data of the subject using a polysomnography monitor to obtain sleep physiological data, wherein the sleep physiological data includes electroencephalogram, oculogram, electromyogram, electrocardiogram and accelerometer signals;

[0052] The embodiment of the present invention uses a polysomnography monitor to continuously collect data from the object under test, and the collected data includes electroencephalogram (EEG), oculogram (EOG), electromyogram (EMG), electrocardiogram (ECG) and accelerometer signals. In specific operations, the electrodes for collecting EEG signals are placed on the scalp according to the international 10-20 electrode placement system, and the electrode contact resistance is controlled within 5kΩ. The sampling rate of EEG and EOG is set to 500Hz, the sampling rate of EMG and ECG is 1000Hz, and the accelerometer signal is recorded at a frequency of 100Hz. During the collection process, ensure that the electrodes are in good contact, and monitor the stability of the signal in real time to avoid artifact interference. The collection process lasts for the entire sleep cycle, recording the physiological data of each sleep stage, and providing a complete sleep physiological basis for subsequent analysis.

[0053] Step S12: collecting body motion signals through an ultrasonic sensor and collecting environmental sounds through a high-sensitivity microphone, thereby obtaining body motion environmental acoustic data;

[0054] In order to more comprehensively evaluate the sleep state, the embodiment of the present invention uses an ultrasonic sensor to collect body motion signals, and combines it with a high-sensitivity microphone to collect environmental sounds. In a specific application scenario, the ultrasonic sensor is installed at the head of the bed or under the mattress, and the operating frequency is set to 40kHz to detect tiny body movements of the subject, such as breathing and turning over. The high-sensitivity microphone is placed in the sleeping environment, 1.5 meters away from the noise source, with a frequency response range of 20Hz to 20kHz, to collect external noise and the subject's snoring, etc. By synchronously recording body motion signals and environmental acoustic data, we can better understand the interference factors in the sleep process, and provide data support for subsequent body movement and environmental impact analysis.

[0055] Step S13: merging the sleep physiological data and the body motion environment acoustic data into original sleep monitoring data;

[0056] The embodiment of the present invention merges the sleep physiological data collected in step S11 and the body movement and environmental acoustic data obtained in step S12 to generate original sleep monitoring data. When merging data, it is necessary to ensure that the timestamps of various signals are synchronized so that the physiological and environmental data of the same time period can be accurately corresponded in subsequent analysis. The data storage format is a multi-channel time series, and all signals are uniformly recorded with a time resolution of 1ms. By integrating body movement and environmental sound information with physiological data, a global sleep monitoring data set is constructed to facilitate subsequent feature extraction and analysis.

[0057] Step S14: performing signal quality assessment on the original sleep monitoring data to obtain signal quality assessment data, wherein the signal quality assessment includes signal-to-noise ratio calculation and electrode detachment detection; performing electrooculography and electromyography artifact removal on the original sleep monitoring data, and removing baseline drift using wavelet transform, thereby obtaining sleep monitoring data;

[0058] The embodiment of the present invention performs signal quality assessment on the original sleep monitoring data, and first calculates the signal-to-noise ratio of each channel signal. In the specific operation, the signal-to-noise ratio is calculated based on the ratio of noise variance and signal power using the background noise level estimation formula in the signal. The ideal signal-to-noise ratio of EEG and EOG signals is usually maintained above 20 dB. In addition, the contact impedance of the EEG electrode is monitored in real time to detect whether the electrode has fallen off. If the impedance suddenly rises by more than 10 kΩ, it is marked as electrode detachment. Then, the adaptive filter method is used to remove EOG and EMG artifacts to ensure the purity of sleep physiological signals. In order to solve the problem of baseline drift, the wavelet transform method is applied to remove it. In particular, for electroencephalogram signals, the db4 wavelet basis is used to process the low-frequency drift, thereby obtaining sleep monitoring data after artifact removal and baseline correction.

[0059] Step S15: performing signal enhancement based on empirical mode decomposition on the sleep monitoring data according to the signal quality evaluation data, thereby obtaining enhanced sleep monitoring data;

[0060] The embodiment of the present invention applies the empirical mode decomposition (EMD) algorithm to perform signal enhancement for sleep monitoring data with low signal-to-noise ratio or containing artifacts based on the signal quality assessment data. First, the original signal is decomposed into a series of intrinsic mode functions (IMFs), and the main components are retained and the noise components are removed through screening and reconstruction. In specific operations, for EEG signals, usually only the first 3-4 order IMFs are retained to capture the main physiological fluctuations during sleep. EMD processing can enhance the details of the signal, especially the weak signal part, improve the usability of the signal, thereby generating enhanced sleep monitoring data, ensuring that the data has higher credibility in subsequent analysis.

[0061] Step S16: performing signal segmentation on the enhanced sleep monitoring data to obtain multi-channel sleep signal segment data.

[0062] After signal enhancement, the embodiment of the present invention performs signal segmentation on the enhanced sleep monitoring data. The segmentation is based on the standard sleep time window, which is usually set to a time period of 30 seconds to ensure that each segment can fully contain a complete physiological change cycle. For accelerometer signals and body motion signals, the time window can be shortened to 10 seconds to capture more subtle movement changes. The segmented data will be marked as multi-channel sleep signal segments and saved by channel type and time order. These segment data are convenient for subsequent feature extraction and sleep state classification analysis. In actual scenarios, these segments will be used as input in machine learning models or deep learning models to predict and classify different sleep stages or states.

[0063] The present invention uses a polysomnography monitor to comprehensively collect electroencephalogram (EEG), oculogram (EOG), electromyogram (EMG), electrocardiogram (ECG) and accelerometer signals, covering multi-dimensional monitoring of the brain, heart, muscle activity and body posture, ensuring all-round perception of the sleep state. Combined with the collection of ultrasonic and environmental sounds, the detection capability of the effects of breathing, body movement and external environment on sleep is further enhanced. By removing artifacts, correcting baseline drift and enhancing signals for the original data, the purity and signal-to-noise ratio of the signal are effectively improved, ensuring the accuracy and robustness of subsequent data analysis. Signal segmentation divides long-term data into short-term segments, so that the system can analyze the sleep characteristics in each time period in a fine-grained manner and capture subtle dynamic changes during sleep. This process not only improves the accuracy of sleep stage recognition, but also helps identify abnormal events such as sleep apnea. The fusion of multidimensional features and acoustic wave features increases the diversity of data. Combined with nonlinear dynamic feature analysis, the system can more accurately capture complex changes in sleep states. Overall, the present invention improves the accuracy and comprehensiveness of sleep state assessment through rich physiological data, diversified feature extraction and fusion methods, and provides a solid foundation for personalized sleep monitoring and intervention.

[0064] Preferably, step S16 comprises the following steps:

[0065] Step S161: segmenting the enhanced sleep monitoring data according to the preset time window data to obtain initial sleep signal segment data;

[0066] The embodiment of the present invention segments the enhanced sleep monitoring data according to a preset time window. The commonly used time window is 30 seconds, which is the standard time period used to classify different sleep stages in sleep research. In a specific implementation, signals such as the electroencephalogram, oculogram, and electromyogram are divided into continuous time segment segments to ensure that each segment contains synchronized data from all channels. In order to reduce the boundary effect, an overlapping window processing method is adopted, and the window overlap rate is set to 50%, that is, half of the data in each time window segment is shared with the previous and next segments to ensure the continuity of the signal characteristics. The initial sleep signal segment data obtained by this processing method will be used for further multi-channel synchronous processing.

[0067] Step S162: performing multi-channel time synchronization and alignment processing on the initial sleep signal segment data to obtain aligned sleep signal segment data;

[0068] The embodiment of the present invention performs multi-channel time synchronization and alignment processing on the initial sleep signal segment data. Since the sampling rates of different devices may be different, the sampling frequencies of each channel must be unified first. For example, the sampling rate of the electroencephalogram may be 500Hz, while the sampling rate of the accelerometer signal is 100Hz. The data of each channel needs to be synchronized to a unified 500Hz through interpolation and resampling. Then, a cross-correlation algorithm is used to align the signals of each channel to ensure that the data points at the same time on the timeline correspond to the same physiological state and environmental factors. The aligned sleep signal segment data will more accurately reflect the coordinated changes of multiple physiological signals during sleep.

[0069] Step S163: aligning the sleep signal segment data to perform data integrity check, identifying and marking missing or distorted data segments, and obtaining integrity check data;

[0070] The embodiment of the present invention performs a data integrity check on the aligned sleep signal segment data, identifies and marks missing or distorted data segments. In specific operations, the continuity of the signal and the voltage amplitude range are checked to detect whether there is data missing or signal distortion. For example, if the voltage amplitude of a certain segment of the EEG signal is abnormal and exceeds the normal physiological range (such as ±100μV), it can be marked as a distorted data segment. For missing data segments, they can be identified by the data blanks on the timeline. These marked missing or distorted data segments will be used for subsequent data repair or processing, and corresponding integrity check data will be generated to ensure the accuracy and reliability of subsequent data analysis.

[0071] Step S164: performing amplitude normalization processing on the aligned sleep signal segment data according to the integrity check data, and performing fast Fourier transform to obtain sleep signal spectrum data;

[0072] The embodiment of the present invention performs amplitude normalization processing on the aligned sleep signal segment data based on the integrity check data. In the specific operation, the signal amplitude range is mapped to between 0 and 1 by normalizing the data of each channel to eliminate the amplitude difference between different channels, so as to facilitate the subsequent frequency domain analysis. Then, the fast Fourier transform (FFT) is used to perform spectral analysis on the standardized data to convert the time domain signal into a frequency domain signal. The specific operating parameters of FFT are: the spectral resolution is set to 0.1Hz, the frequency band range is 0.5Hz to 50Hz, and the spectral information of common slow waves (such as delta waves, 0.5-4Hz) and fast waves (such as alpha waves, 8-13Hz) in sleep is extracted. The generated sleep signal spectrum data will be used to analyze the physiological activity characteristics of different frequency bands during sleep.

[0073] Step S165: performing multi-channel integration and synchronous processing on the sleep signal spectrum data to obtain multi-channel sleep signal segment data, wherein the multi-channel includes electroencephalogram, oculogram, electromyogram, electrocardiogram, accelerometer signal, ultrasonic signal and acoustic signal.

[0074] The embodiment of the present invention performs multi-channel integration and synchronization processing on the sleep signal spectrum data to generate the final multi-channel sleep signal segment data. This step first aligns and integrates the spectrum data of the electroencephalogram (EEG), oculogram (EOG), electromyogram (EMG), electrocardiogram (ECG), accelerometer signal, ultrasonic signal and acoustic signal in the time and frequency domain through a multi-channel synchronization algorithm. In the specific operation, the linear weighted average method is used to fuse the spectrum information of each channel at the same time point to ensure that the correlation between the signals of each channel is maintained. In practical applications, multi-channel data integration can better capture the overall physiological state and sleep behavior of the subject. Ultimately, the obtained multi-channel sleep signal segment data will be used for subsequent feature extraction and sleep stage classification analysis, which can more comprehensively and accurately reflect the sleep quality and potential problems of the subject.

[0075] The present invention divides the enhanced sleep monitoring data into preset time windows to accurately capture sleep activities in each time segment, ensure fine-grained analysis of signals, and improve the efficiency of real-time monitoring and parallel processing. The time synchronization and alignment processing of multi-channel data ensures the time consistency of signals from different sensors and reduces the time deviation in the analysis. Through data integrity check, the system can identify and process abnormal data to ensure the reliability of the analysis results. Amplitude normalization eliminates the dimensional differences between different sensors and enhances the comparability of cross-channel analysis. Fast Fourier transform (FFT) provides frequency domain features to reveal sleep physiological activities in different frequency bands. The integrated analysis of multi-channel signals enhances the ability to capture complex physiological changes, improves the comprehensive understanding of sleep status and the detection accuracy of abnormal events. On the whole, the system achieves accurate assessment of sleep status and effective identification of sleep disorders through high-quality data processing, multi-dimensional feature extraction and spectrum analysis.

[0076] Preferably, step S2 comprises the following steps:

[0077] Step S21: extracting energy features of the δ, θ, α, β and γ bands from the EEG signals in the multi-channel sleep signal segment data, and performing waveform detection based on K complexes and sleep spindles to obtain EEG feature data;

[0078] Step S22: performing fast and slow eye movement identification on the eye movement signal in the multi-channel sleep signal segment data, and extracting the eye movement frequency and amplitude characteristics in the rapid eye movement and non-rapid eye movement states to obtain eye movement feature data;

[0079] Step S23: performing muscle tension analysis based on the electromyographic activity intensity and frequency characteristics on the electromyographic signal in the multi-channel sleep signal segment data to obtain electromyographic feature data;

[0080] Step S24: performing heart rate variability analysis on the electrocardiogram signal in the multi-channel sleep signal segment data, extracting time domain and frequency domain heart rate variability indicators, and obtaining electrocardiogram feature data;

[0081] Step S25: performing body motion analysis on the accelerometer signal in the multi-channel sleep signal segment data, extracting body motion frequency, intensity and duration characteristics, and obtaining body motion feature data;

[0082] Step S26: merging the EEG feature data, the eye movement feature data, the myoelectric feature data, the electrocardiographic feature data, and the body movement feature data into multi-dimensional feature data;

[0083] Step S27: performing Doppler frequency shift analysis on the ultrasonic signal in the multi-channel sleep signal segment data, extracting micro-body motion features, and obtaining ultrasonic feature data;

[0084] Step S28: Detect and classify sound events on the ambient sound signals in the multi-channel sleep signal segment data, extract sleep sound features based on breathing sounds and snoring sounds, and obtain acoustic feature data; merge the ultrasonic feature data and the acoustic feature data into sound wave feature data.

[0085] As an embodiment of the present invention, refer to Figure 3 As shown, Figure 1 Detailed step flow diagram of step S2 in the embodiment of the present invention, step S2 includes the following steps:

[0086] Step S21: extracting energy features of the δ, θ, α, β and γ bands from the EEG signals in the multi-channel sleep signal segment data, and performing waveform detection based on K complexes and sleep spindles to obtain EEG feature data;

[0087] In the embodiment of the present invention, the electroencephalogram (EEG) signal is analyzed in detail in the multi-channel sleep signal segment data to extract the energy characteristics of the δ (0.5-4Hz), θ (4-8Hz), α (8-13Hz), β (13-30Hz) and γ (30-50Hz) bands. In the specific operation, a bandpass filter is first applied to filter out the signals of each frequency band respectively, and then the power spectral density (PSD) of each frequency band is calculated to quantify the energy characteristics. Subsequently, the waveform characteristics of K complexes and sleep spindles are identified using an automatic detection algorithm. K complex detection adopts a template matching-based method to identify by comparing the correlation with a predefined K complex template; sleep spindle detection adopts a wavelet transform combined with a threshold judgment method to capture its unique frequency and duration characteristics. In the specific application scenario, the sampling rate of the EEG signal is set to 500Hz, and the bandwidth of the filter is δ (0.5-4Hz), θ (4-8Hz), α (8-13Hz), β (13-30Hz) and γ (30-50Hz) to ensure the accurate extraction of the energy characteristics of each frequency band. Finally, EEG feature data including the energy value of each frequency band and the detection results of K complex and sleep spindle are generated for subsequent sleep stage classification and analysis.

[0088] Step S22: performing fast and slow eye movement identification on the eye movement signal in the multi-channel sleep signal segment data, and extracting the eye movement frequency and amplitude characteristics in the rapid eye movement and non-rapid eye movement states to obtain eye movement feature data;

[0089] The embodiment of the present invention identifies rapid and slow eye movements in the oculogram (EOG) signal in the multi-channel sleep signal segment data, and extracts the frequency and amplitude characteristics of eye movements in the rapid eye movement (REM) and non-rapid eye movement (NREM) states. In the specific operation, a low-pass filter (such as 30Hz) is first applied to remove high-frequency noise, and then a method based on the time domain and frequency domain is used to identify rapid eye movements and slow eye movements. Rapid eye movements are identified by short-time energy and zero crossing rate analysis, and slow eye movements are detected by signal fluctuations over a long time span. After the identification is completed, the eye movement frequency (times / minute) and amplitude (μV) in each time window are calculated to quantify the eye movement characteristics. In practical applications, the sampling rate of the EOG signal is set to 500Hz, the time window for rapid eye movement detection is 2 seconds, and the time window for slow eye movement detection is 10 seconds. Finally, eye movement feature data including the frequency and amplitude of eye movements in the REM and NREM states are generated for evaluating sleep stages and monitoring sleep abnormalities related to eye movements.

[0090] Step S23: performing muscle tension analysis based on the electromyographic activity intensity and frequency characteristics on the electromyographic signal in the multi-channel sleep signal segment data to obtain electromyographic feature data;

[0091] The embodiment of the present invention performs muscle tension analysis based on the intensity and frequency characteristics of electromyographic activity on the electromyogram (EMG) signal in the multi-channel sleep signal segment data to obtain electromyographic feature data. In the specific operation, the EMG signal is first high-pass filtered (such as 20Hz) to remove the DC component and low-frequency noise, and then the root mean square value (RMS) is applied to calculate the intensity of electromyographic activity, and the frequency domain features such as the main frequency and the energy distribution of the frequency band are extracted by fast Fourier transform (FFT). Then, the threshold judgment method is used to identify changes in muscle tension, such as sudden muscle activity or sustained muscle relaxation. In a specific application scenario, the sampling rate of the EMG signal is set to 1000Hz, the RMS calculation window is 1 second, and the spectral resolution of the FFT is set to 1Hz. Finally, electromyographic feature data including the intensity of electromyographic activity, the main frequency and the frequency domain distribution are generated for analyzing muscle activity during sleep and evaluating muscle-related sleep disorders.

[0092] Step S24: performing heart rate variability analysis on the electrocardiogram signal in the multi-channel sleep signal segment data, extracting time domain and frequency domain heart rate variability indicators, and obtaining electrocardiogram feature data;

[0093] The embodiment of the present invention performs heart rate variability (HRV) analysis on the electrocardiogram (ECG) signal in the multi-channel sleep signal segment data, extracts the heart rate variability index in the time domain and frequency domain, and obtains ECG feature data. In the specific operation, the ECG signal is firstly QRS wave detected, and the heartbeat interval (RR interval) is accurately identified using the Pan-Tompkins algorithm. In the time domain analysis, indicators such as standard deviation (SDNN) and root mean square deviation (RMSSD) are calculated; in the frequency domain analysis, the fast Fourier transform (FFT) or power spectrum density estimation method is applied to extract the low frequency (LF, 0.04-0.15Hz) and high frequency (HF, 0.15-0.40Hz) power and LF / HF ratio. In the specific application scenario, the sampling rate of the ECG signal is set to 1000Hz, and the threshold of the QRS detection is dynamically adjusted according to the heart rate range of the object being measured to ensure accurate identification of the heartbeat. Finally, ECG feature data containing time domain and frequency domain HRV indicators are generated for evaluating the activity of the autonomic nervous system and the health of the heart.

[0094] Step S25: performing body motion analysis on the accelerometer signal in the multi-channel sleep signal segment data, extracting body motion frequency, intensity and duration characteristics, and obtaining body motion feature data;

[0095] The embodiment of the present invention performs body motion analysis on the accelerometer signal in the multi-channel sleep signal segment data, extracts the body motion frequency, intensity and duration characteristics, and obtains body motion feature data. In the specific operation, the accelerometer signal is first low-pass filtered (such as 5Hz) to remove high-frequency noise, and then the amplitude and change rate of the signal are calculated to quantify the body motion intensity. The time domain analysis method is used to identify the frequency and duration of body motion events, such as the number of body movements per minute and the duration of a single body movement. Further, a detection algorithm based on statistical thresholds is applied to distinguish normal body movements from abnormal body movements (such as frequent turning over or violent movements). In a specific application scenario, the sampling rate of the accelerometer signal is set to 100Hz, and the threshold of body motion detection is adjusted individually according to the activity habits of the subject. Finally, body motion feature data including body motion frequency, intensity and duration are generated for monitoring body motion patterns during sleep and identifying abnormal body motion behaviors.

[0096] Step S26: merging the EEG feature data, the eye movement feature data, the myoelectric feature data, the electrocardiographic feature data, and the body movement feature data into multi-dimensional feature data;

[0097] The embodiment of the present invention integrates the EEG feature data, eye movement feature data, electromyographic feature data, electrocardiographic feature data and body movement feature data extracted in steps S21-S25 to form multidimensional feature data. In the specific operation, each type of feature data is first standardized (such as Z-score standardization) to eliminate the dimensional differences between different features. Then, the various types of feature data are arranged in order according to the time window (such as 30 seconds) to form a multidimensional vector containing all feature dimensions. For example, within a 30-second time window, the EEG features include the energy value and waveform detection results of each frequency band, the eye movement features include frequency and amplitude, the electromyographic features include activity intensity and frequency, the electrocardiographic features include HRV indicators, and the body movement features include frequency, intensity and duration. Finally, a comprehensive multidimensional feature data set is generated as an input for subsequent sleep stage classification and abnormality detection to ensure the comprehensive analysis and evaluation of various physiological signals.

[0098] Step S27: performing Doppler frequency shift analysis on the ultrasonic signal in the multi-channel sleep signal segment data, extracting micro-body motion features, and obtaining ultrasonic feature data;

[0099] The embodiment of the present invention performs Doppler frequency shift analysis on the ultrasonic signal in the multi-channel sleep signal segment data to extract the characteristics of micro body movement and obtain ultrasonic feature data. In the specific operation, the ultrasonic signal is first bandpass filtered (such as 20kHz to 40kHz) to remove background noise, and then the Doppler frequency shift algorithm is applied to detect micro movements of the object under test, such as breathing and slight turning over. By comparing the frequency changes of the transmitted and received ultrasonic signals, the speed and direction of the body movement are calculated, thereby quantifying the frequency and amplitude of the micro body movement. In the specific application scenario, the operating frequency of the ultrasonic sensor is set to 40kHz, and the signal sampling rate is 100kHz to ensure high-precision body movement detection. Finally, ultrasonic feature data containing the frequency and amplitude of micro body movements is generated to supplement and enhance the body movement information in traditional physiological signals and improve the accuracy of overall sleep monitoring.

[0100] Step S28: Detect and classify sound events on the ambient sound signals in the multi-channel sleep signal segment data, extract sleep sound features based on breathing sounds and snoring sounds, and obtain acoustic feature data; merge the ultrasonic feature data and the acoustic feature data into sound wave feature data.

[0101] The embodiment of the present invention performs sound event detection and classification on the environmental sound signal in the multi-channel sleep signal segment data, extracts the sleep sound features based on breathing sounds and snoring sounds, and obtains acoustic feature data; then, the ultrasonic feature data in step S27 is merged with the acoustic feature data to generate sound wave feature data. In the specific operation, the environmental sound signal is first preprocessed, including noise reduction and normalization processing, and then a sound event detection algorithm based on machine learning (such as support vector machine SVM or deep neural network DNN) is applied to identify and classify different sound events, such as normal breathing, snoring and environmental noise. For breathing sounds and snoring sounds, their characteristic parameters such as frequency, amplitude and duration are further extracted. In the specific application scenario, the sampling rate of the microphone is set to 44.1kHz, and the sound event classification model is trained with a large amount of sleep environment data to improve the recognition accuracy. Finally, the extracted acoustic feature data is fused with the ultrasonic feature data obtained by Doppler frequency shift analysis, and a method such as feature splicing or weighted averaging is used to generate comprehensive sound wave feature data for further analyzing the impact of the environment on sleep and identifying abnormal sound events related to sleep.

[0102] The present invention can accurately analyze the quality changes of each stage of sleep by extracting the energy characteristics of frequency bands such as δ, θ, α, β, and γ, and identify key events in deep sleep and rapid eye movement (REM) stages by combining K complex and sleep spindle detection. Analysis of eye movement characteristics helps to distinguish REM and NREM stages, and evaluate the depth of sleep, dream activity, and recovery of body functions. The intensity and frequency characteristics of electromyographic activity effectively identify the muscle state of different sleep stages, help determine the degree of muscle relaxation in deep sleep, and detect sleep disorders such as periodic limb movements. The extraction of heart rate variability (HRV) characteristics evaluates the balance of the autonomic nervous system and reveals the activity changes of sympathetic and parasympathetic nerves during sleep. Analysis of body movement characteristics can monitor the number of awakenings and the level of nighttime activity, providing a basis for judging shallow sleep and deep sleep. Multi-channel signal fusion, including EEG, eye movement, electromyography, electrocardiography, and body movement feature data, forms a comprehensive perspective, conducts a comprehensive analysis of physiological changes in sleep, and improves the ability to analyze complex sleep states. Ultrasonic signals are used to detect tiny body movements, especially in deep sleep, and to analyze breathing sounds and snoring to identify breathing disorders, such as sleep apnea syndrome. Acoustic signals can also detect the impact of environmental noise on sleep quality. Through the comprehensive analysis of multiple physiological signals and external interference, the system improves the accuracy of sleep quality assessment and enhances the ability to identify sleep disorders.

[0103] Preferably, step S3 comprises the following steps:

[0104] Step S31: reconstructing the multidimensional characteristic data in phase space to obtain reconstructed phase space data; calculating the maximum Lyapunov exponent based on the reconstructed phase space data to obtain system stability characteristic data;

[0105] The embodiment of the present invention reconstructs the phase space of multidimensional feature data. The reconstruction method is based on Takens' theorem, and the time series data is mapped into a high-dimensional phase space by selecting a suitable time delay and embedding dimension. The time delay is usually determined by an autocorrelation function or a mutual information method, and the embedding dimension can be estimated by a pseudo nearest neighbor algorithm. In practical applications, for the reconstruction of multidimensional feature data including EEG, ECG, body movement, etc., the time delay is set to 10 sampling points and the embedding dimension is 3. After reconstructing the phase space, the maximum Lyapunov exponent is calculated based on the reconstructed phase space data to measure the chaos and stability of the system. The Wolf algorithm is used to calculate the Lyapunov exponent to measure the exponential divergence rate of the trajectory in the phase space. Finally, the system stability characteristic data is obtained, which reflects the dynamic stability characteristics of physiological signals in the sleep state.

[0106] Step S32: performing recursive quantitative analysis on the multi-dimensional feature data, and extracting the recursive rate and deterministic features to obtain recursive feature data;

[0107] The embodiment of the present invention performs recursive quantitative analysis (RQA) on multidimensional feature data. First, the feature data is normalized to adapt it to the construction of the recursive graph. The recursive graph is generated by calculating the distance matrix between the trajectory points in the phase space, and a fixed distance threshold (such as 0.1) is used to determine the recursiveness of the point. Based on the recursive graph, the recursive rate (i.e., the frequency at which the trajectory returns to the previous state) and deterministic features (i.e., the proportion of continuous recursive lines) are extracted, which quantify the complexity and determinism of the system. In actual operation, the distance threshold and the resolution of the recursive graph are adjusted according to the characteristics of the data. Finally, recursive feature data containing the recursive rate and deterministic features are generated for further evaluating the periodicity and stability of the multidimensional feature data.

[0108] Step S33: performing sample entropy and approximate entropy calculations on the multidimensional feature data to obtain entropy feature data; performing detrending fluctuation analysis on the multidimensional feature data and extracting long-range correlation features to obtain fluctuation feature data;

[0109] The embodiment of the present invention calculates sample entropy (SampEn) and approximate entropy (ApEn) for multidimensional feature data. Sample entropy measures the complexity and randomness of data by comparing the similarity of subsequences of different lengths in a time series; approximate entropy is used to detect the predictability of signals. In practical applications, the embedding dimension of sample entropy and approximate entropy is set to 2, and the tolerance is set to 20% of the standard deviation of the data. Subsequently, a detrended fluctuation analysis (DFA) is performed on the multidimensional feature data. This method extracts long-range correlation features by segmenting the time series and calculating the fluctuation amplitude of each segment to reflect the self-similarity of the signal. In the application scenario, the segment length is set to a multiple of 4 seconds to 16 seconds. Finally, feature data containing entropy features and fluctuation features are obtained, which are used to identify the complexity and long-term correlation of physiological signals in the sleep state.

[0110] Step S34: fusing the system stability characteristic data, the recursive characteristic data, the entropy characteristic data and the fluctuation characteristic data to obtain nonlinear characteristic data;

[0111] The embodiment of the present invention performs feature fusion on the system stability feature data, recursive feature data, entropy feature data and fluctuation feature data extracted in steps S31-S33. In the specific operation, all feature data are first standardized to ensure the dimensional consistency of different types of features. Then, different features are combined into a nonlinear feature vector using a dimensionality reduction method such as weighted averaging or principal component analysis (PCA). In practical applications, the number of principal components of PCA is set to a number that can explain more than 90% of the variance to maximize information retention. Finally, the fused nonlinear feature data is generated to reflect the nonlinear dynamic characteristics of the entire system as input for subsequent steps.

[0112] Step S35: performing wavelet packet transformation on the sound wave characteristic data, and extracting time-frequency domain features to obtain time-frequency sound wave characteristic data;

[0113] The embodiment of the present invention performs wavelet packet transform on the acoustic wave feature data extracted in step S28, and decomposes the signal into different time-frequency domains. First, a suitable mother wavelet (such as Daubechies wavelet) is selected, and the number of decomposition layers (generally 3 layers) is determined, and then the original signal is decomposed into sub-signals of different frequency bands by multi-level wavelet packet decomposition. Each sub-signal extracts its time domain and frequency domain features, such as energy, frequency center, and time-frequency energy distribution. In the actual application scenario, the sampling rate of the acoustic wave signal is 44.1kHz, the number of layers of wavelet packet decomposition is 3 layers, and the extracted time-frequency features include the energy proportion and frequency change trend of each frequency band. Finally, the time-frequency acoustic wave feature data is generated to accurately reflect the changing law of the acoustic wave signal in different time and frequency dimensions.

[0114] Step S36: using the nonlinear characteristic data to perform characteristic weighting on the time-frequency sound wave characteristic data to obtain weighted sound wave characteristic data;

[0115] The embodiment of the present invention utilizes the nonlinear feature data in step S34 to perform feature weighting on the time-frequency sound wave feature data in step S35. In the specific operation, the nonlinear feature data and the time-frequency sound wave feature data are first feature matched to ensure their alignment in the time dimension. Then, a feature weighting algorithm (such as a weighted average method based on correlation) is applied to highlight the sound wave features that are sensitive to sleep state by applying the weights of the nonlinear features to the time-frequency sound wave features. In actual application scenarios, the weight calculation uses a linear regression model to dynamically adjust the weights according to the correlation between the nonlinear features and the time-frequency features. Finally, weighted sound wave feature data is generated to enhance the sound wave signal features related to changes in sleep state.

[0116] Step S37: performing feature fusion based on a deep neural network on the nonlinear feature data and the weighted sound wave feature data to obtain mixed sleep feature data.

[0117] The embodiment of the present invention performs feature fusion based on deep neural network (DNN) on the nonlinear feature data in step S34 and the weighted acoustic wave feature data in step S36. First, a deep neural network model is constructed, which includes an input layer, multiple hidden layers and an output layer. The number of neurons in the hidden layer is set according to the dimension and complexity of the feature data (such as 128 neurons per layer). The input layer receives the nonlinear features and weighted acoustic wave features respectively, and performs feature fusion through the nonlinear activation function (such as ReLU) of the hidden layer. The model is trained by the back propagation algorithm, the loss function uses the mean square error (MSE), and the optimizer selects Adam. In practical applications, the training data is divided into a training set and a validation set by cross-validation. The learning rate is set to 0.001 and the number of iterations is 1000 during the training process. Finally, the fused mixed sleep feature data is generated to more accurately describe the sleep state and its related physiological and environmental characteristics, and provide input for subsequent sleep quality assessment and abnormality detection.

[0118] The present invention reveals the complex dynamic behavior during sleep through phase space reconstruction, and uses the maximum Lyapunov exponent to measure the stability of the sleep system, especially in detecting sleep disorders such as apnea. Recursive quantitative analysis (RQA) effectively identifies repeated sleep stages and abnormal events, such as body movements and apnea, and enhances the analysis of the stability of sleep structure. Nonlinear features such as sample entropy and approximate entropy quantify the complexity of physiological signals, and detrended fluctuation analysis reveals long-range dependencies and identifies potential sleep problems. Wavelet packet transform extracts the time-frequency features of breathing sounds and snoring, which helps to detect sleep breathing disorders at an early stage. Combining nonlinear features, time-frequency features and deep neural networks (DNNs) further improves the detection accuracy of sleep events. DNN automatically extracts key patterns in complex data and optimizes the identification and classification of abnormal sleep events. Overall, the present invention provides an efficient and accurate sleep monitoring solution, especially in sleep disorder detection and sleep quality assessment.

[0119] Preferably, step S4 comprises the following steps:

[0120] Step S41: performing a sleep state spatiotemporal evolution process according to the mixed sleep characteristic data to generate sleep state spatiotemporal evolution event data;

[0121] The embodiment of the present invention performs spatiotemporal evolution processing of sleep state based on mixed sleep feature data to generate spatiotemporal evolution event data of sleep state. In the specific operation, the mixed sleep feature data is first input into the spatiotemporal evolution model, which adopts deep learning architectures such as long short-term memory network (LSTM) or graph convolution network (GCN) to capture the dynamic changes of sleep state in time and space. The model identifies the transition trend and pattern of sleep state by analyzing the feature changes in the continuous time window. For example, in the actual application scenario, the number of hidden layer units of the LSTM network is set to 128, the learning rate is 0.001, and the Adam optimizer is used for training, and the number of iterations is 200 rounds to ensure that the model can effectively capture the temporal characteristics of the sleep state. During the processing, the model generates event data including sleep state transfer, duration and spatial distribution. These data reflect the evolution dynamics of different stages in the sleep process and provide a basis for subsequent abnormal event identification.

[0122] Step S42: identifying potential sleep interruptions or abnormal events in the mixed sleep feature data through the sleep state spatiotemporal evolution event data, and performing event impact assessment based on feature importance to generate sleep event assessment data;

[0123] The embodiment of the present invention uses the spatiotemporal evolution event data of sleep state generated in step S41 to identify potential sleep interruptions or abnormal events in the mixed sleep feature data, and performs event impact assessment based on feature importance to generate sleep event assessment data. In the specific operation, firstly, an abnormal detection algorithm (such as an isolation forest or a support vector machine SVM) is applied to identify events that are significantly different from normal sleep patterns in the spatiotemporal evolution event data, such as frequent sleep interruptions or abnormal sleep stage transitions. Subsequently, an assessment method based on feature importance (such as a SHAP value or a feature importance score) is used to analyze the degree of influence of these abnormal events on the overall sleep quality. For example, in practical applications, the number of trees of the isolation forest algorithm is set to 100, and the maximum number of features is 5 to improve the accuracy and efficiency of detection. Through these methods, sleep event assessment data including the type of abnormal event, the frequency of occurrence, and the degree of influence on sleep quality are generated, providing a basis for further screening and intervention of sleep disorders.

[0124] Step S43: Screening potential sleep disorder events according to the sleep event evaluation data, and collecting real-time physiological indicators to generate real-time sleep monitoring indicator data;

[0125] Embodiment of the present invention: According to the sleep event evaluation data generated in step S42, potential sleep disorder events are screened out, and real-time physiological indicators are collected to generate sleep real-time monitoring indicator data. In the specific operation, the sleep event evaluation data is first clustered (such as K-means or spectral clustering) to identify potential sleep disorder patterns with similar characteristics. Then, a screening threshold is set based on these patterns, for example, the event frequency exceeds 3 times per hour or the score of a specific abnormal event exceeds a certain threshold (such as 0.8) as a screening criterion. After screening out potential disorder events, the acquisition parameters of the multimodal sensor are configured, focusing on monitoring physiological indicators related to these disorders, such as heart rate, respiratory rate and blood oxygen saturation. In a specific application, the real-time acquisition parameters are set to heart rate once per second, respiratory rate twice per second, and blood oxygen saturation once per second. Data is transmitted in real time through wireless sensors, and edge computing devices are used for real-time signal processing and analysis. Finally, real-time sleep monitoring indicator data containing key physiological indicators is generated for dynamic monitoring and timely intervention in sleep disorder events.

[0126] Step S44: performing sleep stage fluctuation analysis on the multi-channel sleep signal segment data to generate sleep stage fluctuation data; and performing physiological disturbance association processing on the sleep stage fluctuation data through the real-time sleep monitoring index data to generate physiological sleep disturbance association data.

[0127] The embodiment of the present invention performs a sleep stage fluctuation analysis on the multi-channel sleep signal segment data to generate sleep stage fluctuation data; and performs physiological disturbance association processing on the sleep stage fluctuation data through the sleep real-time monitoring index data to generate physiological sleep disturbance association data. In the specific operation, the dynamic time warping (DTW) algorithm is first applied to perform fluctuation analysis on the sleep stages in the multi-channel sleep signal segment data to identify the frequent conversion of sleep stages and the change of duration. For example, the window length of DTW is set to 30 seconds, and a time offset within a certain range is allowed to match the sleep stage pattern of different time periods. Then, in combination with the sleep real-time monitoring index data generated in step S43, correlation analysis or causal inference methods (such as Granger causality test) are used to evaluate the correlation between physiological index changes and sleep stage fluctuations. For example, in practical applications, the correlation threshold is set to 0.6, and significantly associated physiological disturbance factors are identified, such as frequent conversion of heart rate sudden changes and REM stages. Through this association processing, physiological sleep disturbance association data containing sleep stage fluctuation patterns and related physiological disturbance factors are generated, providing data support for in-depth understanding of sleep quality and formulating personalized intervention strategies.

[0128] The present invention can monitor the changes in sleep state in real time and capture the dynamic evolution of the entire sleep cycle by performing spatiotemporal evolution processing on mixed sleep feature data. In particular, in the transition process between different sleep stages such as rapid eye movement (REM) and non-rapid eye movement (NREM), spatiotemporal evolution analysis can reveal the duration, frequency and pattern of each stage. Spatiotemporal evolution processing can keenly identify sleep interruptions or small abnormal changes, such as short awakenings or micro-movements. This is particularly useful for identifying phenomena that affect sleep quality, such as frequent micro-awakenings and periodic limb movements at night. Through spatiotemporal evolution event data, potential interruptions or abnormal events that may occur in sleep, such as apnea, sleep interruption, and night awakening, can be automatically identified. Combined with the multidimensional features in the mixed sleep feature data, the time and intensity of sleep events can be more accurately located. Using the feature importance evaluation method, the system can automatically analyze which features have an important impact on the occurrence of sleep interruptions or abnormal events. For example, abnormal changes in heart rate fluctuations, apnea or electromyographic activity. Through this type of evaluation, the system can identify key physiological factors and help further screen potential sleep disorders. After analyzing the sleep event assessment data, possible sleep disorder events such as sleep apnea syndrome, insomnia or periodic limb movements can be automatically screened out. This automatic screening can reduce human intervention and help to more efficiently discover potential problems in sleep. During the sleep monitoring process, the system can perform dynamic evaluation based on real-time collected physiological indicators (such as heart rate, respiratory rate, body movement, etc.). This real-time feedback mechanism can immediately respond to new abnormal events to ensure immediate adjustment and early warning of sleep status. Through real-time physiological indicator collection, the system can perform personalized monitoring based on individual sleep characteristics and identify potential health risks based on individual specific physiological conditions. Sleep stage fluctuation analysis can identify and track fluctuations in different stages of sleep (such as light sleep, deep sleep and REM stages). This analysis helps to understand the duration of each stage, the regularity of mutual transformation, and the fluctuation characteristics between stages, which is crucial for studying the stability and quality of different sleep stages. By correlating the real-time sleep monitoring indicator data with the sleep stage fluctuation data, the system can identify the correlation between physiological disturbances (such as apnea, heart rate disorders) and specific sleep stages. This helps reveal the impact of specific physiological events on changes in sleep stages, and further evaluates whether these disturbances may lead to decreased sleep quality or disorders. By analyzing the relationship between sleep stage fluctuations and physiological disturbances, the system can effectively integrate multi-channel sleep signal data (such as EEG, EMG, ECG, accelerometer data, etc.) to generate high-precision physiological sleep disturbance correlation data. This data can be used to diagnose complex sleep disorders such as sleep apnea syndrome and REM behavior disorder.

[0129] Preferably, step S4 comprises the following steps:

[0130] Step S411: extracting sleep state transition features based on time series from the mixed sleep feature data, thereby obtaining state transition sequence data;

[0131] The embodiment of the present invention extracts sleep state transition features based on time series from mixed sleep feature data, thereby obtaining state transition sequence data. In the specific operation, the mixed sleep feature data is first sorted in chronological order, and the sleep features in each time period are marked as corresponding sleep states (such as NREM, REM or awake). Then, the data is processed segment by segment using a sliding window method (such as a window size set to 30 seconds and a step size of 10 seconds), the state transition features in each time window are extracted, and the start time, end time and frequency of state transitions of the sleep state are recorded. The conversion features of each time window are recorded as state transition sequence data through a timestamp for subsequent modeling and analysis.

[0132] Step S412: Modeling the state transition sequence data using a hidden Markov model to obtain sleep state probability distribution data;

[0133] The embodiment of the present invention uses a hidden Markov model (HMM) to model the state transition sequence data, thereby obtaining sleep state probability distribution data. In the specific operation, the parameters of the hidden Markov model are first initialized, including the number of states (set to 3, corresponding to NREM, REM and awake states), as well as the transition probability matrix between states and the initial state probability distribution. Next, the state transition sequence data extracted in step S411 is input into the model, and the Baum-Welch algorithm is used for parameter estimation and model training. After the training is completed, the model can output the sleep state probability distribution at each time point according to the input feature sequence. In practical applications, the number of hidden states of the HMM is set to 3, and the maximum number of iterations is 100 times to ensure that the model can accurately capture the state transition characteristics of different sleep stages, and the generated sleep state probability distribution data is used for further state transition analysis.

[0134] Step S413: constructing a sleep state transition matrix based on the sleep state probability distribution data to obtain state transition feature data;

[0135] The embodiment of the present invention constructs a sleep state transition matrix based on sleep state probability distribution data to obtain state transition feature data. In the specific operation, the sleep state probability distribution of each time period is first statistically analyzed to calculate the transition frequency between each state. For example, the probability of transitioning from the NREM state to the REM state is determined by the transition frequency in the state probability distribution. Through this statistics, a sleep state transition matrix is ​​constructed, and each element in the matrix represents the transition probability between two sleep states. In order to ensure the accuracy of the matrix, a large amount of historical data can be used for estimation. In practical applications, the rows of the matrix represent the current state (such as NREM, REM), and the columns represent the next state. The data of the transfer matrix is ​​used as the basis for subsequent time series modeling.

[0136] Step S414: extracting local time features of the mixed sleep feature data based on a sliding time window, thereby obtaining time window feature data;

[0137] The embodiment of the present invention extracts local time features based on a sliding time window for mixed sleep feature data, thereby obtaining time window feature data. In the specific operation, the length and step size of the sliding time window are first defined. For example, the window size is set to 60 seconds and the step size is set to 20 seconds, and the data of the entire sleep cycle is processed in segments. The feature data (such as heart rate, eye movement, electromyography, etc.) in each time window are statistically analyzed to extract local time features, such as the mean, standard deviation, maximum value, minimum value, etc. of the features. Then, the local time features in each time window are saved as time window feature data, which are used to perform time series modeling in combination with state transition features. In practical applications, the feature extraction strategies will be different for different physiological signals. For example, the feature extraction of heart rate signals will focus on variability, while the feature extraction of eye movement signals will focus on the frequency of fast and slow eye movements.

[0138] Step S415: performing time series modeling based on a long short-term memory network according to the state transition feature data and the time window feature data to generate a sleep state spatiotemporal evolution model, wherein the sleep state spatiotemporal evolution model includes transition probabilities and duration distributions of different sleep stages;

[0139] The embodiment of the present invention uses a long short-term memory network (LSTM) to perform time series modeling based on state transition feature data and time window feature data to generate a spatiotemporal evolution model of sleep state. In the specific operation, the state transition feature data and the time window feature data are first input into the LSTM model. The input layer of the model is multidimensional feature data, and the hidden layer is set to 128 neurons to capture the temporal dependency of the sleep state. During the training process, LSTM adjusts the weight parameters through back propagation so that the model can accurately predict the sleep state at the next moment. In practical applications, the training data set of the LSTM model is divided into 80% for training and 20% for verification. The optimizer uses Adam and the learning rate is set to 0.001. After the training is completed, the generated spatiotemporal evolution model of the sleep state contains the transition probability and duration distribution of different sleep stages, which provides a basis for predicting the evolution trend of the sleep state.

[0140] Step S416: Predict the sleep state evolution trend according to the sleep state spatiotemporal evolution model and the nonlinear characteristic data, and classify and mark the events to generate different types of sleep state spatiotemporal evolution event data, wherein the sleep state spatiotemporal evolution event data includes normal sleep events, abnormal sleep events, and external environment impact events.

[0141] The embodiment of the present invention predicts the evolution trend of the sleep state according to the spatiotemporal evolution model of the sleep state and the nonlinear characteristic data, and classifies and marks the events to generate different types of spatiotemporal evolution event data of the sleep state. In the specific operation, the spatiotemporal evolution model generated in step S415 is first used to perform a time series prediction of the future sleep state, and the prediction result includes the sleep state at each moment and its transition probability. Then, combined with the nonlinear characteristic data (such as the maximum Lyapunov exponent, recursive features, etc.), trend analysis is performed to determine the stability of the sleep state and possible abnormal changes. Based on these analysis results, the system will classify and mark different types of events (such as normal sleep events, abnormal sleep events, or external environment impact events). For example, abnormal events can be frequent sleep interruptions or abnormal sleep state transitions. Finally, the spatiotemporal evolution event data of sleep states generated by the system can be used for further sleep quality assessment and intervention decisions.

[0142] The present invention can finely depict the changes and transition processes of different sleep stages, such as the sequence and duration of light sleep, deep sleep and rapid eye movement (REM) stages, by extracting time series features from mixed sleep feature data. This information is the basis for subsequent sleep state transition modeling. The extracted state transition sequence provides accurate time series data for subsequent modeling, so that changes in sleep states can be dynamically monitored and evaluated. Hidden Markov model (HMM) is a probability-based time series modeling method that can effectively capture the randomness and implicit laws of sleep states. By modeling state transition sequence data, the probability distribution of sleep stages can be obtained, such as the transition probability of light sleep, deep sleep and REM sleep. HMM helps to reveal the sleep stage transition process that is difficult to observe directly and model it, which can greatly improve the understanding of complex sleep states. By constructing a sleep state transition matrix, the transition rules between different sleep stages, such as the probability of transitioning from the REM stage to light sleep, can be clearly displayed. This helps to better understand the overall structure of the sleep cycle. Transition matrix data can be used to evaluate the stability and predictability of sleep states. For example, a higher probability of continuous deep sleep indicates a more stable sleep state, which helps to evaluate sleep quality. By extracting local time features from mixed sleep feature data using a sliding time window, changes in sleep state over a shorter period of time can be captured. This is important for identifying short sleep events or stage changes, especially when detecting sudden abnormal sleep events. The sliding time window method enables the system to perform more sophisticated time series analysis to avoid ignoring short but important sleep fluctuations or anomalies. Long short-term memory networks (LSTMs) are powerful tools for processing time series data, and are particularly good at modeling long-term dependent sequences. By using LSTM to perform time series modeling on state transition features and time window feature data, the system can capture the long-term dependencies of different sleep stages and their transition trends, and build an accurate spatiotemporal evolution model of sleep states. This spatiotemporal evolution model can not only describe the transition probability of each sleep stage, but also predict the duration distribution of each stage, thereby providing a strong basis for evaluating the integrity of the sleep cycle. By combining nonlinear feature data and the spatiotemporal evolution model of sleep states, the system can predict trends in future sleep states. This feature is particularly helpful for early warning of potential sleep interruptions or abnormal events, such as apnea or sudden awakening. The system can classify and label different types of sleep states based on the prediction results, including normal sleep events, abnormal sleep events (such as frequent micro-awakenings or sleep disorders), and external environmental influences (such as noise, light, etc.). This helps users or doctors better understand the root causes of sleep problems. For abnormal events that may cause health risks (such as apnea or abnormal heart rate), the system can detect and issue warnings early through spatiotemporal evolution predictions, which can help to intervene in time and improve sleep safety.

[0143] Preferably, step S43 includes the following steps:

[0144] Step S431: performing cluster analysis on the sleep event assessment data to identify potential sleep disorder patterns, and obtaining sleep disorder pattern data;

[0145] The embodiment of the present invention performs cluster analysis on sleep event assessment data to identify potential sleep disorder patterns, and obtains sleep disorder pattern data. In the specific operation, the sleep event assessment data is first standardized to ensure that the ranges of different feature data are consistent. Then, the data is clustered using the unsupervised learning K-means clustering algorithm, and the initial number of clusters is set to 5. By iteratively optimizing the positions of cluster centers, similar sleep disorder events are grouped together. The optimal number of clusters is determined by the elbow rule. The clustering results generate multiple sleep disorder patterns, and the main features in each cluster are recorded (such as abnormal sleep interruption frequency, rapid eye movement abnormalities, etc.). The sleep disorder pattern data finally generated is used for subsequent event screening and threshold setting, and can be used to analyze personalized sleep disorder patterns of different users in application scenarios.

[0146] Step S432: setting a sleep disorder event screening threshold based on the sleep disorder pattern data, thereby obtaining screening threshold data; screening the sleep event assessment data for sleep disorder events according to the screening threshold data, thereby obtaining potential disorder event data;

[0147] The embodiment of the present invention sets a sleep disorder event screening threshold based on sleep disorder pattern data, thereby obtaining screening threshold data. In specific operations, each sleep disorder pattern is first analyzed to determine the abnormal value range of specific physiological signals (such as heart rate, respiratory rate) in each type of disorder mode. By performing statistical analysis on historical data, the abnormal range of each physiological signal is calculated, and a screening threshold is set. For example, for heart rate variability, the heart rate variability coefficient is set to exceed 90% of the standard range as the screening threshold for sleep disorder events. Then, new sleep event evaluation data is screened according to these thresholds, and events with abnormalities exceeding the set threshold are retained as potential disorder events. The screening threshold data provides a standard for efficiently screening potential sleep disorders.

[0148] Step S433: determining the physiological indicators that need to be monitored based on the potential obstacle event data, and generating a monitoring indicator list;

[0149] The embodiment of the present invention determines the physiological indicators that need to be monitored based on the potential obstacle event data, and generates a list of monitoring indicators. In the specific operation, the screened potential obstacle event data is first analyzed to identify the physiological characteristics that may cause sleep disorders. For example, heart rate, respiratory rate, eye movement frequency, etc. may be indicators that are highly correlated with potential obstacles. Then, through correlation analysis and expert system recommendations, the physiological indicators most relevant to the specific obstacle pattern are determined. In an application scenario, abnormal rapid eye movement frequency may indicate certain neurological problems, so the eye movement frequency will be included in the list of monitoring indicators. The generated monitoring indicator list contains all the physiological indicators that need to be monitored, and serves as the basis for subsequent sensor acquisition.

[0150] Step S434: configuring the acquisition parameters of the multimodal sensor according to the monitoring indicator list, thereby obtaining sensor configuration data;

[0151] The embodiment of the present invention configures the acquisition parameters of the multimodal sensor according to the monitoring indicator list, thereby obtaining sensor configuration data. In the specific operation, first, according to the monitoring indicator list generated in step S433, select a suitable multimodal sensor for physiological data acquisition. For example, an electrocardiogram sensor is selected for heart rate monitoring, an acceleration sensor is selected for body movement, and an electrooculogram sensor is selected for eye movement. Then, according to the characteristics of different sensors, set the acquisition parameters, such as sampling frequency, transmission interval, etc. For electrocardiogram data, the sampling frequency can be set to 500Hz, and the sampling frequency of acceleration data can be set to 50Hz. The sensor configuration data contains the acquisition parameters of all devices, which are used to ensure that the device can efficiently and accurately collect physiological data.

[0152] Step S435: using the sensor configuration data to collect real-time physiological indicators of the subject, and performing real-time signal processing, so as to obtain real-time sleep monitoring indicator data.

[0153] The embodiment of the present invention utilizes the sensor configuration data to collect real-time physiological indicators of the object under test, and performs real-time signal processing, so as to obtain real-time sleep monitoring indicator data. In the specific operation, the configured sensor device is first worn on the object under test, and the sensor begins to collect physiological signals in real time according to the set parameters. The collected data is transmitted to the data processing terminal via wireless transmission, and real-time signal processing is performed on the terminal. The processing process includes operations such as filtering, denoising, and feature extraction. For example, the electrocardiogram signal can be filtered out of high-frequency noise through a low-pass filter, and then the heart rate data is extracted through a QRS wave detection algorithm. The real-time sleep monitoring indicator data finally generated contains the real-time changes of multiple physiological signals, such as heart rate variability, respiratory rate, eye movement frequency, etc., which are used for further sleep quality analysis and intervention.

[0154] The present invention performs pattern recognition on sleep event assessment data through cluster analysis, and can discover different sleep disorder patterns (such as apnea, frequent awakening, REM behavioral disorder, etc.). This pattern recognition not only helps to automatically classify different types of sleep disorders, but also can reveal the potential associations between various disorders. Cluster analysis helps to generate personalized sleep disorder patterns based on the sleep data of different individuals, thereby providing a basis for individualized sleep problem diagnosis. Based on the results of cluster analysis, the system can set a screening threshold and perform event screening for specific sleep disorder patterns. For example, for sleep apnea, a specific breathing interruption frequency or heart rate fluctuation amplitude threshold can be set to ensure that the screened events are highly relevant. The setting of the screening threshold can be adaptively adjusted according to the different sleep characteristics of each person, avoiding the limitations of a single threshold and improving the screening accuracy. Determination of key monitored physiological indicators: Based on the potential disorder events screened, the system can automatically generate a list of physiological indicators that need to be monitored. By selectively selecting key physiological indicators for monitoring, unnecessary monitoring items are reduced, the monitoring process is optimized, and the monitoring efficiency and data processing accuracy are improved. According to the generated list of monitoring indicators, the system can dynamically configure the acquisition parameters of the multimodal sensor to ensure that the sensor can accurately collect the key monitored physiological indicators. This not only optimizes the use of sensors, but also improves the effectiveness of data collection. Through precise parameter configuration, the system can avoid unnecessary data redundancy and acquisition noise, ensure the clarity and accuracy of the signal, and thus improve the effect of subsequent data processing and analysis. With the optimized sensor configuration, the system can collect real-time physiological indicators of the measured object to ensure that when a potential sleep disorder event is detected, the relevant physiological data can be accurately captured, thereby improving the detection and confirmation rate of the event. The system can perform real-time signal processing on the collected physiological data and quickly identify abnormal conditions. This real-time processing capability is crucial for dealing with sudden sleep disorders (such as sleep apnea, arrhythmia, etc.), and can provide effective support for timely intervention. The data collected and processed in real time can provide feedback for subsequent sleep state monitoring and sleep disorder assessment, forming a closed-loop monitoring mechanism to ensure that the system continuously optimizes monitoring strategies and configurations.

[0155] Preferably, step S44 includes the following steps:

[0156] Step S441: performing sleep cycle change analysis based on a dynamic time warping algorithm on the multi-channel sleep signal segment data to obtain sleep cycle data; and performing statistics on the duration and conversion frequency of each sleep stage based on the sleep cycle data to obtain sleep structure data;

[0157] The embodiment of the present invention performs a sleep cycle change analysis based on a dynamic time warping algorithm on multi-channel sleep signal segment data, thereby obtaining sleep cycle data. In the specific operation, first obtain multi-channel sleep signal segments, including electroencephalogram (EEG), eye movement signals (EOG), electromyography (EMG), etc. Then, use the dynamic time warping (DTW) algorithm to align and analyze the cycles of these multidimensional signals, calculate the similarities and differences between different channels, and then identify the boundaries and change patterns of each sleep cycle. By analyzing the signal features in multiple consecutive sleep cycles, accurate sleep cycle data is obtained. Then, based on this cycle data, the duration and conversion frequency of each sleep stage are statistically analyzed to generate sleep structure data. In the application scenario, this method is suitable for identifying multiple sleep cycles of a user in one night and their characteristic changes, such as the transition from rapid eye movement (REM) to deep sleep (NREM).

[0158] Step S442: predicting the sleep stage change trend based on the time series of the sleep structure data, thereby obtaining sleep stage fluctuation data;

[0159] The embodiment of the present invention predicts the change trend of sleep stages based on the time series of sleep structure data, thereby obtaining sleep stage fluctuation data. In the specific operation, firstly, the sleep structure data obtained in the early stage is modeled by time series, and prediction models such as the autoregressive moving average (ARIMA) model or the long short-term memory (LSTM) network are used to predict the change trend of each sleep stage in the future based on historical data. By analyzing the periodic characteristics and volatility in the time series, the fluctuation of the user's sleep stage in the future is predicted. The sleep stage fluctuation data finally generated reflects the change trend of different stages (such as light sleep, deep sleep, REM) that may occur in the future sleep. This step is particularly suitable for monitoring the long-term changes in the user's sleep stages, so as to identify potential sleep quality problems in advance.

[0160] Step S443: performing physiological disturbance correlation analysis on the sleep stage fluctuation data according to the real-time sleep monitoring index data, identifying the correlation between the sleep stage and the physiological signal change, and generating physiological disturbance correlation data;

[0161] The embodiment of the present invention performs a physiological disturbance correlation analysis on the sleep stage fluctuation data based on the real-time sleep monitoring index data, identifies the correlation between the sleep stage and the physiological signal changes, and generates physiological disturbance correlation data. In the specific operation, the real-time sleep monitoring index data (such as heart rate, respiratory rate, blood oxygen saturation, etc.) are first obtained, and then these data are correlated with the sleep stage fluctuation data obtained in the previous step. The relationship between the change of each physiological indicator and the specific sleep stage is analyzed by statistical methods such as the Pearson correlation coefficient and mutual information, and patterns such as whether the heart rate acceleration is related to the REM stage, or whether the respiratory rate change is related to the light sleep stage are identified. The physiological disturbance correlation data finally generated reflects the specific correlation between the sleep stage and the changes in key physiological signals, which is convenient for subsequent risk assessment.

[0162] Step S444: performing pattern classification on the physiological disturbance associated data according to a preset disturbance pattern library to generate disturbance pattern data; performing physiological disturbance risk assessment on the physiological signal disturbance of each sleep stage according to the disturbance pattern data, thereby obtaining physiological disturbance assessment data;

[0163] The embodiment of the present invention performs pattern classification on the physiological disturbance associated data according to a preset disturbance pattern library to generate disturbance pattern data. In the specific operation, a physiological disturbance pattern library is first established, which contains common physiological signal disturbance patterns (such as awakening corresponding to accelerated heart rate, sleep apnea corresponding to irregular breathing, etc.). Then, the physiological disturbance associated data is matched with the template in the pattern library using a pattern matching algorithm to identify the type of the current physiological disturbance and generate the corresponding disturbance pattern data. Based on the disturbance pattern data, a risk assessment is performed on the physiological signal disturbances occurring in each sleep stage. The assessment method can generate physiological disturbance assessment data based on a weighted scoring system, combined with the frequency, duration and severity of the disturbance. In the application scenario, this step can identify high-risk sleep problems, such as frequent apnea or excessive heart rate fluctuations.

[0164] Step S445: generating physiological sleep disturbance associated data including physiological index impact weights, sleep stage transition probabilities, and disturbance degrees according to the physiological disturbance assessment data.

[0165] The embodiment of the present invention generates physiological sleep disturbance association data including physiological indicator impact weights, sleep stage transition probabilities, and disturbance levels based on physiological disturbance assessment data. In specific operations, the physiological disturbance assessment data is first weighted and analyzed, and the impact weight of each physiological indicator is set (e.g., the impact weight of heart rate may be higher, while the weight of body movement is lower). Then, based on the sleep stage transition probability data and the degree of disturbance, the disturbance events that may occur in each sleep stage are evaluated. The physiological sleep disturbance association data finally generated includes the degree of influence of different physiological indicators in different sleep stages, and provides the transition probability of disturbances in each stage. These data can be used to further optimize sleep intervention measures, help users improve sleep quality and reduce potential health risks.

[0166] The dynamic time warping (DTW) algorithm of the present invention can process asynchronous time series data and is suitable for processing sleep signals of different time periods or different individuals. By analyzing multi-channel sleep signal fragment data through the algorithm, subtle changes in the sleep cycle can be accurately captured, and the reliability of sleep cycle data can be improved. By statistically analyzing the duration and conversion frequency of each sleep stage (such as light sleep, deep sleep, REM, etc.), the sleep structure of the subject can be more comprehensively reflected, providing basic data for subsequent sleep quality assessment. Through time series analysis, the system can predict the changing trend of the sleep stage. For example, the system can identify whether the light sleep state is about to turn into a deep sleep or REM stage. This predictive ability can help users or doctors understand the fluctuation of the sleep state in advance and provide an opportunity for intervention in sleep problems. Predicting the fluctuation trend of the sleep stage allows the system to respond in time when abnormal fluctuations are identified, thereby dynamically adjusting the monitoring strategy to avoid data loss or monitoring delays caused by sudden problems. By correlating the real-time monitored physiological indicators with the sleep stage fluctuation data, the potential connection between specific physiological signals (such as heart rate, respiratory rate, etc.) and changes in sleep stages can be revealed. For example, the system can identify the phenomenon that heart rate variability increases significantly in the REM stage, thus providing a reference for subsequent intervention measures. Each person's physiological disturbance may have a unique association pattern with the sleep stage. This analysis step can identify individual physiological disturbance characteristics through association analysis to provide support for personalized sleep monitoring. By classifying the physiological disturbance association data, the system can quickly identify common or abnormal physiological disturbance patterns in the sleep stage based on the disturbance pattern library. For example, the system can identify physiological signal disturbances based on apnea patterns. This standardized pattern classification helps to improve the recognition efficiency and accuracy of the system. By performing risk assessment on the identified disturbance patterns, the system can quantify the potential impact of different physiological disturbances on sleep health. This assessment mechanism can not only provide users with clear risk levels (such as low risk, medium risk or high risk), but also provide doctors with further diagnostic basis. This step combines the impact weight of physiological signals, the transition probability of sleep stages, and the degree of disturbance to generate a complete set of physiological sleep disturbance association data. Through this comprehensive analysis, the system can more accurately quantify the impact of physiological signals on different sleep stages and provide a comprehensive and quantitative assessment result for sleep health. Physiological sleep disturbance association data provides a reliable basis for subsequent intervention measures. Based on this data, doctors can determine which abnormal fluctuations in physiological indicators may have a significant impact on specific sleep stages, thereby formulating more targeted treatment or monitoring plans. Through dynamic time warping algorithms and time series analysis, the system can accurately track changes in sleep cycles and predict fluctuations in sleep stages. This comprehensive analysis helps to fully understand the health of sleep structure.Through correlation analysis and pattern classification of physiological disturbances, the system can identify potential sleep disorders and physiological abnormalities, which is crucial for early detection and prevention of sleep-related diseases; each person's sleep and physiological characteristics are different, and the design of steps S441 to S445 can generate personalized risk assessment reports based on individual data, thereby providing a scientific basis for precise intervention; physiological sleep disturbance correlation data can not only provide doctors with comprehensive diagnostic information, but also support real-time monitoring systems, helping doctors to adjust monitoring plans or take intervention measures in a timely manner when necessary.

[0167] The present invention also provides a feature information extraction system for a sleep state monitoring model, which is used to execute the feature information extraction method for the sleep state monitoring model. The feature information extraction system for the sleep state monitoring model comprises:

[0168] The signal preprocessing module is used to collect multimodal physiological signals of the subject to obtain original sleep monitoring data; remove noise and enhance the signal of the original sleep monitoring data, and perform signal segmentation to obtain multi-channel sleep signal segment data;

[0169] A multi-dimensional feature extraction module is used to extract features of multi-channel sleep signal fragment data based on electroencephalogram, oculogram, electromyogram, electrocardiogram and accelerometer signals to obtain multi-dimensional feature data; and to extract ultrasonic features and acoustic features of multi-channel sleep signal fragment data to generate sound wave feature data;

[0170] A feature fusion module is used to extract nonlinear dynamic features based on multidimensional feature data to generate nonlinear feature data; and to perform complementary fusion processing on the sound wave feature data through the nonlinear feature data to generate mixed sleep feature data;

[0171] The sleep dynamic analysis module is used to perform spatiotemporal evolution processing and event impact assessment on mixed sleep feature data to generate sleep event assessment data; screen potential sleep disorder events based on the sleep event assessment data, collect real-time physiological indicators, and generate real-time sleep monitoring indicator data; perform sleep stage fluctuation analysis and physiological disturbance correlation processing on multi-channel sleep signal fragment data through real-time sleep monitoring indicator data to generate physiological sleep disturbance correlation data;

[0172] The environmental-physiological coupling module is used to identify the sleep environment impact area of ​​the sleep stage fluctuation data and generate sleep environment impact area data; to couple the sleep environment impact area data with the physiological-environment impact pattern to obtain the physiological-environment coupling pattern data; and to merge the mixed sleep feature data, the physiological sleep disturbance association data and the physiological-environment coupling pattern data into feature information extraction data.

[0173] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.

[0174] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be 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 present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A method for extracting feature information of a sleep state monitoring model, characterized in that: The following steps are involved: Step S1: collecting multimodal physiological signals of the subject to obtain original sleep monitoring data; The original sleep monitoring data is subjected to noise removal and signal enhancement processing, and signal segmentation is performed to obtain multi-channel sleep signal segment data; Step S2: extracting features of the multi-channel sleep signal segment data based on electroencephalogram, oculogram, electromyogram, electrocardiogram and accelerometer signals to obtain multi-dimensional feature data; extracting ultrasonic features and acoustic features of the multi-channel sleep signal segment data to generate sound wave feature data; Step S3: extracting nonlinear dynamic features according to the multidimensional feature data to generate nonlinear feature data; performing complementary fusion processing on the sound wave feature data through the nonlinear feature data to generate mixed sleep feature data; Step S4: performing spatiotemporal evolution processing and event impact assessment on the mixed sleep feature data to generate sleep event assessment data; screening potential sleep disorder events based on the sleep event assessment data, and collecting real-time physiological indicators to generate real-time sleep monitoring indicator data; performing sleep stage fluctuation analysis and physiological disturbance association processing on the multi-channel sleep signal segment data through the real-time sleep monitoring indicator data to generate physiological sleep disturbance association data; Step S5: Identify the sleep environment impact area based on the sleep stage fluctuation data to generate sleep environment impact area data; The sleep environment impact area data is coupled with the physiological-environmental impact pattern to obtain the physiological-environmental coupling pattern data; the mixed sleep feature data, the physiological sleep disturbance associated data and the physiological-environmental coupling pattern data are combined into feature information extraction data.

2. The method for extracting characteristic information of a sleep state monitoring model according to claim 1, characterized in that: Step S1 includes the following steps: Step S11: continuously collecting sleep data of the subject using a polysomnography monitor to obtain sleep physiological data, wherein the sleep physiological data includes electroencephalogram, oculogram, electromyogram, electrocardiogram and accelerometer signals; Step S12: collecting body motion signals through an ultrasonic sensor and collecting environmental sounds through a high-sensitivity microphone, thereby obtaining body motion environmental acoustic data; Step S13: merging the sleep physiological data and the body motion environment acoustic data into original sleep monitoring data; Step S14: performing signal quality assessment on the original sleep monitoring data to obtain signal quality assessment data, wherein the signal quality assessment includes signal-to-noise ratio calculation and electrode detachment detection; performing electrooculography and electromyography artifact removal on the original sleep monitoring data, and removing baseline drift using wavelet transform, thereby obtaining sleep monitoring data; Step S15: performing signal enhancement based on empirical mode decomposition on the sleep monitoring data according to the signal quality evaluation data, thereby obtaining enhanced sleep monitoring data; Step S16: performing signal segmentation on the enhanced sleep monitoring data to obtain multi-channel sleep signal segment data.

3. The method for extracting characteristic information of a sleep state monitoring model according to claim 2, characterized in that: Step S16 includes the following steps: Step S161: segmenting the enhanced sleep monitoring data according to the preset time window data to obtain initial sleep signal segment data; Step S162: performing multi-channel time synchronization and alignment processing on the initial sleep signal segment data to obtain aligned sleep signal segment data; Step S163: aligning the sleep signal segment data to perform data integrity check, identifying and marking missing or distorted data segments, and obtaining integrity check data; Step S164: performing amplitude normalization processing on the aligned sleep signal segment data according to the integrity check data, and performing fast Fourier transform to obtain sleep signal spectrum data; Step S165: performing multi-channel integration and synchronous processing on the sleep signal spectrum data to obtain multi-channel sleep signal segment data, wherein the multi-channel includes electroencephalogram, oculogram, electromyogram, electrocardiogram, accelerometer signal, ultrasonic signal and acoustic signal.

4. The method for extracting characteristic information of a sleep state monitoring model according to claim 3, characterized in that: Step S2 includes the following steps: Step S21: extracting energy features of the δ, θ, α, β and γ bands from the EEG signals in the multi-channel sleep signal segment data, and performing waveform detection based on K complexes and sleep spindles to obtain EEG feature data; Step S22: performing fast and slow eye movement identification on the eye movement signal in the multi-channel sleep signal segment data, and extracting the eye movement frequency and amplitude characteristics in the rapid eye movement and non-rapid eye movement states to obtain eye movement feature data; Step S23: performing muscle tension analysis based on the intensity and frequency characteristics of electromyographic activity on the electromyographic signals in the multi-channel sleep signal segment data to obtain electromyographic characteristic data; Step S24: performing heart rate variability analysis on the electrocardiogram signal in the multi-channel sleep signal segment data, extracting time domain and frequency domain heart rate variability indicators, and obtaining electrocardiogram feature data; Step S25: performing body motion analysis on the accelerometer signal in the multi-channel sleep signal segment data, extracting body motion frequency, intensity and duration characteristics, and obtaining body motion feature data; Step S26: merging the EEG feature data, the eye movement feature data, the myoelectric feature data, the electrocardiographic feature data, and the body movement feature data into multi-dimensional feature data; Step S27: performing Doppler frequency shift analysis on the ultrasonic signal in the multi-channel sleep signal segment data, extracting micro-body motion features, and obtaining ultrasonic feature data; Step S28: Detect and classify sound events on the ambient sound signals in the multi-channel sleep signal segment data, extract sleep sound features based on breathing sounds and snoring sounds, and obtain acoustic feature data; merge the ultrasonic feature data and the acoustic feature data into sound wave feature data.

5. The method for extracting characteristic information of a sleep state monitoring model according to claim 4, characterized in that: Step S3 includes the following steps: Step S31: reconstructing the multidimensional characteristic data in phase space to obtain reconstructed phase space data; calculating the maximum Lyapunov exponent based on the reconstructed phase space data to obtain system stability characteristic data; Step S32: performing recursive quantitative analysis on the multi-dimensional feature data, and extracting the recursive rate and deterministic features to obtain recursive feature data; Step S33: performing sample entropy and approximate entropy calculations on the multidimensional feature data to obtain entropy feature data; performing detrending fluctuation analysis on the multidimensional feature data and extracting long-range correlation features to obtain fluctuation feature data; Step S34: fusing the system stability characteristic data, the recursive characteristic data, the entropy characteristic data and the fluctuation characteristic data to obtain nonlinear characteristic data; Step S35: performing wavelet packet transformation on the sound wave characteristic data, and extracting time-frequency domain features to obtain time-frequency sound wave characteristic data; Step S36: using the nonlinear characteristic data to perform characteristic weighting on the time-frequency sound wave characteristic data to obtain weighted sound wave characteristic data; Step S37: performing feature fusion based on a deep neural network on the nonlinear feature data and the weighted sound wave feature data to obtain mixed sleep feature data.

6. The method for extracting characteristic information of a sleep state monitoring model according to claim 5, characterized in that: Step S4 includes the following steps: Step S41: performing a sleep state spatiotemporal evolution process according to the mixed sleep characteristic data to generate sleep state spatiotemporal evolution event data; Step S42: identifying potential sleep interruptions or abnormal events in the mixed sleep feature data through the sleep state spatiotemporal evolution event data, and performing event impact assessment based on feature importance to generate sleep event assessment data; Step S43: Screening potential sleep disorder events based on the sleep event assessment data, and collecting real-time physiological indicators to generate real-time sleep monitoring indicator data; Step S44: performing sleep stage fluctuation analysis on the multi-channel sleep signal segment data to generate sleep stage fluctuation data; and performing physiological disturbance association processing on the sleep stage fluctuation data through the real-time sleep monitoring index data to generate physiological sleep disturbance association data.

7. The method for extracting characteristic information of a sleep state monitoring model according to claim 6, characterized in that: Step S4 includes the following steps: Step S411: extracting sleep state transition features based on time series from the mixed sleep feature data, thereby obtaining state transition sequence data; Step S412: Modeling the state transition sequence data using a hidden Markov model to obtain sleep state probability distribution data; Step S413: constructing a sleep state transition matrix based on the sleep state probability distribution data to obtain state transition feature data; Step S414: extracting local time features of the mixed sleep feature data based on a sliding time window, thereby obtaining time window feature data; Step S415: performing time series modeling based on a long short-term memory network according to the state transition feature data and the time window feature data to generate a sleep state spatiotemporal evolution model, wherein the sleep state spatiotemporal evolution model includes transition probabilities and duration distributions of different sleep stages; Step S416: Predict the sleep state evolution trend according to the sleep state spatiotemporal evolution model and the nonlinear characteristic data, and classify and mark the events to generate different types of sleep state spatiotemporal evolution event data, wherein the sleep state spatiotemporal evolution event data includes normal sleep events, abnormal sleep events, and external environment impact events.

8. The method for extracting characteristic information of a sleep state monitoring model according to claim 7, characterized in that: Step S43 includes the following steps: Step S431: performing cluster analysis on the sleep event assessment data to identify potential sleep disorder patterns, and obtaining sleep disorder pattern data; Step S432: setting a sleep disorder event screening threshold based on the sleep disorder pattern data, thereby obtaining screening threshold data; screening the sleep event assessment data for sleep disorder events according to the screening threshold data, thereby obtaining potential disorder event data; Step S433: determining the physiological indicators that need to be monitored based on the potential obstacle event data, and generating a monitoring indicator list; Step S434: configuring the acquisition parameters of the multimodal sensor according to the monitoring indicator list, thereby obtaining sensor configuration data; Step S435: using the sensor configuration data to collect real-time physiological indicators of the subject, and performing real-time signal processing, so as to obtain real-time sleep monitoring indicator data.

9. The method for extracting characteristic information of a sleep state monitoring model according to claim 8, characterized in that: Step S44 includes the following steps: Step S441: performing sleep cycle change analysis based on a dynamic time warping algorithm on the multi-channel sleep signal segment data to obtain sleep cycle data; and performing statistics on the duration and conversion frequency of each sleep stage based on the sleep cycle data to obtain sleep structure data; Step S442: predicting the sleep stage change trend based on the time series of the sleep structure data, thereby obtaining sleep stage fluctuation data; Step S443: performing physiological disturbance correlation analysis on the sleep stage fluctuation data according to the real-time sleep monitoring index data, identifying the correlation between the sleep stage and the physiological signal change, and generating physiological disturbance correlation data; Step S444: performing pattern classification on the physiological disturbance associated data according to a preset disturbance pattern library to generate disturbance pattern data; performing physiological disturbance risk assessment on the physiological signal disturbance of each sleep stage according to the disturbance pattern data, thereby obtaining physiological disturbance assessment data; Step S445: generating physiological sleep disturbance associated data including physiological index impact weights, sleep stage transition probabilities, and disturbance degrees according to the physiological disturbance assessment data.

10. A feature information extraction system for a sleep state monitoring model, characterized in that: Used to execute the feature information extraction method of the sleep state monitoring model according to claim 1, the feature information extraction system of the sleep state monitoring model comprises: The signal preprocessing module is used to collect multimodal physiological signals of the subject to obtain original sleep monitoring data; remove noise and enhance the signal of the original sleep monitoring data, and perform signal segmentation to obtain multi-channel sleep signal segment data; A multi-dimensional feature extraction module is used to extract features of multi-channel sleep signal fragment data based on electroencephalogram, oculogram, electromyogram, electrocardiogram and accelerometer signals to obtain multi-dimensional feature data; and to extract ultrasonic features and acoustic features of multi-channel sleep signal fragment data to generate sound wave feature data; A feature fusion module is used to extract nonlinear dynamic features based on multidimensional feature data to generate nonlinear feature data; and to perform complementary fusion processing on the sound wave feature data through the nonlinear feature data to generate mixed sleep feature data; The sleep dynamic analysis module is used to perform spatiotemporal evolution processing and event impact assessment on mixed sleep feature data to generate sleep event assessment data; screen potential sleep disorder events based on the sleep event assessment data, collect real-time physiological indicators, and generate real-time sleep monitoring indicator data; perform sleep stage fluctuation analysis and physiological disturbance correlation processing on multi-channel sleep signal fragment data through real-time sleep monitoring indicator data to generate physiological sleep disturbance correlation data; The environmental-physiological coupling module is used to identify the sleep environment impact area of ​​the sleep stage fluctuation data and generate sleep environment impact area data; to couple the sleep environment impact area data with the physiological-environment impact pattern to obtain the physiological-environment coupling pattern data; and to merge the mixed sleep feature data, the physiological sleep disturbance association data and the physiological-environment coupling pattern data into feature information extraction data.

Images