Intelligent sleep instrument and sleep instrument control method

Abstract

The present application relates to the technical field of sleep intervention, in particular to an intelligent sleep instrument and a sleep instrument control method, the system comprising a respiratory mutation recognition module, a slow wave initiation linkage module, a pharyngeal cavity contraction evaluation module, an intervention section positioning module and a delay trigger control module. In the present application, the extreme points densely appearing in the continuous cycle are accurately positioned as the key nodes of respiratory abnormalities through the identification of the mutation of nasal respiratory flow, and the slow wave initiation stage closely related to the respiratory mutation is screened out and marked as the linkage period in combination with the power curve change trend of the electroencephalogram signal. The pharyngeal cavity state is dynamically evaluated through the change of acoustic echo in the period, the intelligent postponement and delay processing of the intervention behavior are realized through the comparison of the trigger point of the intervention action and the boundary of the platform section, the accuracy of the sleep intervention timeliness and the adaptability of the intervention execution are improved, and the respiratory event control effect and the sleep continuity are significantly improved.

Description

Technical Field

[0001] This invention relates to the field of sleep intervention technology, and in particular to an intelligent sleep device and a sleep device control method. Background Technology

[0002] The field of sleep intervention technology involves methods and devices for intervening in and regulating human sleep behavior and physiological states. Core aspects include sleep monitoring, physiological signal acquisition, state recognition, and intervention execution. It encompasses influencing sleep rhythms through methods such as airflow regulation, sound stimulation, light control, and temperature control. This is a comprehensive technology direction at the intersection of medical devices and physiological behavior regulation, widely used in assisting with sleep disorders such as insomnia, sleep apnea, and sleep disturbances. Among these, the intelligent sleep device refers to a device that uses an air pump to deliver air to achieve continuous positive airway pressure therapy to improve sleep disorders. It addresses the problem of sleep interruption caused by nighttime breathing difficulties. The intelligent sleep device uses a ventilation method that sets a constant air pressure or automatically adjusts the air pressure. It connects to the user's mouth and nose through a mask, continuously delivering air to the respiratory tract to ensure airway patency and achieve sleep adjustment.

[0003] Existing technologies employ constant or automatically regulated air pressure ventilation, primarily relying on mask ventilation to maintain airway patency. While this can alleviate sleep disruption caused by breathing difficulties to some extent, it suffers from recognition lag issues in cases of non-obstructive or intermittent respiratory abrupt changes. Especially when brain electrical activity and respiratory rhythms are not synchronized, a mismatch occurs between intervention and actual physiological fluctuations, resulting in intervention timing deviations. This makes it difficult to effectively cover high-risk periods caused by sudden pharyngeal contractions. For example, in some stages, breathing may be briefly abnormal but has not yet triggered air pressure changes, or the intervention frequency may not correspond to the rhythmic abnormality segment, leading to asynchrony between airway maintenance and neural regulation, resulting in decreased intervention efficiency and sleep recovery interruptions. Summary of the Invention

[0004] To address the technical problems of existing technologies, such as recognition lag in non-obstructive or intermittent respiratory abrupt changes, especially when brain electrical activity and respiratory rhythm are not synchronized, leading to a mismatch between intervention and actual physiological fluctuations, resulting in intervention timing deviations, and an inability to effectively cover high-risk periods caused by sudden pharyngeal contractions (e.g., in some stages where breathing is briefly abnormal but has not yet triggered air pressure changes), or where the intervention frequency fails to correspond to the rhythmic abnormality segment, resulting in asynchrony between airway maintenance and neural regulation, leading to decreased intervention efficiency and interrupted sleep recovery, this invention provides an intelligent sleep device and a sleep device control method. The technical solution is as follows:

[0005] On the one hand, a smart sleep device is provided, the system comprising:

[0006] The respiratory mutation identification module acquires raw data of nasal respiratory flow during the user's nighttime sleep, identifies waveform sequences within the respiratory cycle, identifies extreme points where waveform changes and directions reverse as candidate mutation points, and obtains a respiratory mutation reference node group.

[0007] The slow wave initiation linkage module uses the time tag in the respiratory mutation reference node group to extract the delta band power curve of the corresponding time period in the EEG signal, performs slope change screening on the rising segment of the curve, and generates a list of slow wave linkage time periods.

[0008] The pharyngeal contraction assessment module, based on the time range in the slow wave linkage time period list, emits a fixed sound pressure signal in the pharyngeal region, records the time interval sequence of multiple reverberation peaks and the peak value variation trend in the sound wave return, analyzes the shortening of the time interval and the amplitude change in the continuous cycle, and generates a pharyngeal contraction early warning label.

[0009] The intervention segment positioning module uses the time segment of the pharyngeal contraction warning tag to collect the range of respiratory rate changes, the range of oxygenation level changes, and the heart rate frequency band variation value within the same segment. It compares the stability intervals of the three data. If all three data maintain stable characteristics, the time segment is marked as the sleep intervention adaptation segment, and the sleep intervention adaptation segment sequence is obtained.

[0010] As a further aspect of the present invention, the respiratory mutation reference node group includes extreme value occurrence frequency characteristics, waveform transition statistical characteristics, and periodic stability indicators; the slow wave linkage time period list includes slow wave start position labels, band power trend indicators, and linkage delay estimates; the pharyngeal contraction warning label includes echo peak spacing contraction amplitude, echo peak fluctuation characteristics, and periodic decay rate; and the sleep intervention adaptation segment sequence includes oxygenation level stability indicators, respiratory rate maintenance range, and heart rate frequency band variation control amplitude.

[0011] As a further aspect of the present invention, the respiratory mutation recognition module includes:

[0012] The data segmentation submodule acquires raw data of nasal respiratory flow during the user's nighttime sleep, segments the raw data at fixed time intervals, detects the peak and trough values ​​in each segment, filters segments with amplitude changes exceeding the preset fluctuation range, removes segments with amplitude changes below the noise level, and generates respiratory fluctuation interval value groups.

[0013] The waveform extraction submodule, based on each segment of data in the respiratory fluctuation interval value group, calls the internal continuous numerical point series, locates the transition point between the rising and falling directions within the cycle, calculates the average slope difference before and after the continuous direction change, extracts the peak and trough points in the cycle structure, and generates the cycle fluctuation extreme point series.

[0014] The candidate point identification submodule calculates the periodic change based on the position difference between adjacent extreme points in the periodic fluctuation extreme point sequence, counts the morphological difference and positional movement amplitude of extreme points within the period, identifies point sequences that are reversed in direction and recurring in multiple periods, and generates a dataset of recurring mutation points.

[0015] The node tagging submodule calls all points in the repeating mutation point dataset, calculates the distribution density on the time series, groups the point density in the continuous time interval, assigns tag numbers to the concentrated distribution point sequences, and obtains the respiratory mutation reference node group.

[0016] As a further aspect of the present invention, the slow-wave initiation linkage module includes:

[0017] The power extraction submodule calls the time tag in the respiratory mutation reference node group to extract the data sequence of the corresponding time period in the EEG signal, separates the signal components of the delta band frequency range, calculates the instantaneous energy of the frequency band within the time moment and connects them into a time curve to generate the delta band power change sequence.

[0018] The rising segment filtering submodule arranges the values ​​in the δ band power change sequence, compares the numerical differences between adjacent positions point by point, identifies segments with continuous positive differences, calculates the average slope value inside the rising segment, retains segments with average slope exceeding the set change range, and generates a set of continuously changing segments.

[0019] The starting point positioning submodule calls multiple starting positions in the continuously changing segment set and corresponds them with time tags in the respiratory mutation reference node group. It calculates the offset error characteristic value, filters and marks the points whose offset error characteristic values ​​fall within the set time range, and obtains a list of slow wave linkage time periods.

[0020] The offset error characteristic value is expressed by the formula:

[0021] ;

[0022] in, Representing the The offset error characteristic value of a continuously changing segment. Representing the The time position of the starting point of a continuously changing segment Representative and the The time label of the respiratory mutation reference node corresponding to each fragment. Representing the Temporal weighting factor of class reference node, Represents weighting factors The average value, It is a constant. This represents the average offset distance.

[0023] As a further aspect of the present invention, the pharyngeal contraction assessment module includes:

[0024] The acoustic signal excitation submodule outputs a sound wave signal with a fixed frequency and sound pressure in the pharyngeal region based on the time range in the slow wave linkage time period list, sets the duration and transmission interval of each signal, records the data of the sound wave from transmission to reception, and generates a pharyngeal echo response sequence.

[0025] The reverberation parameter extraction submodule calls the time-domain signal in the pharyngeal reverberation response sequence, identifies the position of all reverberation peaks in each sound wave return, extracts the time interval value between adjacent reverberation peaks and the amplitude value of each peak, connects the same type of parameters within the period to identify the trend trajectory of change, and generates a reverberation feature trend sequence.

[0026] The periodic difference analysis submodule calculates the periodic difference feature value based on the arrangement sequence of time interval values ​​and amplitude values ​​in the residual characteristic trend sequence, marks the periodic periods with synchronous shortening and amplitude decreasing trend as warning periods, and obtains the pharyngeal contraction warning label.

[0027] As a further aspect of the present invention, the periodic difference characteristic value is expressed by the formula:

[0028] ;

[0029] in, Representing the The periodic difference characteristic value between each periodic segment and its adjacent periodic segments. Representing the The starting position of the time interval of each period segment. Representing the The starting position of the time interval of each period segment. Representing the The amplitude value of each period segment, Representing the The amplitude value of each period segment, The first in the representative period The amplitude value of each cycle, This represents the average value of the amplitude of the periodic segment. Represents the total number of periodic segments. It represents the absolute average difference between the amplitude value and the average value.

[0030] As a further aspect of the present invention, the intervention segment positioning module includes:

[0031] The physiological signal recognition submodule uses the time segment marked in the pharyngeal contraction warning tag to collect the instantaneous respiratory rate series, continuous blood oxygen saturation value and electrocardiogram signal data in each segment, and extracts the fluctuation range of respiratory rate, the variation range of oxygenation level and the variation value group of heart rate frequency band to generate a set of sleep state indicators.

[0032] The stability comparison submodule calculates the ratio of the peak difference to the trough difference of the three numerical groups in the sleep state index set according to the range of change of the three numerical groups in the segment, determines whether each ratio is lower than the stability judgment standard, filters the segment numbers that simultaneously meet the stability conditions, locates and summarizes the original time position in the pharyngeal contraction warning label, and obtains the sleep intervention adaptation segment sequence.

[0033] As a further aspect of the present invention, the sleep device also includes a delayed trigger control module:

[0034] The delayed trigger control module extracts the predetermined time of the pressure regulation action based on the start and end points of the segments in the sleep intervention adaptation segment sequence, and compares the position with the end point of the adjacent breathing plateau segment. If the predetermined action falls within the plateau segment, it is postponed to the end point, the delay interval is fixed, the execution time is reset, and the sleep adjustment postponement arrangement result is formed.

[0035] The sleep adjustment postponement arrangement results include the delayed execution time point, the plateau segment termination reference value, and the fixed delay adjustment interval.

[0036] As a further aspect of the present invention, the delay triggering control module includes:

[0037] The time extraction submodule extracts the preset pressure regulation action time points within the corresponding segment based on the start and end points of the segment in the sleep intervention adaptation segment sequence, arranges the action times of the segments in sequence, and generates an action time node group.

[0038] The position comparison submodule calls the time points in the action time node group, extracts the sequence of respiratory plateau segment termination points adjacent to the time points, judges the order of action time points and corresponding plateau segment termination points, marks all action points falling within the plateau segment range, and generates a list of actions that need to be extended.

[0039] The delay reset submodule moves the position backward to the platform segment termination point after a specified interval based on the time point in the list of actions to be postponed, replaces the original execution time with the new time point, rearranges the action time sequence according to the update order, and obtains the sleep adjustment postponement arrangement result.

[0040] On the other hand, the sleep device control method is based on the aforementioned smart sleep device and includes the following steps:

[0041] S1: Obtain raw data of nasal respiratory flow during the user's nighttime sleep, extract the extreme point sequence in the waveform signal within the respiratory cycle, compare whether the direction of adjacent extreme points in each cycle is reversed, count the frequency of recurrence in continuous cycles, and generate a respiratory mutation reference node group.

[0042] S2: Based on the time tags in the respiratory mutation reference node group, extract the δ band power curve of the corresponding time period in the EEG signal, screen the segments where the power value has a continuous increasing trend on the time axis, and generate a list of slow wave linkage time periods.

[0043] S3: Call the time segment in the slow wave linkage time period list, release the sound wave signal in the pharyngeal region and record the time series of multiple reverberation peaks in the reflected wave, analyze the time interval change trend and amplitude change pattern between adjacent reverberation peaks in the continuous period, and generate a pharyngeal contraction early warning label.

[0044] S4: Call the time segment indicated in the pharyngeal contraction early warning tag, extract the respiratory rate fluctuation range, blood oxygen change range and heart rate variation sequence within the segment, calculate the standard deviation of the three data and compare them, identify the trend of change, and generate a sleep intervention adaptation segment sequence.

[0045] S5: Call the start and end time information in the sleep intervention adaptation segment sequence, extract the preset pressure regulation action time point, compare the time point with the end time of the breathing plateau segment, if the action time point is within the plateau segment, it will be postponed to after the end of the plateau segment and a fixed delay time will be added to generate the sleep adjustment postponement arrangement result.

[0046] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following:

[0047] By identifying abrupt changes in nasal respiratory flow, the extreme points that occur frequently in continuous cycles are precisely located as key nodes of respiratory abnormalities. Simultaneously, by combining the power curve changes of EEG signals, the slow wave initiation phase closely related to respiratory abrupt changes is screened out and labeled as the linkage period. During this period, the pharyngeal state is dynamically assessed by changes in acoustic echo. During the assessment, physiological parameters such as oxygenation level, heart rate band, and respiratory rate are extracted simultaneously. Through multidimensional stability analysis, the optimal time for execution of intervention is precisely identified. By comparing the trigger point of the intervention action with the plateau segment boundary, intelligent follow-up and delay processing of the intervention behavior is achieved, forming an adjustment strategy that is more in line with the physiological rhythm. Through in-depth mining of signal interaction and dynamic allocation of key nodes, the accuracy of sleep intervention timeliness and the adaptability of intervention execution are improved, significantly improving the control effect of respiratory events and sleep continuity. Attached Figure Description

[0048] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0049] Figure 1 This is a schematic diagram of a smart sleep device provided in an embodiment of the present invention;

[0050] Figure 2 This is a schematic diagram of the sleep device frame of the present invention;

[0051] Figure 3 This is a flowchart of a sleep monitor control method provided in an embodiment of the present invention. Detailed Implementation

[0052] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0053] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0054] In the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that when the distinction is not emphasized, their intended meanings are consistent.

[0055] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0056] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0057] This invention provides a smart sleep aid, such as... Figure 1-2 The diagram shown illustrates a smart sleep aid system, which includes:

[0058] The respiratory mutation identification module acquires raw data of nasal respiratory flow during the user's nighttime sleep, identifies waveform sequences within the respiratory cycle, identifies extreme points where waveform changes and reverses direction as candidate mutation points, counts the number of times candidate points recur in consecutive cycles, determines whether they are densely distributed, and marks them if the conditions are met, thus obtaining a respiratory mutation reference node group.

[0059] The slow wave initiation linkage module uses the time tag in the respiratory mutation reference node group to extract the delta band power curve of the corresponding time period in the EEG signal, performs slope change screening on the rising segment of the curve, locates the starting point of the first continuous increase and judges the offset value with the mutation node. If it is within the predetermined time range, it is recorded and a slow wave linkage time period list is generated.

[0060] The pharyngeal contraction assessment module, based on the time range in the slow wave linkage time period list, emits a fixed sound pressure signal in the pharyngeal region, records the time interval sequence of multiple reverberation peaks and the peak value variation trend in the sound wave return, analyzes the shortening of time intervals and amplitude changes in continuous cycles, and generates pharyngeal contraction early warning labels.

[0061] The intervention segment localization module uses the time segment of the pharyngeal contraction warning tag to collect the range of respiratory rate changes, the range of oxygenation level changes, and the heart rate frequency band variation value within the same segment. The stability interval of the three data is compared. If all three data maintain stable characteristics, the time segment is marked as the sleep intervention adaptation segment, and the sleep intervention adaptation segment sequence is obtained.

[0062] The delayed trigger control module extracts the predetermined time of the pressure regulation action based on the start and end points of the segments in the sleep intervention adaptation segment sequence. It then compares the time point with the end point of the adjacent breathing plateau segment. If the predetermined action falls within the plateau segment, it is postponed to the end point, the delay interval is fixed, the execution time is reset, and the sleep adjustment postponement arrangement result is formed.

[0063] The respiratory mutation reference node group includes extreme value occurrence frequency characteristics, waveform transition statistical characteristics, and periodic stability indicators. The slow wave linkage period list includes slow wave start position labels, band power trend indicators, and linkage delay estimates. The pharyngeal contraction warning labels include echo peak spacing contraction amplitude, echo peak fluctuation characteristics, and periodic decay rate. The sleep intervention adaptation segment sequence includes oxygenation level stability indicators, respiratory rate maintenance range, and heart rate frequency band variability control amplitude. The sleep adjustment delay arrangement results include the execution action postponement time point, plateau segment termination reference value, and fixed delay adjustment interval.

[0064] Specifically, such as Figure 2 As shown, the respiratory mutation recognition module includes:

[0065] The data segmentation submodule acquires raw data of nasal respiratory flow during the user's nighttime sleep, segments the raw data at fixed time intervals, detects the peak and trough values ​​in each segment, filters segments with amplitude changes exceeding the preset fluctuation range, removes segments with amplitude changes below the noise level, and generates respiratory fluctuation interval value groups.

[0066] It is necessary to clearly define the criteria for dividing each time period, setting the length of each time period to 5 seconds. Using timestamps, the original data sequence is divided into several subsequences, each corresponding to a time period. Within each time period, signal processing methods (such as peak detection algorithms) are used to obtain the highest and lowest values, i.e., to identify the peak and trough values ​​of the data. The purpose is to capture the extreme values ​​of respiratory fluctuations within each time period, facilitating subsequent filtering and analysis of fluctuation amplitude. By setting a fluctuation amplitude threshold, fluctuations exceeding 10% are considered valid fluctuations, while smaller fluctuations are considered noise and filtered out. Data with fluctuation amplitudes below 10% will be discarded, resulting in a valid respiratory fluctuation interval value group. For example, if the maximum value of the data in a certain time period is 0.85 and the minimum value is 0.30, then the calculated fluctuation amplitude is (0.85-0.30) / 0.30=1.83, which is 183%, meeting the fluctuation threshold standard and being added to the valid data. If the fluctuation amplitude is too small, it is discarded, ensuring the accuracy and effectiveness of subsequent analysis and generating a respiratory fluctuation interval value group.

[0067] The waveform extraction submodule, based on each segment of data in the respiratory fluctuation interval value group, calls the internal continuous numerical point series, locates the transition point between the rising and falling directions within the cycle, calculates the average slope difference before and after the continuous direction change, extracts the peak and trough points in the cycle structure, and generates the cycle fluctuation extreme point series.

[0068] By detecting the turning points between rising and falling directions within a cycle, it is necessary to analyze the continuous numerical point series within each time period. By judging the difference between adjacent points, it can be determined whether the data segment is rising (positive difference) or falling (negative difference). Let's set a data sequence for a certain time period as [0.50, 0.60, 0.55, 0.45]. From 0.50 to 0.60 is the rising segment, and from 0.60 to 0.55 is the falling segment, indicating a further decline. Therefore, by detecting the turning points between rising and falling, the peaks and troughs of the data can be determined. To extract the peaks and troughs within a cycle more accurately, it is necessary to calculate the average slope difference for each cycle. The average slope within a given cycle is set as the ratio of the slope difference to the cycle length. Through this process, the peaks and troughs of each cycle are extracted. If the slope within a certain cycle is 5% / second, then the average slope difference for that cycle is 5%. Based on the location of the peaks and troughs, a series of extreme value points of cycle fluctuations is generated.

[0069] The candidate point identification submodule calculates the periodic change based on the positional difference between adjacent extreme points in the periodic fluctuation extreme point sequence, statistically analyzes the morphological difference and positional movement amplitude of extreme points within the period, identifies point sequences that repeat in multiple periods and reverse direction, and generates a dataset of repeated mutation points.

[0070] The calculation of periodic variation involves setting the period length to T and the time difference between two adjacent extreme points to Δt. The periodic variation can then be calculated as Δt / T. If Δt is less than a set threshold (0.5 seconds), the variation within the period is considered small and noise; if Δt is large, it indicates a significant variation within the period. Based on this, the morphological differences and positional shifts of extreme points within the period are statistically analyzed. Statistical information is used to identify point sequences that exhibit reversed direction and recurrence across multiple periods in certain data sequences. If the morphological difference of extreme points is greater than a set standard (greater than 0.3), it is considered a significant periodic change. Through comprehensive analysis of the information, a dataset of repeating mutation points is identified. These mutation points exhibit significant periodic changes and can serve as markers of abnormal respiratory fluctuations. This process relies not only on the statistical characteristics of the data points but also on the periodic features of the actual scenario to ensure the identification of meaningful abnormal respiratory fluctuations and the generation of a dataset of repeating mutation points.

[0071] The node labeling submodule calls all points in the repeating mutation point dataset, calculates the distribution density on the time series, groups the points in continuous time intervals according to their density, assigns label numbers to the concentrated distribution of point sequences, and obtains the respiratory mutation reference node group.

[0072] The distribution density of points in a time series can be calculated. Statistical methods (such as density estimation) can be used to classify the density of points. If the density of certain points on the time axis is high, it indicates that the points appear in a concentrated manner over time, reflecting a certain pattern or anomaly. Based on this, the density of points in continuous time intervals can be grouped, and a threshold can be set. If the density is greater than a certain standard (such as 2 points / second), it is considered a high-density area; otherwise, it is a low-density area. According to the analysis results of the distribution density, the concentrated distribution of point sequences is assigned a label number. If, within a certain time period, 5 repetitive mutation points appear in the same second, and only 1 to 2 points appear in the remaining time period, then the density of that time period is high. The points can be assigned a specific number to indicate that they are key nodes of respiratory mutations, forming a respiratory mutation reference node group.

[0073] Specifically, such as Figure 2 As shown, the slow-wave initiation linkage module includes:

[0074] The power extraction submodule calls the time stamp in the respiratory mutation reference node group to extract the data sequence of the corresponding time period in the EEG signal, separates the signal components of the delta band frequency range, calculates the instantaneous energy of the frequency band within the time moment and connects them into a time curve to generate the delta band power change sequence.

[0075] Extracting the corresponding time-segment data sequence from the EEG signal, setting the time label of a certain reference node as t1=5 seconds and t2=10 seconds, then extracting the EEG signal data sequence from 5 seconds to 10 seconds from the EEG signal. For the data, frequency domain analysis is used to separate the signal components in the delta band frequency range, defined as 0.5-4Hz. The signal in this frequency band is extracted using a bandpass filter. Specific filtering methods can include setting an FIR filter or an IIR filter, etc. The instantaneous energy of the frequency band at each moment is calculated. The instantaneous energy is calculated by integrating the square of the signal. The instantaneous energy E(t) at a certain moment can be expressed by the formula:

[0076] ;

[0077] in, The extracted delta band signal, To calculate the window size;

[0078] By calculating the instantaneous energy over the entire period, a time curve is obtained, representing the power change of the delta band. The instantaneous energy values ​​are connected into a time series. This process not only converts the time-domain data of the EEG signal into frequency-domain data, but also reveals the fluctuation of EEG activity in a specific period by calculating the power change of the signal, providing basic data for subsequent rising segment screening and starting point localization, and generating a delta band power change sequence.

[0079] The rising segment filtering submodule arranges the values ​​in the δ band power change sequence, compares the numerical differences between adjacent positions point by point, identifies segments with continuous positive differences, calculates the average slope value within the rising segment, retains segments with an average slope exceeding the set change range, and generates a set of continuously changing segments.

[0080] The values ​​in the delta-band power change sequence need to be arranged, and the differences between adjacent values ​​need to be compared point by point. If the delta-band power change sequence for a certain period is [1.5, 2.0, 2.4, 2.8, 2.5], then the differences between two adjacent points are 0.5, 0.4, 0.4, and -0.3, respectively. By calculating the differences between adjacent points, continuous positive difference segments can be identified. In the example above, 0.5, 0.4, and 0.4 belong to a continuous positive difference segment, which can be considered a rising segment. After identifying the continuous rising segment, the average slope value within the rising segment needs to be calculated. The average slope can be obtained by calculating the ratio of the power change between the start and end points of the rising segment to the time difference. The formula is:

[0081] ;

[0082] If a certain upward segment starts from =0 seconds to =5 seconds, power from =1.0 increased to =2.5, then its average slope is (2.5-1.0) / (5-0)=0.3. The segments with average slope exceeding the set change range are retained. The set change range threshold is 0.2. The segments with slope greater than 0.2 will be retained to generate a set of continuously changing segments.

[0083] The starting point positioning submodule calls multiple starting positions in the continuously changing segment set, matches them with time labels in the respiratory mutation reference node group, calculates the offset error characteristic value, filters and marks the points whose offset error characteristic value falls within the set time range, and obtains a list of slow wave linkage time periods.

[0084] The offset error characteristic value is calculated using the following formula:

[0085] ;

[0086] in, Representing the The offset error characteristic value of a continuously changing segment. Representing the The time position of the starting point of a continuously changing segment Representative and the The time label of the respiratory mutation reference node corresponding to each fragment. Representing the Temporal weighting factor of class reference node, Represents weighting factors The average value, It is a constant. This represents the average offset distance.

[0087] The formula calculation logic is as follows: Extract the starting point of change in the EEG signal and the time position of the reference node of the mutation in the respiratory signal, calculate the time difference between the two, and introduce the weight factor set of the reference node for weighted correction so that the difference is comprehensively considered in terms of the strength of the temporal contribution. The standard deviation of the weighted term is introduced into the denominator for normalization. Combined with the constant smoothing term, the interference of the denominator being zero is eliminated. The average offset distance of the current sample reference group is subtracted as the center correction term to obtain the individual offset difference between each segment and the reference node. This logic uses the time difference, weighting factors, and variance structure to form a composite distance measure, which measures the degree of abnormality of each segment and the reference node in temporal linkage under the standard scale, and is used to screen the effective linkage interval.

[0088] The offset error characteristic value reflects the degree of correction deviation in timing alignment between a continuously changing segment and its corresponding reference node. The larger the value, the less the segment conforms to the average linkage characteristics and the less it has linkage synchronization.

[0089] Meaning of parameters and derivation of formulas:

[0090] parameter For the first The starting point of each continuously changing segment was determined by detecting time-domain sampling points of the EEG signal. The starting point was obtained by extracting the inflection point of change in the dense slow-wave region. The monitoring frequency was set to 250Hz. The starting point of the third continuously changing segment in a certain data set was marked at the 6250th sampling point. ;

[0091] parameter As the time label for the corresponding respiratory mutation reference node, mutation points in the respiratory signal are detected using a chest-strap respiratory sensor to locate the mutation time. The third mutation point of the reference node falls at the 5975th sampling point, corresponding to: ;

[0092] parameter For the first The temporal weighting factor for reference nodes is calculated as the product of the node's clustering density in the slow-wave abrupt change region and the node's persistence score. The clustering density is obtained by normalizing the number of reference nodes per unit time, and the persistence score is scored by normalizing the node's change in magnitude over time.

[0093] Table 1: Parameter Table

[0094] ;

[0095] Calculate the average weight: ;

[0096] Calculate the weighted variance and square root terms: ;

[0097] ;

[0098] Smoothing parameters Derived from the lower bound of the standard deviation of the sample data, to avoid the denominator being zero, it is set to 0.005, which is taken from the 5th percentile of the standard deviation of the sample data of the reference node set in the slow wave section;

[0099] Average offset distance To select the mean of the differences between the starting point of all continuously changing segments within the sample and their corresponding reference nodes, 10 sample segment pairs were tested sequentially:

[0100] ;

[0101] The observed values ​​were: 1.1, 0.9, 1.2, 1.3, 1.0, 1.1, 0.8, 1.2, 1.0, 0.9;

[0102] ;

[0103] Substitute all values ​​into the formula to calculate:

[0104] ;

[0105] The results indicate that the characteristic value of the offset error between the current third continuously changing segment and the reference node is 18.544, which means that the alignment deviation between the starting point and the reference respiratory node is far greater than the average offset interval. It is inferred that it does not belong to the slow wave linkage period set and needs to be removed in the subsequent screening process.

[0106] Specifically, such as Figure 2 As shown, the pharyngeal contraction assessment module includes:

[0107] The acoustic signal excitation submodule outputs a sound wave signal with a fixed frequency and sound pressure in the pharyngeal region based on the time range in the slow wave linkage time period list. It sets the duration and transmission interval of each signal, records the data of the sound wave from transmission to reception, and generates a pharyngeal echo response sequence.

[0108] Based on the time range in the slow-wave linkage time period list, a sound wave signal with a fixed frequency and sound pressure level is output to the pharyngeal region. If a certain time period in the list is set to [10 seconds, 15 seconds], the system will select a fixed frequency (set to 500Hz) and a fixed sound pressure level (e.g., 80 dBSPL) to output a sound wave signal within this time period. The duration and transmission interval of the signal need to be set according to actual needs. The duration of each signal is set to 0.5 seconds, and the transmission interval is set to 2 seconds. That is, a 0.5-second sound wave signal is emitted every 2 seconds. During the transmission process, a suitable sound wave transmitting device is used to emit a sound wave signal in the pharyngeal region. In each time period after the sound wave is emitted, the system will record all data from transmission to reception. The specific data includes the time-domain waveform of the sound wave signal and the time delay of the received signal. By analyzing the signal reception delay and waveform, the reverberation characteristics of the sound wave can be analyzed to generate a pharyngeal reverberation response sequence.

[0109] The reverberation parameter extraction submodule calls the time-domain signal in the pharyngeal reverberation response sequence, identifies the position of all reverberation peaks in each sound wave return, extracts the time interval between adjacent reverberation peaks and the amplitude value of each peak, connects similar parameters within the period to identify the trend trajectory of change, and generates a reverberation characteristic trend sequence.

[0110] The system retrieves the time-domain signal from the pharyngeal echo response sequence and extracts the complete waveform data of the sound wave return signal. The time-domain signal contains multiple reverberation peaks, representing the echoes after the sound wave signal is reflected in the pharynx. The system identifies the position of the reverberation peak in each sound wave return using a peak detection algorithm. Peak detection can be performed by calculating the second-order difference (i.e., acceleration) of the signal. If the second-order difference is positive, it indicates that the signal is rising at that point; if the second-order difference is negative, it indicates that the signal is falling. This allows for precise location of each reverberation peak. The system extracts the time interval between adjacent reverberation peaks and the amplitude of each peak. For example, if the first reverberation peak appears at t1=10 seconds and the second reverberation peak appears at t2=11.5 seconds, then the time interval between them is 1.5 seconds. At the same time, the amplitude of each peak is extracted through peak detection; the amplitude of the first peak is 5 units, and the amplitude of the second peak is 4 units. The system arranges and analyzes the time interval and amplitude values ​​of adjacent reverberation peaks, connects the trend trajectories of similar parameters within the period, and generates a reverberation characteristic trend sequence.

[0111] The periodic difference analysis submodule calculates the periodic difference characteristic value based on the arrangement sequence of time interval values ​​and amplitude values ​​in the reverberation characteristic trend sequence, marks the periods with synchronous shortening and amplitude decreasing trends as warning periods, and obtains pharyngeal contraction warning labels;

[0112] The periodic difference characteristic value is calculated using the following formula:

[0113] ;

[0114] in, Representing the The periodic difference characteristic value between each periodic segment and its adjacent periodic segments. Representing the The starting position of the time interval of each period segment. Representing the The starting position of the time interval of each period segment. Representing the The amplitude value of each period segment, Representing the The amplitude value of each period segment, The first in the representative period The amplitude value of each cycle, This represents the average value of the amplitude of the periodic segment. Represents the total number of periodic segments. It represents the absolute mean difference between the amplitude value and the average value.

[0115] The formula's calculation logic is as follows: It is calculated by combining time interval and amplitude change to measure the synchronous change trend of adjacent cycles in the acoustic features of the pharynx. In the formula, the numerator calculates the time interval difference between two adjacent cycle segments, multiplies it by the square root of the absolute difference in amplitude change between the two segments, and comprehensively reflects the coupling relationship between cycle shortening and energy decrease. The denominator introduces the deviation of the cycle amplitude value from the average value as a normalization correction term to suppress the interference of local extreme values ​​on the overall trend identification, making the feature values ​​comparable under different signal intensity levels. The formula includes addition, absolute value, multiplication, square root, summation and fractional operations to ensure that the weight ratio of time change and amplitude change is orderly and unified, and introduces statistics to control the influence of noise. It is used to identify whether there is a coupling feature of amplitude decrease and synchronous time interval shortening in continuous signals. It is an intermediate numerical basic judgment indicator for obtaining pharyngeal contraction warning signals.

[0116] The period difference characteristic value reflects the degree of coupling between adjacent period segments in terms of time interval and amplitude changes. The larger the characteristic value, the more likely the period at that position is to simultaneously experience shortening and energy reduction.

[0117] Meaning of parameters and derivation of formulas:

[0118] parameter , The starting position of the time interval collected in the reverberation characteristic trend sequence of continuous periodic segments is obtained from the high-precision time annotation of the throat reverberation signal by real-time acoustic monitoring equipment. The feature point marking is completed by a 24-bit ADC module at a sampling rate of 16kHz, with a resolution of up to 0.0625 milliseconds and the time starting point error controlled within 0.25ms.

[0119] In the collected samples, the starting time interval of the third period segment is: ;

[0120] The starting time interval of the fourth period is: ;

[0121] therefore: ;

[0122] parameter , The peak amplitude of adjacent period segments, in dB, is obtained by logarithmic quantization of the maximum envelope value extracted after background noise is filtered out by a bandpass filter. In the monitoring sample, the amplitudes of the 3rd and 4th period segments are obtained as follows: ;

[0123] Calculate the amplitude change term: ;

[0124] The square root term is: ;

[0125] parameter The arithmetic mean of the amplitude values ​​of the periodic segments is calculated from the amplitude values ​​of the first M=5 periodic segments of the entire sequence. The collected values ​​are:

[0126] ;

[0127] Calculate the average:

[0128] ;

[0129] The absolute mean difference term is:

[0130] ;

[0131] Substitute all parameters into the calculation formula:

[0132] ;

[0133] The results show that the periodic difference characteristic value between periodic segment 3 and periodic segment 4 is 3.504. The higher the value, the more obvious the amplitude decrease and periodic contraction at this position. This value is one of the comparison items for the synchronous judgment trend threshold. When multiple consecutive periodic difference characteristic values ​​exceed the empirical threshold of 2.5, combined with the synchronous trend, it is marked as a warning cycle, and a pharyngeal contraction warning label is obtained.

[0134] Specifically, such as Figure 2 As shown, the intervention section positioning module includes:

[0135] The physiological signal recognition submodule uses the time segments marked in the pharyngeal contraction warning tag to collect the instantaneous respiratory rate series, continuous blood oxygen saturation value and electrocardiogram signal data in each segment, extracts the fluctuation range of respiratory rate, the variation range of oxygenation level and the variation value group of heart rate frequency band, and generates a set of sleep state indicators.

[0136] Using the time range marked on the pharyngeal constriction warning tag, the system collects instantaneous respiratory rate data, continuous blood oxygen saturation values, and electrocardiogram (ECG) signal data within that time period. The system extracts changes in respiratory rate during this period, which can be calculated by detecting changes in chest cavity or airflow during respiration. The respiratory rate value for a given time period is set to [16, 17, 16, 15, 16] breaths / minute, and the system records this numerical sequence. The system also collects blood oxygen saturation data, setting the blood oxygen saturation value within that time period to [98%, 97%, 9...]. [8%, 98.5%, 98%] This data reflects changes in oxygenation levels. The system also needs to collect electrocardiogram (ECG) signal data. A common method is to use an ECG sensor to extract heart rate frequency band data. The heart rate is set to [75, 76, 74, 78, 75] beats per minute. The data will be organized into heart rate frequency band values. By extracting the fluctuation range of physiological signals, the fluctuation range of respiratory rate, the variation range of oxygenation level, and the variation value group of heart rate frequency band, the changes of various physiological parameters of the body during sleep can be effectively reflected, generating a set of sleep state indicators.

[0137] The stability comparison submodule calculates the ratio of the peak difference to the trough difference of the three numerical groups in the sleep state index set according to the range of change of the three numerical groups in the segment, determines whether each ratio is lower than the stability judgment standard, filters the segment numbers that meet the stability conditions at the same time, locates and summarizes the original time position in the pharyngeal contraction warning label, and obtains the sleep intervention matching segment sequence.

[0138] Based on the variation range of the three numerical groups in the sleep state index set, it is necessary to calculate the ratio of the peak difference to the trough difference for each numerical group within that range. Let the fluctuation range of respiratory rate be [16, 17, 16, 15, 16], then its peak value is 17, its trough value is 15, and the difference is 17-15=2; the fluctuation range of blood oxygen saturation is [98%, 97%, 98%, 98.5%, 98%], then its peak value is 98.5%, its trough value is 97%, and the difference is 98.5%-97%=1.5%; the fluctuation range of heart rate frequency band is [75, 76, 74, 78, 75], then its peak value is 78, its trough value is 74, and the difference is 78-74=4. Calculate the ratio of the peak difference to the trough difference for each numerical group. For the respiratory rate range, the ratio is 2 / 16 = 0.125. For the blood oxygen saturation range, the ratio is calculated to be 1.5 / 98 = 0.0153. For the heart rate frequency range, the ratio is calculated to be 4 / 76 = 0.0526. Through calculation, it can be determined whether the ratio is lower than the preset stability judgment standard. The standard is set to 0.1. If a certain ratio is less than 0.1, then that item is considered stable; otherwise, it is unstable. During the screening process, the segment numbers that simultaneously meet the stability conditions are filtered out. The stability of some segments meets the standard, while the remaining segments do not. The system will locate and summarize the original time position in the pharyngeal contraction warning label to provide effective time points and target ranges for subsequent sleep management and intervention, ensuring that intervention measures can be carried out at appropriate times to help improve sleep quality and obtain the sleep intervention appropriate segment sequence.

[0139] Specifically, such as Figure 2 As shown, the delay trigger control module includes:

[0140] The time extraction submodule extracts the preset pressure regulation action time points within the corresponding segment based on the start and end points of the segment in the sleep intervention adaptation segment sequence, arranges the action times of the segments in sequence, and generates an action time node group.

[0141] Based on the start and end points of the sleep intervention adaptation segment sequence, it is necessary to extract the preset pressure regulation action time points within the segment. The start and end points of a certain adaptation segment are set as [10 seconds, 15 seconds]. Within this segment, the preset pressure regulation action time points are [11 seconds, 13 seconds, 14 seconds]. These time points represent the various adjustment actions that need to be performed within the segment. The system will arrange the action times within the segment in sequence, setting them to ascending order to obtain [11 seconds, 13 seconds, 14 seconds], forming an adjustment action schedule for a specific segment. This provides data support for subsequent action comparison and delay processing, generating action time node groups.

[0142] The position comparison submodule calls the time points in the action time node group, extracts the sequence of respiratory plateau segment termination points adjacent to the time points, judges the order of action time points and corresponding plateau segment termination points, marks all action points falling within the plateau segment range, and generates a list of actions that need to be extended.

[0143] The system calls each time point in the action time node group and extracts the sequence of respiratory plateau segment termination points adjacent to the time point. The respiratory plateau segment termination point represents the end point of a period in which the respiratory rate or oxygenation level remains stable. If a time point is set to 13 seconds, the sequence of respiratory plateau segment termination points is [12 seconds, 16 seconds, 18 seconds]. The system needs to determine the order of each action time point and its corresponding respiratory plateau segment termination point. If an action time point is set to 13 seconds, and its corresponding respiratory plateau segment termination point is 12 seconds, then obviously, if the 13-second action time point falls after the 12-second termination point, the system will mark all action points within the plateau segment range for that action time point. If an action time point of 14 seconds falls after the 16-second termination point of a plateau segment, then no extension is needed. If the 13-second action time point falls before the termination point of a plateau segment, then that time point will be considered an action point that needs to be extended, and a list of actions that need to be extended will be generated.

[0144] The Delay Reset submodule moves the position of the action to the end of the platform segment after the specified interval based on the time point in the action list that needs to be postponed, and replaces the original execution time with the new time point. It then rearranges the action time sequence according to the update order and obtains the sleep adjustment postponement arrangement result.

[0145] Based on the time points in the list of actions to be postponed, the system moves their positions backward to a specified interval after the end point of the plateau segment. If the time point to be postponed is set to 13 seconds, the end point of the plateau segment is 16 seconds, and the specified interval is 1 second, then the action time point of 13 seconds will be adjusted to 16 seconds + 1 second = 17 seconds. The system replaces the original execution time with the new time point and rearranges the action time sequence according to the updated order. If the updated time points are [11 seconds, 13 seconds, 17 seconds], the sequence will be sorted according to the new adjustment time, providing a new time point arrangement for subsequent sleep intervention adjustments. This ensures that the adjustment actions are performed in the correct order, guaranteeing that the intervention measures do not affect the stability of the breathing plateau segment, and obtaining the sleep adjustment postponement arrangement result.

[0146] Please see Figure 3 The sleep aid control method is based on the aforementioned smart sleep aid and includes the following steps:

[0147] S1: Obtain raw data of nasal respiratory flow during the user's nighttime sleep, extract the extreme point sequence in the waveform signal within the respiratory cycle, compare whether the direction of adjacent extreme points in each cycle is reversed, count the frequency of recurrence in continuous cycles, and generate a respiratory mutation reference node group.

[0148] S2: Based on the time tags in the respiratory mutation reference node group, extract the delta band power curve of the corresponding time period in the EEG signal, screen the segments where the power value has a continuous increasing trend on the time axis, and generate a list of slow wave linkage time periods.

[0149] S3: Call the time segment in the slow wave linkage time period list, release the sound wave signal in the pharyngeal region and record the time series of multiple reverberation peaks in the reflected wave, analyze the time interval change trend and amplitude change pattern between adjacent reverberation peaks in the continuous period, and generate pharyngeal contraction early warning label.

[0150] S4: Call the time segment indicated in the pharyngeal contraction warning label, extract the respiratory rate fluctuation range, blood oxygen change range and heart rate variation sequence within the segment, calculate the standard deviation of the three data and compare them, identify the trend of change, and generate a sleep intervention adaptation segment sequence.

[0151] S5: Call the start and end time information in the sleep intervention adaptation segment sequence, extract the preset pressure regulation action time point, compare the time point with the end time of the breathing plateau segment, if the action time point is within the plateau segment, it will be postponed to after the end of the plateau segment and a fixed delay time will be added to generate the sleep adjustment postponement arrangement result.

[0152] The above are merely specific embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A smart sleep aid, characterized in that, The intelligent sleep device includes: The respiratory mutation identification module acquires raw data of nasal respiratory flow during the user's nighttime sleep, identifies waveform sequences within the respiratory cycle, identifies extreme points where waveform changes and directions reverse as candidate mutation points, and obtains a respiratory mutation reference node group. The slow wave initiation linkage module uses the time tag in the respiratory mutation reference node group to extract the delta band power curve of the corresponding time period in the EEG signal, performs slope change screening on the rising segment of the curve, and generates a list of slow wave linkage time periods. The pharyngeal contraction assessment module, based on the time range in the slow wave linkage time period list, emits a fixed sound pressure signal in the pharyngeal region, records the time interval sequence of multiple reverberation peaks and the peak value variation trend in the sound wave return, analyzes the shortening of the time interval and the amplitude change in the continuous cycle, and generates a pharyngeal contraction early warning label. The intervention segment positioning module uses the time segment of the pharyngeal contraction early warning tag to collect the range of respiratory rate changes, the range of oxygenation level changes, and the heart rate frequency band variation value within the same segment. The stability interval of the three data is compared. If all of them maintain stable characteristics, the time segment is marked as the sleep intervention adaptation segment, and the sleep intervention adaptation segment sequence is obtained. The delayed trigger control module extracts the predetermined time of the pressure regulation action based on the start and end points of the segments in the sleep intervention adaptation segment sequence, and compares the position with the end point of the adjacent breathing plateau segment. If the predetermined action falls within the plateau segment, it is postponed to the end point, the delay interval is fixed, the execution time is reset, and the sleep adjustment postponement arrangement result is formed. The sleep adjustment postponement arrangement results include the execution action postponement time point, the plateau segment termination reference value, and the fixed delay adjustment interval; The delay trigger control module includes: The time extraction submodule extracts the preset pressure regulation action time points within the corresponding segment based on the start and end points of the segment in the sleep intervention adaptation segment sequence, arranges the action times of the segments in sequence, and generates an action time node group. The position comparison submodule calls the time points in the action time node group, extracts the sequence of respiratory plateau segment termination points adjacent to the time points, judges the order of action time points and corresponding plateau segment termination points, marks all action points falling within the plateau segment range, and generates a list of actions that need to be extended. The delay reset submodule moves the position backward to the platform segment termination point after a specified interval based on the time point in the list of actions to be postponed, replaces the original execution time with the new time point, rearranges the action time sequence according to the update order, and obtains the sleep adjustment postponement arrangement result.

2. The intelligent sleep device according to claim 1, characterized in that, The respiratory mutation reference node group includes extreme value occurrence frequency characteristics, waveform transition statistical characteristics, and periodic stability indicators. The slow wave linkage time period list includes slow wave start position labels, band power trend indicators, and linkage delay estimates. The pharyngeal contraction warning labels include echo peak spacing contraction amplitude, echo peak fluctuation characteristics, and periodic decay rate. The sleep intervention adaptation segment sequence includes oxygenation level stability indicators, respiratory rate maintenance range, and heart rate frequency band variation control amplitude.

3. The intelligent sleep device according to claim 1, characterized in that, The respiratory mutation identification module includes: The data segmentation submodule acquires raw data of nasal respiratory flow during the user's nighttime sleep, segments the raw data at fixed time intervals, detects the peak and trough values ​​in each segment, filters segments with amplitude changes exceeding the preset fluctuation range, removes segments with amplitude changes below the noise level, and generates respiratory fluctuation interval value groups. The waveform extraction submodule, based on each segment of data in the respiratory fluctuation interval value group, calls the internal continuous numerical point series, locates the transition point between the rising and falling directions within the cycle, calculates the average slope difference before and after the continuous direction change, extracts the peak and trough points in the cycle structure, and generates the cycle fluctuation extreme point series. The candidate point identification submodule calculates the periodic change based on the positional difference between adjacent extreme points in the periodic fluctuation extreme point sequence, statistically analyzes the morphological difference and positional movement amplitude of extreme points within the period, identifies point sequences that repeat in multiple periods and reverse direction, and generates a dataset of repeated mutation points. The node tagging submodule calls all points in the repeating mutation point dataset, calculates the distribution density on the time series, groups the points in continuous time intervals according to their density, assigns tag numbers to the concentrated distribution of point sequences, and obtains the respiratory mutation reference node group.

4. The intelligent sleep device according to claim 3, characterized in that, The slow-wave initiation linkage module includes: The power extraction submodule calls the time tag in the respiratory mutation reference node group to extract the data sequence of the corresponding time period in the EEG signal, separates the signal components of the delta band frequency range, calculates the instantaneous energy of the frequency band within the time moment and connects them into a time curve to generate the delta band power change sequence. The rising segment filtering submodule arranges the values ​​in the δ band power change sequence, compares the numerical differences between adjacent positions point by point, identifies segments with continuous positive differences, calculates the average slope value inside the rising segment, retains segments with average slope exceeding the set change range, and generates a set of continuously changing segments. The starting point positioning submodule calls multiple starting positions in the continuously changing segment set and corresponds them with time tags in the respiratory mutation reference node group. It calculates the offset error characteristic value, filters and marks the points whose offset error characteristic values ​​fall within the set time range, and obtains a list of slow wave linkage time periods. The offset error characteristic value is expressed by the formula: ; in, Representing the The offset error characteristic value of a continuously changing segment. Representing the The time position of the starting point of a continuously changing segment Representative and the The time label of the respiratory mutation reference node corresponding to each fragment. Representing the Temporal weighting factors of class reference nodes, Represents weighting factors The average value, It is a constant. This represents the average offset distance.

5. The intelligent sleep device according to claim 4, characterized in that, The pharyngeal contraction assessment module includes: The acoustic signal excitation submodule outputs a sound wave signal with a fixed frequency and sound pressure in the pharyngeal region based on the time range in the slow wave linkage time period list, sets the duration and transmission interval of each signal, records the data of the sound wave from transmission to reception, and generates a pharyngeal echo response sequence. The reverberation parameter extraction submodule calls the time-domain signal in the pharyngeal reverberation response sequence, identifies the position of all reverberation peaks in each sound wave return, extracts the time interval value between adjacent reverberation peaks and the amplitude value of each peak, connects the same type of parameters within the period to identify the trend trajectory of change, and generates a reverberation feature trend sequence. The periodic difference analysis submodule calculates the periodic difference feature value based on the arrangement sequence of time interval values ​​and amplitude values ​​in the residual characteristic trend sequence, marks the periodic periods with synchronous shortening and amplitude decreasing trend as warning periods, and obtains the pharyngeal contraction warning label.

6. The intelligent sleep device according to claim 5, characterized in that, The periodic difference characteristic value is expressed by the formula: ; in, Representing the The periodic difference characteristic value between each periodic segment and its adjacent periodic segments. Representing the The starting position of the time interval of each period segment. Representing the The starting position of the time interval of each period segment. Representing the The amplitude value of each period segment, Representing the The amplitude value of each period segment, The first in the representative period The amplitude value of each cycle, This represents the average value of the amplitude of the periodic segment. Represents the total number of periodic segments. It represents the absolute average difference between the amplitude value and the average value.

7. The intelligent sleep device according to claim 5, characterized in that, The intervention segment positioning module includes: The physiological signal recognition submodule uses the time segment marked in the pharyngeal contraction warning tag to collect the instantaneous respiratory rate sequence, continuous blood oxygen saturation value and electrocardiogram signal data in each segment, extracts the fluctuation range of respiratory rate, the variation range of oxygenation level and the variation value group of heart rate frequency band, and generates a set of sleep state indicators. The stability comparison submodule calculates the ratio of the peak difference to the trough difference of the three numerical groups in the sleep state index set according to the range of change of the three numerical groups in the segment, determines whether each ratio is lower than the stability judgment standard, filters the segment numbers that simultaneously meet the stability conditions, locates and summarizes the original time position in the pharyngeal contraction warning label, and obtains the sleep intervention matching segment sequence.

Images