A sleep staging method, device, equipment and storage medium

Abstract

The present invention relates to a sleep staging method, device, equipment and storage medium. The sleep staging method comprises: obtaining a low-load physiological signal of a target user; extracting the heartbeat features of the beat-by-beat interval of the physiological signal to obtain a beat-by-beat interval sequence; wherein the beat-by-beat interval refers to the time interval between two adjacent heartbeats; analyzing the heart rate variability according to the beat-by-beat interval sequence, and extracting the physiological features of the target user from multiple angles; classifying the physiological features through a pre-trained machine learning model to determine the sleep staging result of the target user. The method provided by the present application improves the accuracy and acceptability of sleep staging for children.

Description

Technical Field

[0001] The present invention relates to the technical field of sleep monitoring, and in particular to a sleep staging method, device, equipment and storage medium. Background Art

[0002] Traditional polysomnography equipment has many challenges in children's sleep staging, including complex operation, high cost, and low comfort and acceptability for children. In addition, there are differences in physiological signals between adults and children. Traditional adult sleep staging methods may not be fully applicable to children, resulting in poor sleep monitoring results for children. Therefore, there is an urgent need to provide a sleep staging method to improve the accuracy and acceptability of children's sleep staging. Summary of the invention

[0003] In order to solve the above technical problems, the embodiments of the present disclosure provide a sleep staging method, apparatus, device and storage medium.

[0004] In a first aspect, an embodiment of the present disclosure provides a sleep staging method, comprising:

[0005] Obtain low-load physiological signals from target users;

[0006] Extract the heartbeat features of the physiological signal during the beat-by-beat interval to obtain a beat-by-beat interval sequence; wherein the beat-by-beat interval refers to the time interval between two adjacent heartbeats;

[0007] Analyze heart rate variability based on the beat-to-beat interval sequence and extract physiological characteristics of the target user from multiple perspectives;

[0008] Physiological features are classified through a pre-trained machine learning model to determine the sleep stage results of the target user.

[0009] In a second aspect, an embodiment of the present disclosure provides a sleep staging device, comprising:

[0010] An acquisition unit, used for acquiring a low-load physiological signal of a target user;

[0011] An extraction unit is used to extract heartbeat features of the physiological signal during the beat-by-beat interval to obtain a beat-by-beat interval sequence; wherein the beat-by-beat interval refers to the time interval between two adjacent heartbeats;

[0012] An analysis unit, used to analyze the heart rate variability according to the beat-by-beat interval sequence and extract physiological characteristics of the target user from multiple angles;

[0013] The classification unit is used to classify the physiological characteristics through a pre-trained machine learning model to determine the sleep stage results of the target user.

[0014] In a third aspect, an embodiment of the present disclosure provides an electronic device, including:

[0015] Memory;

[0016] Processor; and

[0017] Computer programs;

[0018] The computer program is stored in the memory and is configured to be executed by the processor to implement the sleep staging method as described above.

[0019] In a fourth aspect, an embodiment of the present disclosure provides a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the sleep staging method as described above are implemented.

[0020] The sleep staging method provided by the present application includes: obtaining low-load physiological signals of the target user; extracting the heartbeat features of the beat-by-beat interval of the physiological signals to obtain a beat-by-beat interval sequence; wherein the beat-by-beat interval refers to the time interval between two adjacent heartbeats; analyzing the heart rate variability according to the beat-by-beat interval sequence, and extracting the physiological characteristics of the target user from multiple angles; classifying the physiological characteristics through a pre-trained machine learning model to determine the sleep staging results of the target user. The method provided by the present application fully considers the detection comfort of children, collects low-load physiological signals, and completes children's sleep staging based on low-load physiological signals, thereby improving the accuracy and acceptability of children's sleep staging and facilitating relevant users to understand children's sleep conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present disclosure and, together with the description, serve to explain the principles of the present disclosure.

[0022] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0023] Figure 1 A flowchart of a sleep staging method provided by an embodiment of the present disclosure;

[0024] Figure 2 A schematic diagram of a process for extracting beat-by-beat intervals of an electrocardiogram signal provided in an embodiment of the present disclosure;

[0025] FIG3 (a) is a diagram of an electrocardiogram signal feature point location and a beat-by-beat interval sequence result provided by an embodiment of the present disclosure;

[0026] FIG3( b ) is another result diagram of ECG signal feature point positioning and beat-by-beat interval sequence provided by an embodiment of the present disclosure;

[0027] FIG3( c ) is another electrocardiogram signal feature point location and beat-by-beat interval sequence result diagram provided by an embodiment of the present disclosure;

[0028] Figure 4 A flow chart of a porous decomposition algorithm provided for an embodiment of the present disclosure;

[0029] Figure 5 A frequency amplitude response curve diagram of an equivalent filter provided in an embodiment of the present disclosure at a sampling frequency of 250 Hz;

[0030] Figure 6 A schematic diagram of a 5-layer decomposition signal provided by an embodiment of the present disclosure;

[0031] Figure 7 A spectrum diagram of a 5-layer decomposed signal provided by an embodiment of the present disclosure;

[0032] Figure 8 A schematic diagram of a clustering template and an average template provided in an embodiment of the present disclosure;

[0033] Fig. 9 A result diagram of the characteristic point positioning and beat-by-beat interval sequence of the heartbeat signal provided by the embodiment of the present disclosure;

[0034] Fig.10 A flow chart of heart rate variability analysis and feature extraction provided by an embodiment of the present disclosure;

[0035] Fig.11 A schematic diagram of the structure of a sleep staging device provided in an embodiment of the present disclosure;

[0036] Fig.12 A schematic diagram of the structure of an electronic device provided in an embodiment of the present disclosure. DETAILED DESCRIPTION

[0037] In order to more clearly understand the above-mentioned objectives, features and advantages of the present disclosure, the scheme of the present disclosure will be further described below. It should be noted that the embodiments of the present disclosure and the features in the embodiments can be combined with each other without conflict.

[0038] In the following description, many specific details are set forth to facilitate a full understanding of the present disclosure, but the present disclosure may also be implemented in other ways different from those described herein; it is obvious that the embodiments in the specification are only part of the embodiments of the present disclosure, rather than all of the embodiments.

[0039] Specifically, the existing adult sleep staging methods are not effective for children's sleep monitoring. First, the ECG signal characteristics of children are significantly different from those of adults. Secondly, the existing deep sleep and REM (rapid eye movement) stage classification performance is poor, and the existing sleep staging methods are susceptible to interference under low signal-to-noise ratio conditions, which may lead to erroneous staging results. In addition, traditional sleep monitoring equipment is not suitable for daily monitoring of children, is not friendly to children's wearing experience, and is not conducive to popularization among children. Therefore, the existing adult sleep staging methods have great difficulty in extracting children's sleep signal features, especially under low-load monitoring conditions, when extracting beat-to-beat intervals and performing HRV analysis, it is easy to introduce errors, affecting the accuracy of sleep staging.

[0040] In view of the above technical problems, the embodiments of the present disclosure provide a sleep staging method, which extracts features from children's physiological signals, such as heart rate variability, and uses a variety of traditional machine learning algorithms to classify physiological signals based on the extracted features, thereby realizing automatic staging of children's sleep stages, which can be applied to children's sleep monitoring equipment to help relevant personnel understand children's sleep conditions. This is described in detail through one or more of the following embodiments.

[0041] The sleep staging method provided in the embodiments of the present disclosure can be applied to children's sleep monitoring scenarios. The method can be performed by a sleep staging device, which can be implemented by software and / or hardware, and the system can be integrated in an electronic device. Among them, the electronic device can include but is not limited to mobile terminals such as smart phones, laptops, digital broadcast receivers, personal digital assistants (PDA), tablet computers (Tablet Personal Computer, Tablet PC), PMP (portable multimedia player), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), wearable devices, etc., and fixed terminals such as digital televisions, desktop computers, smart home devices, etc.

[0042] Figure 1 A sleep staging method provided by the embodiment of the present disclosure specifically includes the following steps: Figure 1 Steps shown:

[0043] S101. Obtain a low-load physiological signal of a target user.

[0044] It is understandable that low-load physiological signals of target users are obtained. Target users may refer to children and other relevant users who are not convenient to use traditional sleep detection equipment. Specifically, low-load physiological signals can be collected through monitoring equipment suitable for children. Suitable for children means that factors such as children's monitoring comfort and acceptability are fully considered. Among them, low-load physiological signals refer to physiological signals whose collection process has less impact on the subjects and are easier to monitor for a long time. These signals are usually more comfortable and less invasive, and can be continuously collected in daily living environments without affecting the subjects' normal activities.

[0045] S102, extracting heartbeat features of beat-to-beat intervals of physiological signals to obtain a beat-to-beat interval sequence.

[0046] The beat-to-beat interval refers to the time interval between two adjacent heartbeats.

[0047] It can be understood that, based on the above S101, the heartbeat feature extraction of the beat-to-beat interval of the physiological signal is performed to obtain a beat-to-beat interval sequence to reduce noise and interference in the physiological signal, ensure the high robustness of the physiological signal, and effectively extract features under low signal-to-noise ratio conditions. Among them, the beat-to-beat interval refers to the time interval between two adjacent heartbeats, which is usually measured in milliseconds (ms) and is mostly used for heart rate variability analysis. Heart rate variability (HRV) refers to the time change of the heartbeat interval and is one of the important indicators for evaluating the activity of the autonomic nervous system and cardiovascular health.

[0048] Among them, physiological signals include electrocardiogram signals.

[0049] As you can understand, an electrocardiogram (ECG) is a method of recording the electrical activity of the heart by placing electrodes on the surface of the skin to detect the electrical changes in the heart during each cardiac cycle.

[0050] Optionally, the above-mentioned extraction of heartbeat features of beat-by-beat intervals of physiological signals to obtain a beat-by-beat interval sequence can be specifically implemented through the following steps:

[0051] The heartbeat features of the ECG signal are extracted by quadratic spline wavelet decomposition and R wave peak location to obtain the beat-to-beat interval sequence.

[0052] It is understandable that the ECG signal is decomposed based on the quadratic spline wavelet decomposition method, wherein the coefficients of the quadratic spline wavelet are simple, and when the ECG signal waveform is decomposed, not only the noise can be filtered out, but also the singular values ​​of the ECG signal can be highlighted, wherein the singular values ​​reflect the main components of the ECG signal, for example, larger singular values ​​correspond to the main components of the ECG signal, and smaller singular values ​​usually correspond to noise or secondary components. Therefore, the main components of the ECG signal can be extracted and the noise can be removed by wavelet decomposition. Subsequently, the peak advantage of the decomposed waveform is increased by the square operation, and the R wave peak value in the ECG signal is found with the help of the obvious peak. This method is simple to calculate and has strong robustness. Finally, the beat-to-beat interval sequence of the ECG signal is determined based on the R wave peak value.

[0053] Among them, physiological signals also include cardiac shock signals.

[0054] As you can understand, Ballistocardiogram (BCG) is a technology that records heart activity by detecting the tiny mechanical vibrations generated by the human body every time the heart beats. The BCG signal reflects the changes in force and acceleration generated during the heart pumping blood, which can be detected by highly sensitive sensors.

[0055] Optionally, the above-mentioned extraction of heartbeat features of beat-by-beat intervals of physiological signals to obtain a beat-by-beat interval sequence can be specifically implemented through the following steps:

[0056] The heartbeat features of the electrocardiogram signal are extracted in the beat-to-beat interval by means of quadratic spline wavelet decomposition and R-wave peak location to obtain a first sequence; the heartbeat features of the cardiac impulse signal are extracted in the beat-to-beat interval by means of matching clustering templates to obtain a second sequence; wherein the beat-to-beat interval sequence includes the first sequence and the second sequence.

[0057] Another feasible way is to extract the heartbeat characteristics of the ECG signal and the cardiac impact signal to obtain a beat-by-beat interval sequence, that is, the beat-by-beat interval sequence of multiple physiological signals can be extracted. Specifically, the ECG signal is subjected to beat-by-beat interval extraction in the above manner to obtain a first sequence. The cardiac impact signal is unstable and has a variable shape relative to the ECG signal, and the main peak J wave of the cardiac impact signal is easily affected by other simple harmonic waves with larger amplitudes. Therefore, after the collected cardiac impact signal is preprocessed by a bandpass filter, an adaptive template matching algorithm based on clustering is used to extract the beat-by-beat interval of the cardiac impact signal to obtain a second sequence, and the clustering algorithm can be Affinity Propagation (AP). Among them, the beat-by-beat interval sequence includes a first sequence and a second sequence, and the beat-by-beat interval sequence can be understood as a set of sequences after the beat-by-beat interval extraction of each signal.

[0058] Optionally, the above-mentioned extraction of heartbeat features of the electrocardiogram signal in the beat-by-beat interval by means of quadratic spline wavelet decomposition and R-wave peak location to obtain the first sequence can be specifically implemented by the following steps:

[0059] The electrocardiogram signal is preprocessed using a sliding average filter to remove high-frequency noise in the electrocardiogram signal; the preprocessed electrocardiogram signal is decomposed using a quadratic spline wavelet to obtain a wavelet decomposition signal; the wavelet decomposition signal is squared to obtain a square decomposition signal; the target peak point of the square decomposition signal is located; wherein the target peak point refers to a maximum value point in the square decomposition signal that exceeds a preset threshold; based on the target peak point, the R wave peak point of the electrocardiogram signal is searched within a preset window; wherein the R wave peak point is the maximum value point of all target value points included in the electrocardiogram signal within the preset window range; the time interval between two adjacent R wave peak points is calculated to obtain a first sequence.

[0060] It can be understood that the ECG signal is preprocessed using a sliding average filter to remove high-frequency noise in the ECG signal to reduce the influence of high-frequency noise on the positioning of ECG signal waveform feature points, and the preprocessed ECG signal is obtained. Specifically, a tenth-order sliding average filter can be used for denoising. The formula of the tenth-order sliding average filter is shown in formula (1). Subsequently, the preprocessed ECG signal is decomposed by a quadratic spline wavelet to obtain a wavelet decomposition signal. When performing wavelet decomposition, a scale with the characteristics of few coefficients, low computational complexity, and good robustness can be selected. Subsequently, the wavelet decomposition signal is squared to obtain a square decomposition signal (square wavelet decomposition ECG, SWECG) to highlight the peak amplitude of the decomposed signal waveform. The square operation can also widen the amplitude gap between the singular value peak and other interference. When using threshold positioning in the subsequent process, it is less susceptible to interference and has higher robustness than the differential threshold method. Then, locate the target peak point of the square decomposition signal, where the target peak point is the maximum value point in the square decomposition signal waveform that exceeds the preset threshold. The setting of the preset threshold is related to the maximum amplitude value of the preset window sampling point. The preset window size is determined by the normal range of heart rate and sampling rate, usually covering 1-3 heartbeat cycles, where the dynamic update of the preset threshold is shown in formula (2). Then, with the target peak point as the reference, find the R wave peak point of the ECG signal within the preset window, where the R wave peak point is the maximum value point of all target value points included in the ECG signal within the preset window range. The quadratic spline wavelet decomposition has a linear phase characteristic. There is a fixed time delay between the target peak point of the decomposed signal waveform and the R wave peak point of the ECG signal waveform. Therefore, the preset window size can be set according to the sampling rate. The maximum value point of the ECG signal within the preset window is the R wave peak point. Subsequently, the time interval between two adjacent R wave peak points is calculated to obtain the first sequence, where in the ECG signal, the R wave peak point marks the beginning of the ventricular depolarization process, that is, the starting point of heart contraction, and the beat-to-beat interval usually refers to the time interval between two adjacent R wave peaks, that is, the RR interval (RR interval). By detecting the R wave peak points in the ECG signal, a series of RR intervals can be obtained, and a series of RR intervals constitute a beat-to-beat interval sequence.

[0061] Formula (1)

[0062] In the formula, is the preprocessed ECG signal Middle The value of the sampling point, is the original ECG signal, and i is the index variable of the sliding window.

[0063] Formula (2)

[0064] In the formula, is the preset threshold, For a SWECG with a window size of 400 samples, is the value of SWECG at the nth sampling point.

[0065] Optionally, the above calculation of the time interval between two adjacent R wave peak points to obtain the first sequence can be specifically implemented by the following steps:

[0066] Using the sampling rate corresponding to the preset window and the heart rate range of the target user, determine whether the found R wave peak point has been missed or detected multiple times; if missed or detected multiple times, determine the historical interval sequence of the target user as the first sequence; or, if there is no missed or detected multiple times, calculate the time interval between two adjacent R wave peak points to obtain the first sequence.

[0067] It is understandable that when the ECG signal waveform is distorted, missed detection or multiple detection may occur during the search for the R wave peak point. In order not to affect the beat-by-beat interval sequence, the sampling rate corresponding to the preset window and the heart rate range of the target user are used to determine whether there is a missed detection or multiple detection. If there is a missed detection or multiple detection, the historical interval sequence of the target user is determined as the first sequence, that is, the historical numerical value is used to replace the RR interval value; if there is no missed detection or multiple detection, the time interval between the two R wave peaks is calculated to obtain the first sequence, that is, the RR interval value is calculated in real time.

[0068] For example, see Figure 2 , Figure 2 The schematic diagram of the process of extracting the beat-by-beat interval of the ECG signal provided by the embodiment of the present disclosure is as follows: Figure 2As shown in the figure, the beat-by-beat interval sequence extraction process includes: 1) obtaining a single-lead ECG signal, where a single-lead ECG signal refers to an ECG signal recorded using one lead (i.e., a pair of electrodes) and a single-lead ECG device is usually small in size and light in weight, easy to carry and use, and suitable for home health monitoring and personal health management; 2) performing a sliding mean filter on the single-lead ECG signal; 3) performing a quadratic spline wavelet decomposition on the filtered ECG signal; 4) performing a square on the decomposed waveform; 5) Locate the maximum point of the square decomposition signal; 6) Determine whether the peak value in the square decomposition waveform is greater than the set amplitude threshold; 7) If the peak value is greater than the set amplitude threshold, determine the target peak on the square decomposition waveform; 8) Based on the target peak, find the maximum value point on the ECG within the preset window and determine it as the R wave peak point; 8) Determine whether there is missed detection or multiple detection; 9) If there is no missed detection or multiple detection, calculate the RR interval based on the R wave peak point to obtain a beat-by-beat interval sequence; 10) If the peak value is less than the set amplitude threshold or there is missed detection or multiple detection, determine the historical interval as the RR interval. The detailed description of 1) to 10) involved in the above-mentioned beat-by-beat interval sequence extraction process can be found in the above-mentioned embodiment and will not be repeated here.

[0069] For example, see the result diagram of ECG signal feature point positioning and beat-by-beat interval sequence, specifically the result diagram of R wave peak point positioning and beat-by-beat interval sequence based on wavelet decomposition, including the result diagram under normal conditions as shown in Figure 3 (a), the result diagram under large disturbance conditions as shown in Figure 3 (b), and the result diagram under obvious noise conditions as shown in Figure 3 (c). Among them, the result diagram in each case represents the waveforms of ECG, SWECG and beat-by-beat interval sequence. When the ECG has a large disturbance, the R wave positioning based on wavelet decomposition is prone to peak deviation. In the ECG shown in Figure 3 (b), a large disturbance occurs in the range of 12000-8000 sampling points, resulting in the SWECG obtained by subsequent wavelet decomposition and square operation with peak deviation, but the subsequent error detection method avoids large disturbances in the beat-by-beat interval sequence. When the ECG high-frequency noise is large, as shown in Figure 3 (c), SWECG still has obvious peak advantages and a higher signal-to-noise ratio. Based on this advantage, the ECG beat-to-beat interval sequence with large noise is the same as the ECG beat-to-beat interval sequence under normal conditions, and can accurately locate each cardiac cycle.

[0070] Optionally, the preprocessed ECG signal is subjected to quadratic spline wavelet decomposition to obtain a wavelet decomposition signal, which can be specifically achieved through the following steps:

[0071] The equivalent filters corresponding to the wavelets at various scales are constructed based on low-pass filters and high-pass filters; the target scale is determined based on the frequency amplitude response curves of the equivalent filters at multiple scales at a preset sampling frequency and the frequency range of the QRS complex; the preprocessed ECG signal is subjected to wavelet decomposition to obtain a wavelet decomposition signal of the target scale.

[0072] It can be understood that the template of the quadratic spline wavelet is simple and has a reciprocal characteristic, and has a good performance in signal singular value detection. The Fourier transform of the wavelet function and the corresponding high-pass and low-pass filters are shown in formulas (3) to (5).

[0073] Formula (3)

[0074] Formula (4)

[0075] Formula (5)

[0076] In the formula, is the wavelet function, is a low-pass filter, is a high-pass filter, is the angular frequency, j is an imaginary unit, and j satisfies 2 =−1.

[0077] It can be inferred from formula (4) and formula (5) that the high-pass and low-pass filters corresponding to the quadratic spline wavelet function have low complexity and are easy to calculate. They can save running time when preprocessing a large number of ECG signals.

[0078] Understandably, the problem of downsampling will occur in the process of wavelet decomposition layer by layer. In order to ensure that the decomposed signal has the same sampling rate as the original signal for the convenience of subsequent feature point positioning, a porous algorithm is used to insert zero points to ensure that the length of the decomposed signal waveform after wavelet decomposition is consistent with the original signal, where j is the wavelet decomposition scale. For example, see Figure 4 , Figure 4 The flowchart of the multi-pore decomposition algorithm provided in the embodiment of the present disclosure is as follows. The multi-pore decomposition algorithm decomposes the signal into different resolution layers through multi-scale analysis, and each layer represents a different frequency component of the signal, such as Figure 4 The process by which the high-pass filter (G) and the low-pass filter (H) decompose the signal (X) into a plurality of resolution layers is shown, which will not be described in detail here.

[0079] It can be understood that the scale determination of wavelet decomposition is specifically described as follows: the expression of the quadratic spline wavelet scale function in the frequency domain is shown in formula (6), and its relationship with the low-pass filter is shown in formula (7). The relationship between the wavelet function, the high-pass filter and the scale function in the frequency domain is shown in formula (8). The equivalent filter corresponding to the wavelet at each scale is shown in formula (9). Specifically, the equivalent filter in formula (9) can be deduced based on the above formulas (4) and (5).

[0080] Formula (6)

[0081] Formula (7)

[0082] Formula (8)

[0083] Formula (9)

[0084] In the formula, is the scale function, Characterizes the relationship between the low-pass filter and the scaling function, Characterizes the relationship between the high-pass filter and the scaling function.

[0085] For example, see Figure 5 , Figure 5 The frequency amplitude response curve of the equivalent filter provided in the embodiment of the present disclosure at a sampling frequency of 250 Hz, that is, when the sampling rate is 250 Hz, the frequency response curve of the equivalent filter decomposed by the quadratic spline wavelet at the first 5 scales is as follows: Figure 5 As shown, the five scales are j=1, 2,...5.

[0086] For example, see Figure 6 , Figure 6 A schematic diagram of a 5-layer decomposition signal provided in an embodiment of the present disclosure, that is, a schematic diagram obtained by performing wavelet decomposition at a scale of j=1, 2....5, is shown by Figure 6 From the schematic diagram of the multi-scale decomposition signal shown, it can be seen that high-frequency information is mainly concentrated in the low-scale layer, while low-frequency information is mainly concentrated in the high-scale layer.

[0087] For example, see Figure 7 , Figure 7 The spectrum diagram of the 5-layer decomposition signal provided by the embodiment of the present disclosure is as follows: Figure 7 The figure shows the ECG and the spectrum of the decomposed waveform at 5 scales. Since the frequency range of the QRS complex is usually between 5Hz and 22Hz, Figure 7It can be seen that the energy of the QRS complex is concentrated on scales 3 and 4. Therefore, considering the time complexity and robustness of the algorithm, the decomposition waveform analysis of scale 3 is the best choice. Therefore, scale 3 is determined as the target scale, and the selection of scale 3 for wavelet decomposition analysis has the characteristics of few coefficients, low computational complexity, and good robustness. Among them, the QRS complex reflects the changes in the depolarization potential and time of the left and right ventricles. The first downward wave is the Q wave, the upward wave is the R wave, and the next downward wave is the S wave.

[0088] It can be understood that after determining the target scale, the preprocessed ECG signal is subjected to wavelet decomposition to obtain a wavelet decomposition signal of scale 3. The wavelet decomposition formula of scale 3 is shown in formula (10).

[0089] Formula (10)

[0090] Where s is the filtered signal, ws is the wavelet decomposition signal of scale 3, and n is the current position.

[0091] Optionally, the above-mentioned extraction of beat-by-beat heartbeat features of the heartbeat signal by matching the cluster template to obtain the second sequence can be specifically implemented by the following steps:

[0092] The cardiac impact signal is filtered by a bandpass filter to obtain a subtle vibration signal related to the heartbeat in the cardiac impact signal; all the maximum points in the subtle vibration signal are extracted, and data samples within a preset time period in the subtle vibration signal are intercepted with the maximum point as the center; the data samples are clustered by an affinity propagation clustering algorithm to obtain a clustering template; the heartbeat features of the subtle vibration signal are extracted beat by beat based on the clustering template to obtain a second sequence.

[0093] It is understandable that the beat-to-beat interval extraction of the cardiac impact signal is specifically achieved through the following steps: filter the cardiac impact signal (BCG) with a bandpass filter to obtain the subtle vibration component related to the heartbeat movement in the cardiac impact signal, wherein the bandpass filter can be a Butterworth bandpass filter of 8~24Hz, and the cardiac impact signal can be collected by a piezoelectric ceramic sensor. Subsequently, extract all the maximum points in the filtered cardiac impact signal (subtle vibration signal), and take the maximum point as the center, and intercept the signal of a preset time period from the subtle vibration signal as a sample. , build a data set based on all the intercepted samples, such as building a data sample set ,in, is the number of samples, For the The signal segments intercepted at the extreme value, wherein the preset time period can be the time period around 0.4 seconds. After obtaining the sample set, the sample set is clustered by affinity propagation algorithm. Clustering is performed to obtain at least one cluster template. Subsequently, the midpoint values ​​of each cluster template in at least one cluster template are averaged, and the cluster with the largest average value or within the maximum value range is determined as the target cluster, and then all target clusters are averaged to obtain the BCG average template. Subsequently, the heartbeat features of the subtle vibration signal are extracted beat by beat based on the BCG average template to obtain the second sequence. This algorithm for matching cluster templates ensures the accuracy of identifying the key components of the heartbeat signal and provides a basis for subsequent heart rate variability calculations.

[0094] For example, see Figure 8 , Figure 8 A schematic diagram of a clustering template and an average template provided in an embodiment of the present disclosure, wherein the sample set is clustered by an affinity propagation clustering algorithm. Clustering can be performed to obtain Figure 8 The BCG cluster template shown in . The midpoint values ​​of each cluster template are averaged, and the average value of the target cluster is the largest. Then all target clusters are averaged to obtain the following Figure 8 BCG average template shown.

[0095] For example, see Fig. 9 , Fig. 9 The result diagram of the characteristic point positioning and beat-by-beat interval sequence of the heartbeat signal provided in the embodiment of the present disclosure is as follows. The beat-by-beat interval extraction algorithm of the heartbeat signal adopted is tested by the BCG average template determined above. The original heartbeat signal collected by the piezoelectric ceramic front end, the heartbeat signal obtained after filtering, the correlation coefficient and the extracted cardiac cycle (second sequence) are shown in FIG. Fig. 9 shown.

[0096] S103: Analyze the heart rate variability according to the beat-to-beat interval sequence, and extract physiological characteristics of the target user from multiple angles.

[0097] It is understandable that, based on the above S102, heart rate variability analysis is performed according to the beat-by-beat interval sequence, and physiological characteristics are extracted from multiple angles. Among them, heart rate variability is composed of a continuous beat-by-beat interval sequence in cardiac activity, which describes the slight changes in the heartbeat cycle, reflects the degree of interaction between the sympathetic nerves and the vagus nerves in regulating the heart, and contains physiological information of the human cardiovascular system. HRV will also have different manifestations in different sleep phases. Therefore, HRV and its characteristics are often used as one of the research directions of low physiological load sleep staging.

[0098] Optionally, the above analysis of the heart rate variability according to the beat-to-beat interval sequence and the extraction of physiological characteristics of the target user from multiple angles can be specifically implemented through the following steps:

[0099] The beat-to-beat interval sequence is resampled to obtain a uniform sampling sequence; the DC signal in the uniform sampling sequence is removed to obtain a DC-free sequence; based on the DC-free sequence, the heart rate variability is analyzed from multiple angles to extract the physiological characteristics of the target user; wherein the multiple angles include linear, nonlinear and frequency domain; the physiological characteristics are standardized to obtain physiological characteristics that satisfy the standard normal distribution.

[0100] It is understandable that the beat-by-beat interval sequence is a non-uniformly sampled time series, the horizontal axis of the signal is the cumulative sum of the historical interval sequence at that point, and the vertical axis is the interval value at that point, where the interval value is the difference between the number of sampling points of the two R wave peaks. The premise of the frequency domain analysis of the beat-by-beat interval sequence is that the signal is a uniformly sampled sequence, so it is necessary to resample the beat-by-beat interval sequence to obtain a uniformly sampled sequence. The beat-by-beat interval range of a normal person is generally between 300 and 2000 ms, with a maximum frequency of 0.5 Hz, so the resampling frequency can be set to 2 Hz to satisfy the Shannon-Nyquist sampling theorem, and then the beat-by-beat interval sequence is resampled using cubic spline interpolation to obtain a uniform curve at the target sampling rate. It is understandable that when the beat-by-beat interval sequence includes sequences extracted from the beat-by-beat intervals of multiple physiological signals, heart rate variability analysis is performed on each sequence of each physiological signal, for example, in the case where the first sequence and the second sequence are extracted as described above, heart rate variability analysis is performed on the first sequence and the second sequence respectively, and the specific analysis order is not limited. After signal resampling, the uniform sampling sequence is subjected to DC removal. Specifically, the resampled beat-to-beat interval sequence (uniform sampling sequence) contains DC components. The DC signal needs to be removed before power spectrum analysis to avoid the situation where other AC frequency components cannot be observed due to excessive DC components during frequency domain analysis. After DC removal, HRV is analyzed from three perspectives: time domain, frequency domain, and nonlinearity to extract physiological features. The relevant physiological features are shown in Table 1. After feature extraction, the physiological features are standardized to obtain physiological features that meet the standard normal distribution. Specifically, physiological features include multidimensional features. Since the attributes of each dimensional feature are different and the numerical magnitude is different, there will be large feature differences. When the multidimensional features differ by several orders of magnitude, the learner will focus on features with large magnitudes and will not be able to learn from the remaining features with smaller differences. Therefore, in order to make the features between different dimensions comparable in numerical value and improve the accuracy of the classifier, the features of different dimensions need to be standardized. For example, the z-score (standard score) method can be used, that is, the data is standardized by subtracting the mean and dividing by the standard deviation. The processed data satisfies the standard normal distribution. Among them, Z-score is a statistical concept used to measure the standard deviation of a data point relative to the mean of the data set.

[0101] Table 1:

[0102]

[0103] For example, see Fig.10 , Fig.10 The heart rate variability analysis and feature extraction flow chart provided in the embodiment of the present disclosure is for the beat-by-beat interval sequence of a single physiological signal, specifically through the following steps: Fig.10 The following steps are shown for heart rate variability analysis and feature extraction: 1) obtaining a beat-by-beat interval sequence; 2) resampling the beat-by-beat interval sequence; 3) removing DC from the resampled beat-by-beat interval sequence; 4) extracting features from the DC-removed beat-by-beat interval sequence; 5) obtaining time domain features, frequency domain features, and nonlinear features; 6) standardizing the obtained multi-scale features. It is understandable that the specific description of the steps shown in 1) to 6) above refers to the above embodiment, and will not be repeated here.

[0104] S104. Classify the physiological characteristics through a pre-trained machine learning model to determine the sleep staging results of the target user.

[0105] It can be understood that, on the basis of the above S103, after performing heart rate deformation analysis and feature extraction on the beat-to-beat interval sequence, the physiological characteristics are classified by a pre-trained machine learning model to determine the sleep staging result of the target user, wherein sleep staging is usually divided into five main stages: N1 (non-REM sleep stage 1), N2 (non-REM sleep stage 2), N3 (non-REM sleep) and REM (rapid eye movement sleep) stages. Each stage has different physiological characteristics, and the specific division criteria of different stages are not limited. The sleep stage can also be set according to user needs. The machine learning model can be at least one model such as random forest, multi-layer perceptron, logistic regression, Adaboost, support vector machine, etc. A variety of learning models can be used to respectively perform sleep staging learning on the heart rate variability features extracted based on single-lead ECG signals and cardiac shock signals, and the results can be analyzed in non-independent subject mode and independent subject mode. Among them, the non-independent subject mode directly reads the data of all subjects, and then randomly allocates them in proportion, one part is the training set and the other part is the test set. However, after the data is divided by the non-independent subject mode, the data in the training set and the test set may come from the same subject, while the independent subject mode first divides the subject group into a training population and a test population, and then reads the data of the training population and the test population respectively to form a training set and a test set. Therefore, in the independent subject mode, the data in the training set and the test set cannot have an intersection of subjects, and the training accuracy can also be improved. That is, multiple models can be trained, and the one with higher accuracy is determined as the machine learning model for sleep staging prediction.

[0106] The embodiment of the present disclosure provides a sleep staging method, which collects the physiological signals of the user, extracts the beat-by-beat interval sequence of the physiological signals through quadratic spline wavelet decomposition, and performs heart rate variability analysis and physiological feature extraction. In addition, in the feature extraction stage, HRV is analyzed from multiple angles such as time domain, frequency domain and nonlinearity to extract the physiological characteristics of sleep. Then, the extracted physiological features are classified into children's sleep stages using a variety of traditional machine learning algorithms (such as random forest, support vector machine, logistic regression, etc.). The present disclosure is adapted to the physiological characteristics of children, which can significantly improve the accuracy of children's sleep staging and effectively process physiological signals under low signal-to-noise ratio conditions. On the other hand, compared with the traditional polysomnography monitoring method, the sleep staging method can be integrated into a portable monitoring device and is suitable for daily sleep monitoring of children; and its robust signal processing and sleep staging method can provide more reliable sleep data, provide scientific reference for relevant users, and help early detection and intervention of children's sleep problems.

[0107] Fig.11 The sleep staging device provided by the embodiment of the present disclosure can execute the processing flow provided by the sleep staging method embodiment, such as Fig.11 As shown, the device 1100 includes:

[0108] An acquisition unit 1101 is used to acquire a low-load physiological signal of a target user;

[0109] An extraction unit 1102 is used to extract heartbeat features of beat-by-beat intervals of physiological signals to obtain a beat-by-beat interval sequence; wherein the beat-by-beat interval refers to the time interval between two adjacent heartbeats;

[0110] An analysis unit 1103, configured to analyze the heart rate variability according to the beat-by-beat interval sequence and extract physiological characteristics of the target user from multiple perspectives;

[0111] The classification unit 1104 is used to classify the physiological characteristics through a pre-trained machine learning model to determine the sleep stage results of the target user.

[0112] Optionally, the physiological signals include electrocardiogram signals and cardiac shock signals.

[0113] Optionally, the extraction unit 1102 is used to:

[0114] The heartbeat features of the ECG signal are extracted beat by beat by beat by beat by quadratic spline wavelet decomposition and R wave peak location to obtain a first sequence;

[0115] Extracting the heartbeat features of the heartbeat signal in each beat by beat interval by matching the clustering template to obtain a second sequence;

[0116] The beat-by-beat interval sequence includes a first sequence and a second sequence.

[0117] Optionally, the extraction unit 1102 is used to:

[0118] The ECG signal is preprocessed using a sliding average filter to remove high-frequency noise in the ECG signal;

[0119] Performing quadratic spline wavelet decomposition on the preprocessed ECG signal to obtain a wavelet decomposition signal;

[0120] Square the wavelet decomposition signal to obtain a square decomposition signal;

[0121] Locating a target peak point of the square decomposition signal; wherein the target peak point refers to a maximum point in the square decomposition signal that exceeds a preset threshold;

[0122] Taking the target peak point as a reference, find the R wave peak point of the ECG signal within the preset window; wherein the R wave peak point is the maximum value point of all target value points included in the ECG signal within the preset window range;

[0123] The time interval between two adjacent R wave peak points is calculated to obtain the first sequence.

[0124] Optionally, the extraction unit 1102 is used to:

[0125] Using the sampling rate corresponding to the preset window and the heart rate range of the target user, determine whether the found R wave peak point has been missed or detected multiple times;

[0126] If there is missed detection or multiple detection, the historical interval sequence of the target user is determined as the first sequence; or, if there is no missed detection or multiple detection, the time interval between two adjacent R wave peak points is calculated to obtain the first sequence.

[0127] Optionally, the extraction unit 1102 is used to:

[0128] Construct the equivalent filters corresponding to the wavelets at each scale based on the low-pass filter and the high-pass filter;

[0129] Determine the target scale according to the frequency amplitude response curves of the equivalent filter at multiple scales at the preset sampling frequency and the frequency range of the QRS complex;

[0130] The preprocessed ECG signal is subjected to wavelet decomposition to obtain a wavelet decomposition signal of target scale.

[0131] Optionally, the extraction unit 1102 is used to:

[0132] The heartbeat signal is filtered by a bandpass filter to obtain a subtle vibration signal related to the heartbeat in the heartbeat signal;

[0133] Extract all the maximum points in the subtle vibration signal, and take the maximum point as the center to intercept the data samples within a preset time period in the subtle vibration signal;

[0134] Cluster the data samples using the affinity propagation clustering algorithm to obtain a clustering template;

[0135] The heartbeat features of the subtle vibration signal are extracted beat by beat according to the clustering template to obtain a second sequence.

[0136] Optionally, the analysis unit 1103 is used to:

[0137] Resample the beat-by-beat interval sequence to obtain a uniformly sampled sequence;

[0138] Removing the DC signal from the uniform sampling sequence to obtain a DC-free sequence;

[0139] Based on the DC-free sequence, the heart rate variability is analyzed from multiple angles to extract the physiological characteristics of the target user; the multiple angles include linear, nonlinear and frequency domains;

[0140] The physiological characteristics are standardized to obtain physiological characteristics that satisfy the standard normal distribution.

[0141] Fig.11 The sleep staging device of the illustrated embodiment can be used to implement the technical solution of the above-mentioned method embodiment. Its implementation principle and technical effect are similar and will not be described in detail here.

[0142] Fig.12 This is a schematic diagram of the structure of an electronic device provided by an embodiment of the present disclosure. Fig.12 , which shows a schematic diagram of the structure of an electronic device 1200 suitable for implementing the embodiment of the present disclosure. The electronic device 1200 in the embodiment of the present disclosure may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), vehicle terminals (such as vehicle navigation terminals), wearable electronic devices, etc., and fixed terminals such as digital TVs, desktop computers, smart home devices, etc. Fig.12 The electronic device shown is only an example and should not bring any limitation to the functions and scope of use of the embodiments of the present disclosure.

[0143] like Fig.12As shown, the electronic device 1200 may include a processing device (e.g., a central processing unit, a graphics processor, etc.) 1201, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1202 or a program loaded from a storage device 1208 to a random access memory (RAM) 1203 to implement the sleep staging method of the embodiment described in the present disclosure. In the RAM 1203, various programs and data required for the operation of the electronic device 1200 are also stored. The processing device 1201, the ROM 1202, and the RAM 1203 are connected to each other via a bus 1204. An input / output (I / O) interface 1205 is also connected to the bus 1204.

[0144] Typically, the following devices may be connected to the I / O interface 1205: an input device 1206 including, for example, a touch screen, a touch pad, a keyboard, a mouse, a camera, a microphone, an accelerometer, a gyroscope, etc.; an output device 1207 including, for example, a liquid crystal display (LCD), a speaker, a vibrator, etc.; a storage device 1208 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1209. The communication device 1209 may allow the electronic device 1200 to communicate with other devices wirelessly or by wire to exchange data. Although Fig.12 The electronic device 1200 is shown with various devices, but it should be understood that it is not required to implement or possess all the devices shown. More or fewer devices may be implemented or possessed instead.

[0145] In particular, according to an embodiment of the present disclosure, the process described above with reference to the flowchart can be implemented as a computer software program. For example, an embodiment of the present disclosure includes a computer program product, which includes a computer program carried on a non-transitory computer-readable medium, and the computer program contains program code for executing the method shown in the flowchart, thereby implementing the sleep staging method as described above. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device 1209, or installed from a storage device 1208, or installed from a ROM 1202. When the computer program is executed by the processing device 1201, the above-mentioned functions defined in the method of the embodiment of the present disclosure are performed.

[0146] It should be noted that the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In the present disclosure, a computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, device or device. In the present disclosure, a computer-readable signal medium may include a data signal propagated in a baseband or as part of a carrier wave, which carries a computer-readable program code. This propagated data signal may take a variety of forms, including but not limited to an electromagnetic signal, an optical signal, or any suitable combination of the above. Computer readable signal media may also be any computer readable medium other than computer readable storage media, which may send, propagate or transmit a program for use by or in conjunction with an instruction execution system, apparatus or device. The program code contained on the computer readable medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.

[0147] In some embodiments, the client and the server may communicate using any currently known or future developed network protocol such as HTTP (HyperText Transfer Protocol), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), an internet (e.g., the Internet), and a peer-to-peer network (e.g., an ad hoc peer-to-peer network), as well as any currently known or future developed network.

[0148] The computer-readable medium may be included in the electronic device, or may exist independently without being incorporated into the electronic device.

[0149] Optionally, when the above one or more programs are executed by the electronic device, the electronic device may also execute other steps described in the above embodiments.

[0150] Computer program code for performing the operations of the present disclosure may be written in one or more programming languages ​​or a combination thereof, including, but not limited to, object-oriented programming languages, such as Java, Smalltalk, C++, and conventional procedural programming languages, such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0151] The flow chart and block diagram in the accompanying drawings illustrate the possible architecture, function and operation of the system, method and computer program product according to various embodiments of the present disclosure. In this regard, each square box in the flow chart or block diagram can represent a module, a program segment or a part of a code, and the module, the program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some implementations as replacements, the functions marked in the square box can also occur in a sequence different from that marked in the accompanying drawings. For example, two square boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each square box in the block diagram and / or flow chart, and the combination of the square boxes in the block diagram and / or flow chart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0152] The units involved in the embodiments described in the present disclosure may be implemented by software or hardware, wherein the name of a unit does not, in some cases, limit the unit itself.

[0153] The functions described above herein may be performed at least in part by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: field programmable gate arrays (FPGAs), application specific integrated circuits (ASICs), application specific standard products (ASSPs), systems on chips (SOCs), complex programmable logic devices (CPLDs), and the like.

[0154] In the context of the present disclosure, a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, device, or equipment. A machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium. A machine-readable medium may include, but is not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any suitable combination of the foregoing. A more specific example of a machine-readable storage medium may include an electrical connection based on one or more lines, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.

[0155] It should be noted that, in this article, relational terms such as "first" and "second" are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or gateway that includes a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements inherent to such process, method, article or gateway. In the absence of further restrictions, the elements defined by the sentence "comprise one..." do not exclude the presence of other identical elements in the process, method, article or gateway that includes the elements.

[0156] The above description is only a specific embodiment of the present disclosure, so that those skilled in the art can understand or implement the present disclosure. 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 disclosure. Therefore, the present disclosure will not be limited to the embodiments described herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A sleep staging method, characterized in that: include: Obtain low-load physiological signals from target users; Extracting the heartbeat features of the beat-by-beat interval of the physiological signal to obtain a beat-by-beat interval sequence; wherein the beat-by-beat interval refers to the time interval between two adjacent heartbeats; Analyzing the heart rate variability according to the beat-by-beat interval sequence, and extracting the physiological characteristics of the target user from multiple perspectives; Classifying the physiological characteristics through a pre-trained machine learning model to determine the sleep stage result of the target user; The physiological signal includes an electrocardiogram signal and a cardiac impulse signal, and the step of extracting the beat-by-beat interval heartbeat features of the physiological signal to obtain a beat-by-beat interval sequence includes: The heartbeat features of the electrocardiogram signal are extracted in the beat-to-beat interval by means of quadratic spline wavelet decomposition and R-wave peak location to obtain a first sequence; the heartbeat features of the cardiac impulse signal are extracted in the beat-to-beat interval by means of matching clustering templates to obtain a second sequence; wherein the beat-to-beat interval sequence includes the first sequence and the second sequence.

2. The method according to claim 1, characterized in that The step of extracting the beat-by-beat heartbeat features of the electrocardiogram signal by using quadratic spline wavelet decomposition and R-wave peak location to obtain a first sequence includes: Preprocessing the ECG signal using a sliding average filter to remove high-frequency noise in the ECG signal; Performing quadratic spline wavelet decomposition on the preprocessed ECG signal to obtain a wavelet decomposition signal; Squaring the wavelet decomposition signal to obtain a square decomposition signal; Locating a target peak point of the square decomposition signal; wherein the target peak point refers to a maximum point in the square decomposition signal that exceeds a preset threshold; Taking the target peak point as a reference, searching for the R wave peak point of the electrocardiogram signal within a preset window; wherein the R wave peak point is the maximum value point of all target value points included in the electrocardiogram signal within the preset window range; The time interval between two adjacent R wave peak points is calculated to obtain the first sequence.

3. The method according to claim 2, characterized in that The step of calculating the time interval between two adjacent R wave peak points to obtain a first sequence includes: Using the sampling rate corresponding to the preset window and the heart rate range of the target user, determining whether the found R wave peak point has been missed or detected multiple times; If there is missed detection or multiple detection, the historical interval sequence of the target user is determined as the first sequence; or, if there is no missed detection or multiple detection, the time interval between two adjacent R wave peak points is calculated to obtain the first sequence.

4. The method according to claim 2, characterized in that: The step of performing quadratic spline wavelet decomposition on the preprocessed ECG signal to obtain a wavelet decomposition signal comprises: Construct the equivalent filters corresponding to the wavelets at each scale based on the low-pass filter and the high-pass filter; Determining a target scale according to frequency amplitude response curves of multiple scales of the equivalent filter at a preset sampling frequency and a frequency range of a QRS complex; The preprocessed ECG signal is subjected to wavelet decomposition to obtain a wavelet decomposition signal of the target scale.

5. The method according to claim 1, characterized in that The step of extracting beat-by-beat heartbeat features of the heartbeat signal by matching cluster templates to obtain a second sequence includes: Using a bandpass filter to filter the cardiac impact signal to obtain a subtle vibration signal related to the heartbeat in the cardiac impact signal; Extracting all the maximum value points in the subtle vibration signal, and taking the maximum value point as the center, intercepting data samples within a preset time period in the subtle vibration signal; Clustering the data samples using an affinity propagation clustering algorithm to obtain a clustering template; The heartbeat features of the subtle vibration signal are extracted in the beat-by-beat interval according to the clustering template to obtain a second sequence.

6. The method according to claim 1, characterized in that The analyzing the heart rate variability according to the beat-by-beat interval sequence and extracting the physiological characteristics of the target user from multiple angles includes: Resampling the beat-by-beat interval sequence to obtain a uniformly sampled sequence; Removing a DC signal from the uniform sampling sequence to obtain a DC-removed sequence; Analyzing the heart rate variability from multiple angles based on the DC removal sequence to extract the physiological characteristics of the target user; wherein the multiple angles include linear, nonlinear and frequency domains; The physiological characteristics are standardized to obtain physiological characteristics that satisfy a standard normal distribution.

7. A sleep staging device, characterized in that: include: An acquisition unit, used for acquiring a low-load physiological signal of a target user; An extraction unit, configured to extract the heartbeat features of the beat-by-beat interval of the physiological signal to obtain a beat-by-beat interval sequence; wherein the beat-by-beat interval refers to the time interval between two adjacent heartbeats; an analyzing unit, configured to analyze the heart rate variability according to the beat-to-beat interval sequence and extract physiological characteristics of the target user from multiple angles; A classification unit, used to classify the physiological characteristics through a pre-trained machine learning model to determine the sleep stage result of the target user; Wherein, the physiological signals include electrocardiogram signals and cardiac shock signals; The extraction unit is used to: extract the heartbeat features of the electrocardiogram signal in the beat-to-beat interval by means of quadratic spline wavelet decomposition and R-wave peak location to obtain a first sequence; extract the heartbeat features of the cardiac impulse signal in the beat-to-beat interval by means of matching clustering templates to obtain a second sequence; wherein the beat-to-beat interval sequence includes the first sequence and the second sequence.

8. An electronic device, characterized in that: include: Memory; processor; as well as Computer programs; The computer program is stored in the memory and is configured to be executed by the processor to implement the sleep staging method as claimed in any one of claims 1 to 6.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the sleep staging method according to any one of claims 1 to 6 are implemented.

Images