Sleep staging method, device, equipment and computer-readable storage medium

Abstract

The present disclosure relates to the technical field of brain wave signal processing, and more specifically, to a sleep staging method, apparatus, device, and computer-readable storage medium. The sleep staging method includes: determining the power spectral density of the brain wave signal of each sleep frame during sleep; performing dimensionality reduction processing on the power spectral density to obtain a reduced dimensionality data set; determining multiple characteristic signals of the brain wave signal of the sleep frame based on the power spectral density; for each sleep frame, performing kernel density estimation on the reduced dimensionality data set with weights determined according to each characteristic signal to obtain a probability density corresponding to the characteristic signal; and determining that the sleep frame belongs to the sleep stage corresponding to the characteristic signal when the probability density meets the preset condition corresponding to the characteristic signal, wherein different characteristic signals correspond to different sleep stages.

Description

Technical Field

[0001] The present disclosure relates to the technical field of brain wave signal processing, and in particular to a sleep staging method, apparatus, device, and computer-readable storage medium. Background Art

[0002] Sleep staging is crucial for assessing sleep quality and diagnosing sleep-related disorders. According to the American Academy of Sleep Medicine (AASM) standards, sleep can be divided into five stages: wake, stage N1 (falling asleep), stage N2 (light sleep), stage N3 (deep sleep), and REM (Rapid Eye Movement).

[0003] In the existing technology, sleep staging methods based on supervised learning have significant limitations: first, such methods rely on a large amount of manually labeled sleep data, resulting in high data collection and labeling costs; second, the manual labeling process may introduce subjective bias and label noise, thereby affecting model performance; finally, supervised learning methods often exhibit insufficient generalization ability when applied across data sets, and are difficult to adapt to individual differences and changes in data distribution under different collection environments. Summary of the Invention

[0004] To overcome the problems existing in the related art, the present disclosure provides a sleep staging method, apparatus, device and computer-readable storage medium, which can solve the above problems.

[0005] According to a first aspect of an embodiment of the present disclosure, a sleep staging method is provided, the method comprising: determining a power spectral density of an EEG signal of each sleep frame during sleep; performing dimensionality reduction processing on the power spectral density to obtain a reduced dimensionality data set; determining a plurality of characteristic signals of the EEG signal of the sleep frame based on the power spectral density; for each sleep frame, performing kernel density estimation on the reduced dimensionality data set with a weight determined according to each characteristic signal to obtain a probability density corresponding to the characteristic signal; and determining that the sleep frame belongs to the sleep stage corresponding to the characteristic signal when the probability density meets a preset condition corresponding to the characteristic signal, wherein different characteristic signals correspond to different sleep stages.

[0006] According to a second aspect of an embodiment of the present disclosure, a sleep staging device is provided, the device comprising: a power spectral density determination unit, configured to determine the power spectral density of the brain wave signal of each sleep frame during sleep; a dimensionality reduction unit, configured to perform dimensionality reduction processing on the power spectral density to obtain a dimensionality reduction data set; a characteristic signal determination unit, configured to determine a plurality of characteristic signals of the brain wave signal of the sleep frame based on the power spectral density; a kernel density estimation unit, configured to perform kernel density estimation on the dimensionality reduction data set for each sleep frame with a weight determined according to each characteristic signal, to obtain a probability density corresponding to the characteristic signal; a condition judgment unit, configured to determine that the sleep frame belongs to the sleep stage corresponding to the characteristic signal when the probability density meets a preset condition corresponding to the characteristic signal, wherein different characteristic signals correspond to different sleep stages.

[0007] According to a third aspect of an embodiment of the present disclosure, an electronic device is provided, comprising: a processor and a memory; the memory is used to store a computer program; and the processor is used to execute the sleep staging method as described in the first aspect by calling the computer program.

[0008] According to a fourth aspect of an embodiment of the present disclosure, a computer-readable storage medium is provided, on which a computer program is stored. When the program is executed by a processor, the sleep staging method as described in the first aspect is implemented.

[0009] According to a fifth aspect of an embodiment of the present disclosure, a computer program product is provided, comprising a computer program, wherein when the computer program is executed by a processor, the method as described in the first aspect is implemented.

[0010] The technical solutions provided by the embodiments of the present disclosure may have the following beneficial effects:

[0011] The sleep staging method proposed in the present disclosure is an unsupervised sleep staging method. After determining the power spectral density of the EEG signal for each sleep frame, a dimensionality reduction data set can be obtained through dimensionality reduction processing. Multiple characteristic signals of the EEG signal are then determined, with different characteristic signals corresponding to different sleep stages. For each sleep frame, the characteristic signal is used to determine the weight corresponding to the characteristic signal. Based on the weight, kernel density estimation is then performed on the dimensionality reduction data set to obtain a probability density distribution corresponding to the characteristic signal. If the probability density of any sleep frame meets the preset conditions corresponding to the characteristic signal, the sleep frame is determined to belong to the sleep stage corresponding to the characteristic signal.

[0012] The sleep staging method disclosed in the present invention has high accuracy and better generalization ability compared with supervised learning algorithms. It is suitable for portable EEG acquisition equipment and does not require technical personnel intervention, which can significantly improve the convenience and accuracy of sleep monitoring.

[0013] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] 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.

[0015] Figure 1 is a schematic flow chart of a sleep staging method according to an exemplary embodiment of the present disclosure.

[0016] Figure 2 FIG. 1 is an algorithm flow chart of a sleep staging method according to an exemplary embodiment of the present disclosure.

[0017] Figure 3 It is a schematic diagram of a dimension reduction data set according to an exemplary embodiment of the present disclosure.

[0018] Figure 4 3 is a schematic diagram of an identification area during the eyes-open stage of the awake period according to an exemplary embodiment of the present disclosure.

[0019] Figure 5 : is a schematic diagram of the identification area of the N2 stage and the N3 stage according to an exemplary embodiment of the present disclosure; wherein, Figure 5 (a) is a schematic diagram showing grouping sleep frames according to fast spindle power according to an exemplary embodiment of the present disclosure. Figure 5 (b) is a schematic diagram of the identification areas of the N2 stage and the N3 stage according to an exemplary embodiment of the present disclosure.

[0020] Figure 6 : is a schematic diagram of the identification area of the N3 phase according to an exemplary embodiment of the present disclosure; wherein, Figure 6 (a) is a schematic diagram of a slow wave signal according to an exemplary embodiment of the present disclosure. Figure 6 (b) is a schematic diagram of the identification area of stage N3 according to an exemplary embodiment of the present disclosure.

[0021] Figure 7 3 is a schematic diagram of the identification area during the eye-closed stage of wakefulness according to an exemplary embodiment of the present disclosure.

[0022] Figure 8 FIG. 1 is a schematic diagram of sleep stage correction according to an exemplary embodiment of the present disclosure.

[0023] Figure 9 The figure is a block diagram of a sleep staging device according to an exemplary embodiment of the present disclosure.

[0024] Figure 10This is a schematic block diagram of a sleep staging device according to an exemplary embodiment of the present disclosure. DETAILED DESCRIPTION

[0025] Exemplary embodiments will be described in detail herein, with examples illustrated in the accompanying drawings. In the following description, when referring to the drawings, identical numerals in different figures represent identical or similar elements, unless otherwise indicated. The embodiments described in the following exemplary embodiments are not intended to represent all possible embodiments consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.

[0026] The terms used in this disclosure are for the purpose of describing specific embodiments only and are not intended to limit the disclosure. As used in this disclosure and the appended claims, the singular forms "a," "an," "the," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.

[0027] It should be understood that although the terms first, second, third, etc. may be used in this disclosure to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other. For example, without departing from the scope of this disclosure, first information may also be referred to as second information, and similarly, second information may also be referred to as first information. Depending on the context, the word "if" as used herein may be interpreted as "at the time of" or "when" or "in response to determining."

[0028] In order to solve the above technical problems, the present disclosure proposes a sleep staging method.

[0029] Figure 1 This is a schematic flow chart illustrating a sleep staging method according to an embodiment of the present disclosure. The sleep staging method can be executed by a terminal. The sleep staging method processes acquired brainwave signals to divide sleep into stages. The terminal includes but is not limited to mobile phones, tablets, wearable devices, sensors, IoT devices, and other communication devices.

[0030] like Figure 1 As shown, sleep staging methods include:

[0031] In step S101, the power spectrum density of the brain wave signal of each sleep frame during sleep is determined;

[0032] In step S102, dimensionality reduction processing is performed on the power spectrum density to obtain a dimensionality reduction data set;

[0033] In step S103, multiple characteristic signals of the brain wave signal of the sleep frame are determined according to the power spectrum density;

[0034] In step S104, for each sleep frame, kernel density estimation is performed on the dimensionality reduction data set using weights determined according to each characteristic signal to obtain a probability density corresponding to the characteristic signal;

[0035] In step S105 , when the probability density satisfies a preset condition corresponding to the characteristic signal, it is determined that the sleep frame belongs to the sleep stage corresponding to the characteristic signal, wherein different characteristic signals correspond to different sleep stages.

[0036] In some embodiments, the power spectral density of the brainwave signal is determined for each sleep frame during sleep.

[0037] Electroencephalogram (EEG) signals can be collected from the target subject during sleep. The EEG signals collected from the target subject are continuous. To facilitate subsequent data processing, the continuous EEG signals can be segmented into multiple sleep frames of a certain duration (for example, 30 seconds).

[0038] After the sleep frame is determined, the power spectral density (PSD) of the brain wave signal in the sleep frame may be determined.

[0039] In some embodiments, the power spectral density is subjected to dimensionality reduction processing to obtain a dimensionality reduced data set.

[0040] The determined power spectral density is high-dimensional data, which is relatively difficult to process. Therefore, the power spectral density (PSD) can be processed to reduce the dimensionality of the data to obtain a reduced-dimensionality dataset.

[0041] For example, the Uniform Manifold Approximation and Projection (UMAP) algorithm can be used to reduce the dimensionality of the EEG power spectral density. Based on the reduced dimensionality dataset, sleep frames can be visualized in the PSD feature space. The reduced dimensionality dataset is the projection of the high-dimensional PSD in the two-dimensional space. The projected PSD vector can be simply referred to as PSD. UMAP .

[0042] In some embodiments, multiple characteristic signals of the brain wave signal of the sleep frame are determined according to the power spectrum density.

[0043] Based on the power spectral density, a variety of characteristic signals can be determined. The characteristic signals can be used to describe the characteristics of the brain wave signal in a certain aspect, such as describing the characteristics of the brain wave signal in a specific frequency band.

[0044] The target subject's brain wave signals will have different performances in specific frequency bands during different sleep stages. Therefore, based on the characteristic signals of each sleep frame during sleep, the sleep stage corresponding to the target subject's sleep frame can be determined.

[0045] In some embodiments, for each sleep frame, kernel density estimation is performed on the dimensionality reduction dataset using a weight determined according to each characteristic signal to obtain a probability density corresponding to the characteristic signal.

[0046] The probability density value of each sleep frame relative to the entire sleep period can be determined. Since the characteristic signals corresponding to the brain wave signals in different sleep stages are different, the probability density of different sleep frames for specific characteristic signals is different.

[0047] Therefore, we can first determine the weights for the kernel density estimation (KDE) algorithm based on the characteristic signals. Different types of characteristic signals have different ways of determining weights, which will be described in detail later and will not be elaborated here.

[0048] After the weight corresponding to the characteristic signal is determined, kernel density estimation may be performed on each sleep frame based on the weight to determine the probability density of the sleep frame corresponding to the characteristic signal.

[0049] In some embodiments, when the probability density satisfies a preset condition corresponding to the characteristic signal, it is determined that the sleep frame belongs to the sleep stage corresponding to the characteristic signal, wherein different characteristic signals correspond to different sleep stages.

[0050] After determining the probability density of each sleep frame corresponding to a particular characteristic signal, a judgment can be made based on the preset conditions corresponding to the characteristic signal. If the preset conditions corresponding to the characteristic signal are met, the sleep frame belongs to the sleep stage corresponding to the characteristic signal. From all sleep frames, the portion of sleep frames that belong to the sleep stage corresponding to the characteristic signal can be determined. Similarly, multiple characteristic signals can correspond to multiple sleep stages, thus realizing sleep staging.

[0051] It should be noted that the present disclosure may also determine the sleep frames belonging to the sleep stage corresponding to the characteristic signal during a sleep period based on only one characteristic signal. For example, if only sleep frames belonging to stage N2 need to be determined during a sleep period, the determination may be made based solely on the characteristic signal corresponding to stage N2.

[0052] The sleep staging method proposed in this disclosure accurately stages sleep by identifying characteristic signals and the probability density of each sleep frame under the characteristic signals, based on the different characteristic manifestations of EEG signals during sleep at different sleep stages. Unlike supervised classification methods, this method does not require the calculation of large amounts of sample data, resulting in relatively low costs, better generalization, and relatively low terminal computing power requirements. It is suitable for portable EEG acquisition equipment, does not require the participation of technical personnel, and is highly convenient and accurate.

[0053] In some embodiments, the characteristic signal includes at least one of the following: gamma power, fast spindle power, oscillation signal power, slow wave signal.

[0054] Gamma power (gamma) is the power of brain wave signals in the gamma frequency band (25-50Hz).

[0055] Fast spindle power may be the power of a brain wave signal within a fast spindle frequency band.

[0056] The slow oscillation (SO) signal can be determined based on a waveform detection algorithm.

[0057] The power spectrum density can be separated to determine the power of the oscillating signal.

[0058] The specific method for determining the above characteristic signals will be introduced in detail later and will not be elaborated here.

[0059] Since different characteristic signals correspond to different sleep stages, the method proposed in the present disclosure can be to determine the sleep frames belonging to the corresponding sleep stage during the sleep period through any of the above characteristic signals, or to determine the sleep frames corresponding to different sleep stages during the sleep period through the above multiple characteristic signals.

[0060] In some embodiments, the sleep stage corresponding to the gamma power is the eyes-open stage of the awake period; the sleep stage corresponding to the fast spindle power is stage N2 and stage N3; the sleep stage corresponding to the oscillation signal power is stage N2, stage N3 and the eyes-closed stage of the awake period; the sleep stage corresponding to the slow wave signal is stage N3.

[0061] Gamma power is used to determine the eyes-open phase of wakefulness open The power of fast spindles is used to determine the N2 and N3 stages. The power of oscillation signals is used to determine the N2 and N3 stages and the closed-eye stage of the wake period. close ); Slow wave signal is used to determine N3 stage.

[0062] Figure 21 is an algorithm flow chart of a sleep staging method according to an embodiment of the present disclosure.

[0063] like Figure 2 As shown in the figure, for the original signal (original brain wave signal), the power spectral density (PSD) can be determined first, and then the determined PSD is subjected to dimensionality reduction (UMAP) processing to obtain a dimensionality reduction data set (PSD UMAP Based on the multiple characteristic signals identified, kernel density estimation (KDE) is performed on the dimensionality reduction dataset to determine the probability density of the sleep frame corresponding to the characteristic signal.

[0064] Based on the mutual cooperation of multiple characteristic signals, the accurate sleep stage of the sleep frame can be determined.

[0065] like Figure 2 As shown in the figure, based on the fast spindle power, the sleep frame set belonging to the N2 and N3 stages can be determined, and based on the slow wave signal, the sleep frame set belonging to the N3 stage can be determined, thereby distinguishing the N2 and N3 stages based on the N2 and N3 stages determined by the fast spindle waves; on the other hand, the eye-opening stage of the awake period can be determined based on the gamma power. open ), based on the oscillation signal ( ) can determine the N2 stage, N3 stage and the closed eye stage of the wake period (wake colse ), and then remove the already clear N2 and N3 stages to determine the eyes-closed stage of the wakeful period.

[0066] The remaining sleep frames belong to the N1 stage and the REM stage. In subsequent embodiments of the present disclosure, smoothing processing can be performed, and the N1 stage and the REM stage can be further distinguished based on the smoothing processing results, thereby realizing sleep staging.

[0067] In some embodiments, determining the power spectral density of the brain wave signal of each sleep frame during sleep includes: collecting the brain wave signal during sleep; filtering and resampling the brain wave signal, and dividing the processed brain wave signal into multiple sleep frames lasting a first duration; determining the power spectral density of the brain wave signal of each sleep frame.

[0068] Specifically, a single-channel frontal EEG signal can be extracted from the subject from a certain time before sleep (e.g., 30 minutes) to a certain time after sleep (e.g., 30 minutes). The EEG acquisition equipment can determine the EEG electrode placement according to the international 10-20 system. For example, acceptable EEG lead locations include F4-M1, F3-M2, and Fz-Cz. The EEG signal can be sampled at a frequency of 200 Hz or higher (e.g., 500 Hz).

[0069] Filtering can include bandpass filtering (0.1-50Hz) and EEG artifact removal to improve the quality of EEG signals. After filtering, the processed EEG signals can be resampled to optimize data compatibility and reduce data computation.

[0070] Figure 3 It is a schematic diagram of a dimension reduction data set according to an exemplary embodiment of the present disclosure.

[0071] After noise reduction, the collected EEG signals during the entire sleep period can be segmented into multiple sleep frames, each of which can be of the first duration (e.g., 30 seconds). The Welch method (Welch power spectrum estimation method) can then be used to calculate the power spectral density (PSD) of each sleep frame, and the unit of the power spectral density can be converted to a logarithmic unit of decibels (dB).

[0072] In order to visualize the sleep frames in the PSD feature space, the UMAP technique can be used to project the high-dimensional PSD into the two-dimensional space. The projected PSD vector is referred to as PSD. UMAP . Figure 3 The PSD is shown after dimensionality reduction. UMAP ),in, Figure 3 Each point in the middle graph corresponds to a sleep frame and the EEG signal within the duration of the sleep frame.

[0073] In some embodiments, a kernel density estimation (KDE) algorithm may be used to determine the probability density of sleep frames corresponding to different sleep stages.

[0074] Based on the probability density, the distribution of different sleep stages can be determined.

[0075] The specific kernel density estimation algorithm is as follows:

[0076]

[0077] Any sleep frame The probability density of It can be calculated by the above formula, where is the sample size, is the total number of sleep frames, represents the i-th observation in the sample, is the kernel function, is the bandwidth parameter, Indicates assignment to The weight of different feature signals, weight Not the same either.

[0078] Based on the kernel density estimation algorithm, it is possible to determine the total sleep time of any sleep frame x during the entire sleep period ( , i=1 to n).

[0079] As an example, the Gaussian kernel can be selected as the kernel function. The expression of the Gaussian kernel function is:

[0080]

[0081] Among them, e is the natural logarithm and π is the circumference of a circle.

[0082] The bandwidth parameter in formula (1) Determines the width of the kernel function and affects the estimation of the probability density. In some embodiments, the bandwidth parameter b can be determined using the Scott method, which is:

[0083]

[0084] in, is a spatial dimension. In the embodiment of the present disclosure, The value is 2.

[0085] In formula (1) The PSD corresponding to each sleep frame UMAP Substitute in. Weight The quantitative characteristic values of each sleep stage are substituted into the quantified characteristic values, which can be determined based on characteristic signals, such as gamma power, fast spindles, slow wave oscillations (SO) and (The oscillation component of PSD, i.e. the oscillation signal power). The specific weights corresponding to each characteristic signal The determination method is included in the subsequent introduction of each characteristic signal.

[0086] The following describes how to determine the weight based on each characteristic signal and perform kernel density estimation to determine the corresponding sleep stage.

[0087] (1) Gamma power

[0088] The eyes-open phase of wakefulness can be determined based on gamma power.

[0089] Figure 4 3 is a schematic diagram of an identification area during the eyes-open stage of the awake period according to an exemplary embodiment of the present disclosure.

[0090] Specifically, the power of each sleep frame in the gamma frequency band (25 Hz-50 Hz) can be determined, and the PSD average value can be determined as an indicator to measure the gamma power.

[0091] Furthermore, the Otsu threshold segmentation method (maximum inter-class variance method) can be used to determine the high and low gamma power discrimination threshold th based on the determined gamma power. Based on this threshold, sleep frames can be divided into high and low groups. The Otsu threshold segmentation method is an automatic global threshold selection algorithm based on the image grayscale histogram, which can automatically determine the optimal segmentation threshold without manual intervention.

[0092] The discrimination threshold th (threshold) divides the sleep frames corresponding to the gamma power into two groups, the ones on the left side of the dotted line mark th are the low gamma power group, and the ones on the right side of the dotted line mark th are the high gamma power group.

[0093] After grouping the sleep frames according to gamma power, the average gamma power of the high gamma power group can be calculated ( ). For any sleep frame , the weight corresponding to the gamma power This can be determined by:

[0094] If the sleep frame Gamma power Greater than the average of the high gamma power group , then the weight =1;

[0095] If the gamma power is less than the discrimination threshold th, then =0;

[0096] like , then determine the gamma power of the sleep frame The percentile rank within the high gamma power group is weighted to twice the percentile rank value. For example, if its percentile rank is 30%, then =0.6. It should be noted that in this embodiment, the weight is set to twice the percentile ranking value, which can more intuitively show the difference from the low gamma power group, thereby facilitating subsequent sleep staging. According to actual needs, it can also be adjusted to 1 times, 3 times, etc., and this disclosure does not impose any restrictions on this.

[0097] The weights determined for each sleep frame , substituted into the formula (1) of the above kernel density estimation algorithm to determine each sleep frame separately Probability density value corresponding to gamma power Then determine the sleep frame based on the preset conditions corresponding to the gamma power Whether it belongs to the eyes-open stage of the wakefulness period.

[0098] In some embodiments, when the characteristic signal includes the gamma power, the preset condition is that the probability density of the sleep frame is greater than a first multiple of the maximum probability density of all sleep frames, and the first multiple is less than 1 and greater than 0.

[0099] Specifically, if any sleep frame Probability density value under the weight corresponding to gamma power , greater than the maximum value of the probability density values of all sleep frames under gamma power If the first multiple of (for example, 10%) of any sleep frame is exceeded, it is determined that the sleep frame belongs to the eyes-open stage of the wakefulness period.

[0100] like Figure 4 As shown, determine any sleep frame The ratio of the probability density value to the maximum probability density value is displayed in the dimensionality reduction data set. The larger the ratio, the darker the color. The darker sleep frame cluster on the left side of the figure is the eye-opening stage of the awake period (wake open ), and the remaining sleep frame clusters belong to the sleep stage to be determined (others).

[0101] (2) Fast spindle power

[0102] The N2 and N3 phases can be determined based on the fast spindle power.

[0103] Figure 5 3 is a schematic diagram of identification areas of the N2 and N3 stages according to an exemplary embodiment of the present disclosure.

[0104] In some embodiments, the Irregular Resampling Auto-Spectral Analysis (IRASA) method can be used to decompose the power spectral density (PSD) into oscillatory components ( ) and fractal components ( ).

[0105]

[0106] In EEG signals, fractal components Usually corresponds to background EEG noise, while the oscillatory component This corresponds to the oscillation signal in the EEG signal, such as the spindle wave signal. Since the processing result determined by the IRASA algorithm is noisy, a two-dimensional Gaussian filter can be used to filter the oscillation component. Perform noise reduction processing.

[0107] In some embodiments, the distribution of the oscillation frequency corresponding to the highest oscillation power in each sleep frame may be statistically analyzed.

[0108] For each sleep frame, the noise reduction process The highest oscillation frequency in medium power For this oscillation frequency , use the kernel density estimation (KDE) algorithm mentioned in the above example to fit the frequency distribution, find the personalized peak frequency closest to 14Hz . Define personalized fast spindle power ( ) for ±1 Hz frequency band The average value of .

[0109] After fitting with the kernel density estimation algorithm, two peaks can be found, one of which is closer to 14 Hz. Therefore, the peak closer to 14 Hz can be used as the frequency band within ± 1 Hz. The average value of the fast spindle power .

[0110] The sleep frames can be divided into two groups using a preset first threshold, and the sleep frames of the two groups are respectively Use Gaussian distribution for fitting, where the higher The mean of a set of sleep frames is associated with stages N2 and N3, and this mean can be recorded as The standard deviation of the group is recorded as ; The other set of sleep frames corresponds to other sleep stages, and the mean is recorded as , the standard deviation is recorded as .

[0111] like Figure 5 As shown in (a), Figure 5 (a) With 0 as the threshold, the sleep frame on the left The average value is low, and this group of sleep frames corresponds to other sleep stages; the sleep frames on the right The average value is high, and this group of sleep frames corresponds to stages N2 and N3.

[0112] In some embodiments, when the characteristic signal includes the fast spindle power, the preset condition is that the probability density of the sleep frame under the first weight is greater than the probability density under the second weight, wherein the first weight is determined according to the cumulative distribution of the fast spindle power of the sleep frames in which the fast spindle power is greater than a first threshold, and the second weight is determined according to the cumulative distribution of the fast spindle power of the sleep frames in which the fast spindle power is less than the first threshold.

[0113] The first weight is based on the higher The weight determined by the mean can be recorded as , the second weight is based on the lower The weight determined by the mean can be recorded as .

[0114] The weight of each embedded point can be determined based on the cumulative distribution function of the Gaussian distribution and .

[0115] The formula is:

[0116]

[0117]

[0118]

[0119] In determining the weight and After that, the weight can be substituted into the kernel density estimation algorithm to determine the weight of each sleep frame. and The probability density value under and .

[0120] If any sleep frame Satisfy the preset conditions, the preset conditions are > , it is determined that any sleep frame belongs to the N2 period and the N3 period.

[0121] like Figure 5 (b) shows that Figure 5 In (b), the sleep frames that meet the preset conditions are shown. It can be determined that the sleep frames that meet the preset conditions on the right are stages N2 and N3.

[0122] (3) Slow wave signal

[0123] Based on the slow wave signal, stage N3 can be determined.

[0124] Figure 6 1 is a schematic diagram of an identification area of the N3 phase according to an exemplary embodiment of the present disclosure.

[0125] Slow wave signals (SO) can be detected using the YASA algorithm (Yet Another Spindle Algorithm). Detection criteria include: a frequency range of 0.5 Hz to 2.0 Hz, a peak-to-peak amplitude greater than 75 μV, a negative deflection duration of 300 ms to 1500 ms, a positive deflection duration of 100 ms to 1000 ms, a negative peak value less than -10 μV, and a positive peak value greater than 10 μV.

[0126] Based on the above-mentioned slow wave detection criteria of amplitude, the time period in which slow waves appear in the brain wave signal during sleep can be detected. Figure 6 As shown in (a), in the EEG signal, the time period marked is the time period when slow waves (SO) appear.

[0127] After determining the time period of the slow wave, the slow wave percentage of each sleep frame can be determined. The slow wave percentage is the ratio of the slow wave duration to the sleep frame duration (eg, 30 seconds).

[0128] Because electrooculogram (EOG) and motion artifacts can affect SO detection, the KDE weight for each sleep frame can be determined as the difference between the slow wave percentage and a preset value (e.g., 10%). Specifically, the KDE weight for each sleep frame can be determined by subtracting the preset value from the slow wave percentage. This embodiment eliminates interference caused by motion artifacts, thereby improving the accuracy of the determined weights.

[0129] After determining the weight corresponding to each sleep frame, the weight of each sleep frame can be determined based on the weight. The probability density of . And in the probability density When the preset condition is met, it is determined that the sleep frame belongs to the N3 phase.

[0130] In some embodiments, when the characteristic signal includes the slow wave signal, the preset condition is that the probability density of the sleep frame is greater than the second multiple of the maximum probability density among all sleep frames, and the distance to the identified N2 and N3 stages is closer than other identified sleep stages, wherein the second multiple is less than 1 and greater than 0.

[0131] In determining each sleep frame The probability density of After that, if any sleep frame Greater than the maximum probability density in all sleep frames , and the second multiple is less than 1 and greater than 0, then any sleep frame can be determined as a slow wave candidate frame.

[0132] For each slow wave candidate frame, its relative position in the UMAP space with the identified eye-open phase of the wake period can be further determined. open ), N2 and N3, and the remaining unidentified stages (usually corresponding to Wake close , REM and N1 phases) and assigns the slow wave candidate frames to the nearest region.

[0133] If the slow wave candidate frames are allocated to the N2 and N3 periods, that is, the slow wave candidate frames are closest to the identified N2 and N3 periods, then these slow wave candidate frames are determined to belong to the N3 period.

[0134] Since the N2 stage and the N3 stage have been determined, the other sleep frames that are not identified as the N3 stage in the N2 stage and the N3 stage can be determined to belong to the N2 stage.

[0135] like Figure 6 As shown in (b), N3 stage can be determined based on slow wave signal, combined with Figure 5 The N2 stage and N3 stage determined in (b) can determine the N2 stage.

[0136] (4) Oscillation signal power

[0137] Based on the oscillation signal power, the N2 phase, N3 phase and Wake phase can be determined. close Expect.

[0138] Figure 7 3 is a schematic diagram of the identification area during the eye-closed stage of wakefulness according to an exemplary embodiment of the present disclosure.

[0139] In an embodiment of determining the fast spindle power, the PSD can be decomposed based on the IRASA algorithm to obtain In this After Gaussian filtering, the frequency of each sleep frame in the 5Hz-20Hz frequency band can be determined. Standard deviation , the standard deviation can be used to quantify the intensity of oscillation activity in this frequency band.

[0140] Different sleep stages There are differences, wake close Both N2 and N3 have strong oscillation power, so wake close and N2 and N3 It is also higher than other sleep stages.

[0141] In determining Afterwards, you can The sleep frames are divided into high oscillation activity group and low oscillation activity group by fitting the distribution of the two groups of sleep frames. The mean and standard deviation of the high oscillation activity group can be recorded as and , the mean and standard deviation of the low oscillatory activity group can be written as and . According to the cumulative distribution function of the Gaussian distribution, a weight is assigned to each embedded point and :

[0142]

[0143]

[0144] Each sleep frame can be based on the weight (third weight) and (Fourth weight), two probability densities are determined by the above KDE algorithm and , when the two probability densities of any sleep frame meet certain preset conditions, it can be determined that any sleep frame belongs to stage N2, stage N3 and wake close After excluding sleep frames that have been identified as stages N2 and N3, the remaining sleep frames have alpha wave activity and can be determined to belong to wake close Expect.

[0145] In some embodiments, when the characteristic signal includes the oscillation signal power, the preset condition is that the probability density of the sleep frame under the third weight is greater than the probability density under the fourth weight, wherein the third weight is determined based on the cumulative distribution of the oscillation signal power of the sleep frames whose oscillation signal power is greater than the second threshold, and the fourth weight is determined based on the cumulative distribution of the oscillation signal power of the sleep frames whose oscillation signal power is less than the second threshold.

[0146] Specifically, according to the third weight ( ) determines the probability density ( ), and according to the fourth weight ( ) determines the probability density ( ), if any sleep frame satisfy , then the sleep frame ( ) is identified as a sleep frame with oscillatory activity, which belongs to stage N2, stage N3 and wake close Expect.

[0147] like Figure 7 As shown, based on the oscillation signal power, it can be determined that it belongs to the N2 period, N3 period and wake period. close The remaining wake frames can be determined by combining the sleep frames of N2 and N3 determined based on the fast spindle signal power. close Expect.

[0148] Based on the relevant embodiments of the above characteristic signals, sleep frames belonging to corresponding sleep stages can be determined. It should be noted that the above characteristic signals can be used alone to determine corresponding sleep stages, or can be combined with other characteristic signals to more accurately determine more sleep stages. This disclosure does not impose any restrictions on this.

[0149] In some embodiments, when the characteristic signal includes gamma power, fast spindle power, oscillation signal power and slow wave signal, the method further includes: determining a sleep frame for which the sleep stage has not been determined as a sleep frame to be identified; determining the sleep stage corresponding to the sleep frame to be identified as the REM stage, and smoothing the sleep frame to be identified; and adjusting the sleep stage corresponding to some sleep frames in the sleep frame to be identified whose distance from the N1 stage is less than a distance threshold to the N1 stage according to the smoothing result.

[0150] In the case where the characteristic signals include gamma power, fast spindle power, oscillation signal power and slow wave signal, based on the above embodiment, it can be determined that the sleep frame belongs to N2 stage, N3 stage, wake stage, open Period and wake close period, the remaining sleep frames belong to the REM period or N1 period.

[0151] In this regard, in this embodiment, these unrecognized sleep frames may be determined as sleep frames to be recognized, and all the sleep frames to be recognized may be preliminarily assumed to be REM periods.

[0152] Since the sleep stage does not jump during sleep and is relatively continuous with the previous and next sleep stages, it is possible to determine whether the sleep frame to be identified belongs to the REM stage or the N1 stage through smoothing processing.

[0153] In some embodiments, after assuming that all sleep frames to be identified belong to the REM period, a convolution smoothing process may be performed on the sleep stage curve.

[0154] For example, the convolution kernel length may be 20, and the corresponding duration may be 10 minutes. After the convolution smoothing process, a convolution smoothed state curve may be obtained.

[0155] The position of the smooth curve belonging to the REM period is judged. If the sleep frame is closer to the N1 period and the current REM period duration is less than a preset value (for example, 10 minutes), the sleep frame is readjusted from the REM period to the N1 period.

[0156] Figure 8 FIG. 1 is a schematic diagram of sleep stage correction according to an exemplary embodiment of the present disclosure. Figure 8As shown, all sleep frames to be identified are assumed to be REM periods, and the dotted lines in the figure represent the sleep frames assumed to be REM periods; then, the sleep stage curve is smoothed to obtain a smooth curve, and the smooth curve in the figure is represented by a curved dotted line; finally, based on the smoothing result, for the sleep frames assumed to be REM periods, it is determined whether the sleep frames are close to N1 period or close to REM period, and those close to N1 period are adjusted to N1 period. The black solid line in the figure represents the final adjustment result. It can be seen that in the entire smooth curve graph, there are multiple sleep frames with dotted lines in REM period that are eventually readjusted to N1 period.

[0157] In some embodiments, adjusting the sleep stages corresponding to some sleep frames in the sleep frames to be identified whose distance from stage N1 is less than a distance threshold to stage N1 based on the smoothing processing result includes: adjusting the sleep stages corresponding to some sleep frames one by one, and re-performing smoothing processing after each adjustment.

[0158] The sleep frames to be identified can be adjusted one by one in chronological order.

[0159] When a sleep frame to be identified is adjusted from REM to N1, a smooth curve can be re-determined, and based on the newly determined smooth curve, it is determined whether the next sleep frame to be identified needs to be adjusted from REM to N1. This process is repeated until all sleep frames to be identified are traversed.

[0160] When any sleep frame is adjusted from REM stage to N1 stage, the adjusted sleep frame will cause certain interference to the smooth curve. Therefore, the smooth curve is re-determined based on the adjusted sleep frame, and the next sleep frame to be identified is adjusted based on the newly determined smooth curve. This method has higher accuracy.

[0161] Corresponding to the embodiments of the sleep staging method disclosed herein, the present disclosure also provides embodiments of corresponding sleep staging devices.

[0162] See Figure 9 , Figure 9 FIG. 1 is a block diagram of a sleep staging device according to an embodiment of the present disclosure. Figure 9 As shown, the sleep staging device includes:

[0163] a power spectrum density determination unit 910 , configured to determine the power spectrum density of the brain wave signal of each sleep frame during sleep;

[0164] A dimensionality reduction unit 920 is configured to perform dimensionality reduction processing on the power spectrum density to obtain a reduced dimensionality data set;

[0165] a characteristic signal determining unit 930, configured to determine a plurality of characteristic signals of the brain wave signal of the sleep frame according to the power spectrum density;

[0166] a kernel density estimation unit 940 configured to perform kernel density estimation on the dimensionality reduction dataset for each sleep frame using a weight determined according to each characteristic signal to obtain a probability density corresponding to the characteristic signal;

[0167] The condition judgment unit 950 is configured to determine that the sleep frame belongs to the sleep stage corresponding to the characteristic signal when the probability density meets the preset condition corresponding to the characteristic signal, wherein different characteristic signals correspond to different sleep stages.

[0168] In some embodiments, the characteristic signal includes at least one of the following: gamma power, fast spindle power, oscillation signal power, slow wave signal.

[0169] In some embodiments, the sleep stage corresponding to the gamma power is the eyes-open stage of the awake period; the sleep stage corresponding to the fast spindle power is stage N2 and stage N3; the sleep stage corresponding to the oscillation signal power is stage N2, stage N3 and the eyes-closed stage of the awake period; the sleep stage corresponding to the slow wave signal is stage N3.

[0170] In some embodiments, when the characteristic signal includes the gamma power, the preset condition is that the probability density of the sleep frame is greater than the first multiple of the maximum probability density of all sleep frames, and the first multiple is less than 1 and greater than 0; when the characteristic signal includes the fast hammer wave power, the preset condition is that the probability density of the sleep frame under the first weight is greater than the probability density under the second weight, wherein the first weight is determined according to the cumulative distribution of the fast spindle wave power of the sleep frame whose fast spindle wave power is greater than the first threshold, and the second weight is determined according to the cumulative distribution of the fast spindle wave power of the sleep frame whose fast spindle wave power is less than the first threshold; when the characteristic signal includes the slow wave In the case of a sleep signal, the preset condition is that the probability density of the sleep frame is greater than the second multiple of the maximum probability density of all sleep frames, and the distance to the identified N2 and N3 stages is closer than other identified sleep stages, wherein the second multiple is less than 1 and greater than 0; in the case where the characteristic signal includes the oscillation signal power, the preset condition is that the probability density of the sleep frame under the third weight is greater than the probability density under the fourth weight, wherein the third weight is determined according to the cumulative distribution of the oscillation signal power of the sleep frames whose oscillation signal power is greater than the second threshold, and the fourth weight is determined according to the cumulative distribution of the oscillation signal power of the sleep frames whose oscillation signal power is less than the second threshold.

[0171] In some embodiments, when the characteristic signal includes gamma power, fast spindle power, oscillation signal power and slow wave signal, the device is further configured to: determine a sleep frame for which the sleep stage has not been determined as a sleep frame to be identified; determine the sleep stage corresponding to the sleep frame to be identified as the REM stage, and smooth the sleep frame to be identified; and according to the smoothing result, adjust the sleep stage corresponding to some sleep frames in the sleep frame to be identified whose distance from the N1 stage is less than a distance threshold to the N1 stage.

[0172] In some embodiments, adjusting the sleep stages corresponding to some sleep frames in the sleep frames to be identified whose distance from stage N1 is less than a distance threshold to stage N1 based on the smoothing processing result includes: adjusting the sleep stages corresponding to some sleep frames one by one, and re-performing smoothing processing after each adjustment.

[0173] In some embodiments, determining the power spectral density of the brain wave signal of each sleep frame during sleep includes: collecting the brain wave signal during sleep; filtering and resampling the brain wave signal, and dividing the processed brain wave signal into multiple sleep frames lasting a first duration; determining the power spectral density of the brain wave signal of each sleep frame.

[0174] The implementation process of the functions and effects of each unit in the above-mentioned device is specifically described in the implementation process of the corresponding steps in the above-mentioned method, and will not be repeated here.

[0175] An embodiment of the present disclosure further provides an electronic device, comprising: a processor and a memory; the memory is used to store a computer program; and the processor is used to execute the sleep staging method as described in any of the above embodiments by calling the computer program.

[0176] An embodiment of the present disclosure further provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the sleep staging method as described in any of the above embodiments.

[0177] An embodiment of the present disclosure further provides a computer program product, including a computer program, which implements the method described in any of the above embodiments when executed by a processor.

[0178] Figure 10 FIG1 is a schematic block diagram of a sleep staging apparatus 1000 according to an embodiment of the present disclosure. For example, the apparatus 1000 may be a mobile phone, a computer, a digital broadcast terminal, a messaging device, a game console, a tablet device, a medical device, a fitness device, a personal digital assistant, etc.

[0179] Reference Figure 10The device 1000 may include one or more of the following components: a processing component 1002 , a memory 1004 , a power component 1006 , a multimedia component 1008 , an audio component 1010 , an input / output (I / O) interface 1012 , a sensor component 1014 , and a communication component 1016 .

[0180] The processing component 1002 generally controls the overall operation of the device 1000, such as operations associated with display, phone calls, data communications, camera operation, and recording operations. The processing component 1002 may include one or more processors 1020 to execute instructions to perform all or part of the steps of the sleep staging method described above. In addition, the processing component 1002 may include one or more modules to facilitate interaction between the processing component 1002 and other components. For example, the processing component 1002 may include a multimedia module to facilitate interaction between the multimedia component 1008 and the processing component 1002.

[0181] The memory 1004 is configured to store various types of data to support the operations of the device 1000. Examples of such data include instructions for any application or method operating on the device 1000, contact data, phone book data, messages, pictures, videos, etc. The memory 1004 can be implemented by any type of volatile or non-volatile storage device, or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic memory, flash memory, magnetic disk, or optical disk.

[0182] The power supply component 1006 provides power to the various components of the device 1000. The power supply component 1006 may include a power management system, one or more power supplies, and other components associated with generating, managing, and distributing power to the device 1000.

[0183] The multimedia component 1008 includes a screen that provides an output interface between the device 1000 and the user. In some embodiments, the screen may include a liquid crystal display (LCD) and a touch panel (TP). If the screen includes a touch panel, it may be implemented as a touch screen to receive input signals from the user. The touch panel includes one or more touch sensors to sense touches, slides, and gestures on the touch panel. The touch sensors can not only sense the boundaries of a touch or slide action, but also detect the duration and pressure associated with the touch or slide action. In some embodiments, the multimedia component 1008 includes a front-facing camera and / or a rear-facing camera. When the device 1000 is in an operating mode, such as a capture mode or a video mode, the front-facing camera and / or the rear-facing camera can receive external multimedia data. Each front-facing camera and the rear-facing camera can have a fixed optical lens system or have focal length and optical zoom capabilities.

[0184] The audio component 1010 is configured to output and / or input audio signals. For example, the audio component 1010 includes a microphone (MIC) that is configured to receive external audio signals when the device 1000 is in an operating mode, such as a call mode, a recording mode, and a voice recognition mode. The received audio signals may be further stored in the memory 1004 or transmitted via the communication component 1016. In some embodiments, the audio component 1010 also includes a speaker for outputting audio signals.

[0185] I / O interface 1012 provides an interface between processing component 1002 and peripheral interface modules, such as a keyboard, click wheel, buttons, etc. These buttons may include, but are not limited to, a home button, volume buttons, a start button, and a lock button.

[0186] The sensor assembly 1014 includes one or more sensors for providing various aspects of the status assessment of the device 1000. For example, the sensor assembly 1014 can detect the open / closed state of the device 1000, the relative positioning of components, such as the display and keypad of the device 1000. The sensor assembly 1014 can also detect changes in the position of the device 1000 or a component of the device 1000, the presence or absence of user contact with the device 1000, the orientation or acceleration / deceleration of the device 1000, and changes in the temperature of the device 1000. The sensor assembly 1014 can include a proximity sensor configured to detect the presence of nearby objects without any physical contact. The sensor assembly 1014 can also include an optical sensor, such as a CMOS or CCD image sensor, for use in imaging applications. In some embodiments, the sensor assembly 1014 can also include an accelerometer, a gyroscope, a magnetic sensor, a pressure sensor, or a temperature sensor.

[0187] The communication component 1016 is configured to facilitate wired or wireless communication between the apparatus 1000 and other devices. The apparatus 1000 can access a wireless network based on a communication standard, such as WiFi, 2G, 3G, 4G LTE, 5G NR, or a combination thereof. In an exemplary embodiment, the communication component 1016 receives a broadcast signal or broadcast-related information from an external broadcast management system via a broadcast channel. In an exemplary embodiment, the communication component 1016 further includes a near field communication (NFC) module to facilitate short-range communication. For example, the NFC module can be implemented based on radio frequency identification (RFID) technology, infrared data association (IrDA) technology, ultra-wideband (UWB) technology, Bluetooth (BT) technology, and other technologies.

[0188] In an exemplary embodiment, the apparatus 1000 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the above-described sleep staging method.

[0189] In an exemplary embodiment, a non-transitory computer-readable storage medium including instructions is also provided, such as a memory 1004 including instructions. The instructions can be executed by the processor 1020 of the apparatus 1000 to implement the sleep staging method described above. For example, the non-transitory computer-readable storage medium can be a ROM, a random access memory (RAM), a CD-ROM, a magnetic tape, a floppy disk, an optical data storage device, and the like.

[0190] Other embodiments of the present disclosure will readily occur to those skilled in the art after considering the specification and practicing the disclosure herein. This disclosure is intended to cover any variations, uses, or adaptations of the present disclosure that follow the general principles of the present disclosure and include common knowledge or customary techniques in the art not disclosed herein. The description and examples are to be considered as exemplary only, with the true scope and spirit of the present disclosure being indicated by the following claims.

[0191] It should be understood that the present disclosure is not limited to the exact structures that have been described above and shown in the drawings, and that various modifications and changes can be made without departing from the scope thereof. The scope of the present disclosure is limited only by the appended claims.

[0192] It should be noted that, in this article, relational terms such as first and second, etc. 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. The terms "comprise", "include" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements that are not explicitly listed, or also includes elements that are inherent to such process, method, article or device. In the absence of further restrictions, an element defined by the sentence "comprises a..." does not exclude the presence of other identical elements in the process, method, article or device that includes the element.

[0193] The above is a detailed introduction to the methods and devices provided in the embodiments of the present disclosure. Specific examples are used herein to illustrate the principles and implementation methods of the present disclosure. The description of the above embodiments is only used to help understand the methods and core ideas of the present disclosure. At the same time, for those skilled in the art, according to the ideas of the present disclosure, there may be changes in the specific implementation methods and application scopes. In summary, the contents of this specification should not be understood as limiting the present disclosure.

Claims

1. A sleep staging method, characterized in that: The method comprises: determining the power spectral density of the EEG signal for each sleep frame during sleep; Performing dimensionality reduction processing on the power spectral density to obtain a dimensionality reduced data set; determining a plurality of characteristic signals of the brain wave signal of the sleep frame according to the power spectral density; For each sleep frame, performing kernel density estimation on the dimensionality reduction data set using a weight determined according to each characteristic signal to obtain a probability density corresponding to the characteristic signal; In a case where the probability density satisfies a preset condition corresponding to the characteristic signal, it is determined that the sleep frame belongs to the sleep stage corresponding to the characteristic signal, wherein different characteristic signals correspond to different sleep stages.

2. The method according to claim 1, characterized in that The characteristic signal includes at least one of the following: Gamma power, fast spindle power, oscillation signal power, slow wave signal.

3. The method according to claim 2, characterized in that The sleep stage corresponding to the gamma power is the eyes-open stage of the wakefulness period; The sleep stages corresponding to the fast spindle power are N2 and N3; The sleep stages corresponding to the oscillation signal power are stage N2, stage N3, and the eyes-closed stage of the awake period; The sleep stage corresponding to the slow wave signal is stage N3.

4. The method according to claim 2, characterized in that In a case where the characteristic signal includes the gamma power, the preset condition is that the probability density of the sleep frame is greater than a first multiple of the maximum probability density of all sleep frames, and the first multiple is less than 1 and greater than 0; When the characteristic signal includes the fast spindle power, the preset condition is that a probability density of the sleep frame under a first weight is greater than a probability density under a second weight, wherein the first weight is determined based on a cumulative distribution of fast spindle powers of sleep frames in which the fast spindle power is greater than a first threshold, and the second weight is determined based on a cumulative distribution of fast spindle powers of sleep frames in which the fast spindle power is less than the first threshold; When the characteristic signal includes the slow wave signal, the preset condition is that the probability density of the sleep frame is greater than a second multiple of the maximum probability density of all sleep frames, and the distance to the identified N2 stage and N3 stage is closer than that of other identified sleep stages, wherein the second multiple is less than 1 and greater than 0; In the case where the characteristic signal includes the oscillation signal power, the preset condition is that the probability density of the sleep frame under the third weight is greater than the probability density under the fourth weight, wherein the third weight is determined based on the cumulative distribution of the oscillation signal power of the sleep frames whose oscillation signal power is greater than the second threshold, and the fourth weight is determined based on the cumulative distribution of the oscillation signal power of the sleep frames whose oscillation signal power is less than the second threshold.

5. The method according to claim 2, characterized in that In the case where the characteristic signal includes gamma power, fast spindle power, oscillation signal power, and slow wave signal, the method further includes: determining a sleep frame for which the sleep stage is not determined as a sleep frame to be identified; Determining the sleep stage corresponding to the sleep frame to be identified as the REM stage, and performing smoothing processing on the sleep frame to be identified; According to the smoothing result, the sleep stages corresponding to some sleep frames in the sleep frames to be identified whose distance from stage N1 is less than a distance threshold are adjusted to stage N1.

6. The method according to claim 5, characterized in that The step of adjusting, according to the smoothing result, the sleep stages corresponding to some sleep frames of the sleep frames to be identified whose distance from the N1 stage is less than a distance threshold to the N1 stage, includes: The sleep stages corresponding to the partial sleep frames are adjusted one by one, and smoothing is performed again after each adjustment.

7. The method according to any one of claims 1 to 6, characterized in that Determining the power spectrum density of the brain wave signal of each sleep frame during sleep includes: Collect brain wave signals during sleep; performing filtering and resampling processing on the brain wave signal, and dividing the processed brain wave signal into a plurality of sleep frames lasting a first duration; Determine the power spectral density of the EEG signal for each sleep frame.

8. A sleep staging device, characterized in that: The device comprises: a power spectrum density determination unit configured to determine a power spectrum density of the brain wave signal of each sleep frame during sleep; a dimensionality reduction unit, configured to perform dimensionality reduction processing on the power spectral density to obtain a dimensionality reduced data set; a characteristic signal determining unit, configured to determine a plurality of characteristic signals of the brain wave signal of the sleep frame according to the power spectrum density; a kernel density estimation unit configured to perform kernel density estimation on the dimensionality reduction data set for each sleep frame using a weight determined according to each characteristic signal to obtain a probability density corresponding to the characteristic signal; The condition judgment unit is configured to determine that the sleep frame belongs to the sleep stage corresponding to the characteristic signal when the probability density meets the preset condition corresponding to the characteristic signal, wherein different characteristic signals correspond to different sleep stages.

9. An electronic device, characterized in that: include: Processor, memory; The memory is used to store computer programs; The processor is configured to execute the sleep staging method according to any one of claims 1 to 7 by calling the computer program.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the sleep staging method according to any one of claims 1 to 7 is implemented.

Images