Sleep time point detection method and system based on non-contact sensor

Abstract

The invention relates to a sleep time point detection method and system based on a non-contact sensor. Comprising the following steps: acquiring discrimination characteristics based on historical data and detection data collected by a non-contact sensor; sorting each sub-feature value extracted from the detection data based on a time sequence to form a candidate time sequence corresponding to each sub-feature; calculating a feature weight corresponding to each sub-feature based on discriminant features obtained by the historical data and the detection data; combining the candidate time sequences corresponding to the sub-features to obtain a total candidate time sequence, inputting the total candidate time sequence into a score model to obtain a score-candidate time sequence table, substituting the feature weights corresponding to the sub-features into the score-candidate time sequence table, and calculating to obtain a target prediction sequence table; and selecting the final target candidate time point with the maximum total score value in the target prediction sequence table as the final target sleep time point. According to the method, the difference of different user sleep feature data is considered, and a scientific sleep time point detection result is provided.

Description

Technical Field

[0001] The present invention relates to the technical field of sleep detection, and particularly to a method and system for detecting sleep time points based on non-contact sensors. Background Art

[0002] Sleep quality directly affects human health and quality of life. With the rise of the concept of health preservation, more and more people seek to monitor and understand their own sleep quality. To meet the sleep monitoring needs of users, in the prior art, technicians have developed various sleep detection devices, including wearable devices (such as smart watches and smart bracelets with sleep detection functions) and devices closely combined with bedding (such as smart mattresses).

[0003] The sleep time point is a key indicator for evaluating sleep quality. The sleep time point includes the sleep onset time point and the awakening time point. As the two time endpoints of a sleep process, they need to be identified during the sleep quality monitoring process.

[0004] Based on this, in the prior art, devices such as sleep beds equipped with non-contact sensors are used to detect the sleep time points of users. The sleep data collected by the non-contact sensors can be converted into a single physiological parameter or behavioral parameter. In the prior art, based on the above single parameter, mathematical models are used to find mathematical feature points such as extreme points or inflection points in the source data to determine the sleep time points. This method has low accuracy and does not consider individual differences, so its practical application value is low. Summary of the Invention

[0005] Aiming at the problems existing in the prior art, the present invention provides a method and system for detecting sleep time points based on non-contact sensors. This method uses multi-dimensional data, comprehensively considers the differences in different user sleep characteristic data, and provides scientific sleep time point detection results.

[0006] On the one hand, a method for detecting sleep time points based on non-contact sensors, the method includes: Obtaining discriminant features based on historical data and detection data collected by non-contact sensors; the discriminant features include multiple sub-features, each sub-feature includes multiple sub-feature values, and each sub-feature value corresponds to a time point; the time interval length between any two adjacent time points is the same; Sorting the sub-feature values extracted from the detection data based on the time sequence to form candidate time series corresponding to each sub-feature; Calculating the feature weights corresponding to each sub-feature based on the discriminant features obtained from the historical data and the detection data respectively; Merge the candidate time series corresponding to each sub-feature to obtain a total candidate time series, input the total candidate time series into the score model, and obtain a score-candidate time series table; the score-candidate time series table includes multiple candidate time points and multiple score values ​​corresponding to each candidate time point, and the multiple score values ​​include the score values ​​corresponding to each of the multiple sub-features; The feature weights corresponding to each sub-feature are brought into the score-candidate time series table, and a target prediction sequence table is calculated, the target prediction sequence table includes multiple final target candidate time points and multiple total score values; each final target candidate time point corresponds to a total score value; the total score value is determined based on the score values ​​corresponding to each sub-feature in the score-candidate time series table and the feature weights corresponding to the sub-features; The final target candidate time point with the largest total score value in the target prediction sequence table is selected as the final target sleep time point.

[0007] As a preferred embodiment of the present invention, the feature weight corresponding to the sub-feature is calculated as follows: Obtain sub-feature quality scores and sub-feature fluctuation values ​​based on historical data and current day data; The weight comprehensive index of each sub-feature is calculated based on the sub-feature quality score and the sub-feature fluctuation value; the feature weight of the target sub-feature is calculated based on the weight comprehensive index of the sub-feature.

[0008] As a preferred embodiment of the present invention, one of the multiple sub-features is a body motion feature, and the quality score of the body motion feature is calculated as follows: (1) Among them, Q tw Indicates the quality score of the body movement index, sum_tw mean represents the average of the total number of body movements in the past z days, sum_tw represents the total number of detected body movements, sum_tw i It represents the total body movements on the ith day in the past z days, ε is the correction coefficient, and tw_history represents the maximum difference between the total body movements on a single day and the average of the total body movements within z days in the past z days.

[0009] As a preferred embodiment of the present invention, one of the multiple sub-features is a breathing feature, and the quality score of the breathing feature is calculated as follows: (2) Among them, Q brwhere \(Q_{br}\) is the mass fraction of the breathing index, \(cont\ of\ br\ in[x1,x2]\) represents the number of breathing sub - feature values whose numerical magnitudes fall within the range \([x1,x2]\) in the detected valid sleep data; \(N\) is the total number of breathing sub - feature values in the detected valid sleep data; \([x1,x2]\) is set based on the standard breathing interval of human sleep.

[0010] As a preference of the present invention, one of the multiple sub - features is the heartbeat feature, and the calculation method of the mass fraction of the heartbeat feature is specifically as follows: (3) where \(Q\) he is the mass fraction of the heartbeat index, \(cont\ of\ br\ in[x3,x4]\) represents the number of heartbeat sub - feature values whose numerical magnitudes fall within the range \([x3,x4]\) in the detected valid sleep data; \(N\) is the total number of heartbeat sub - feature values in the detected valid sleep data; \([x3,x4]\) is set based on the standard heartbeat interval of human sleep.

[0011] As a preference of the present invention, before the time point with the highest output score, there is also a threshold verification step: Extract historical sleep time points from the historical data; Add threshold periods in the forward time direction and the backward time direction of the historical sleep time points to form a threshold verification interval; Judge whether the time point with the largest total score value in the target prediction sequence table falls into the threshold verification interval; if it successfully falls in, output this time point as the final target sleep time point; if it fails to fall in, do not output, and continue to judge whether the time point with the second - largest total score value falls into the threshold verification interval until the final target sleep time point is successfully output.

[0012] As a preference of the present invention, the discriminant feature includes body movement sub - features, and the body movement sub - features include multiple body movement sub - feature values. When forming the candidate time series corresponding to the body movement sub - features, the adjacent window iteration method is used for time point selection.

[0013] On the other hand, the present invention provides a sleep time point detection system based on a non - contact sensor, and the system includes: A feature extraction module, configured to obtain discriminant features based on historical data and detection data collected by the non - contact sensor; the discriminant features include multiple sub - features, each sub - feature includes multiple sub - feature values, and each sub - feature value corresponds to a time point; the time interval length between any two adjacent time points is the same; A candidate time series generation module, configured to sort the respective sub - feature values extracted from the detection data based on time sequence to form candidate time series corresponding to the respective sub - features; A weight calculation module, configured to calculate the feature weights corresponding to each sub-feature based on the discriminant features obtained from the historical data and the detection data respectively; A sequence list generation module, configured to merge the candidate time series corresponding to each sub-feature to obtain a total candidate time series, input the total candidate time series into a score model, and obtain a score-candidate time series list; the score-candidate time series list includes a plurality of candidate time points and a plurality of score values corresponding to each candidate time point, and the plurality of score values include the score values corresponding to each of the plurality of sub-features; A calculation module, configured to substitute the feature weights corresponding to each sub-feature into the score-candidate time series list, and calculate and obtain a target prediction sequence list, where the target prediction sequence list includes a plurality of final target candidate time points and a plurality of total score values; each final target candidate time point corresponds to a total score value; the total score value is determined based on the score values corresponding to each sub-feature in the score-candidate time series list and the feature weights corresponding to the sub-features; An output module, configured to select the final target candidate time point with the largest total score value in the target prediction sequence list as the final target sleep time point for output.

[0014] On the other hand, an electronic device includes a processor and a memory; The processor is connected to the memory; The memory is configured to store executable program code; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the above-mentioned sleep time point detection method.

[0015] In summary, the technical solution provided by the present invention has the following beneficial effects: 1. The sleep time point detection method of the present invention obtains the data of the user during the sleep process through a non-contact sensor, and extracts a plurality of different discriminant features from it, such as body movement features, breathing features or heartbeat features, and considers the contribution degree of the features through the calculation of the weights of different features. That is, although there are individual differences in the sleep processes of different users, the individual differences can be quantitatively understood through the feature weight calculation step in the present invention, and the adaptability and recognition accuracy of different user groups can be improved.

[0016] 2. The present sleep time point detection method is different from traditional mathematical models. Instead of simply finding the extreme points of data changes or the extreme points of curve slopes, it forms a candidate time series by extracting the candidate time points corresponding to each sub-feature, then obtains the total candidate time series and scores the subsequent score model to find the point with the greatest possibility (the largest score value) among multiple possible time points as the target sleep time point. Its score evaluation mechanism enables the finally obtained target sleep time point to well adapt to the sleep process states of different users. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0018] Figure 1 is a flowchart of the present sleep time point detection method; Figure 2 is a structural block diagram of the present sleep time point detection system; Figure 3 is a schematic diagram of feature separation processing in the embodiment. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0019] Now, the subject matter described herein will be discussed with reference to example embodiments. It should be understood that discussing these embodiments is only to enable those skilled in the art to better understand and thus implement the subject matter described herein, and is not a limitation on the scope of protection, applicability, or examples set forth in the claims. The functions and arrangements of the elements discussed can be changed without departing from the scope of protection of the content of this specification. Each example can omit, substitute, or add various processes or components as needed. For example, the described method can be executed in an order different from the described order, and each step can be added, omitted, or combined. Additionally, the features described relative to some examples can also be combined in other examples.

[0020] As used herein, the term "comprising" and its variants represent open terms, meaning "including but not limited to". The term "based on" means "at least partially based on". The term "one embodiment" and "an embodiment" mean "at least one embodiment". The term "another embodiment" means "at least one other embodiment". The terms "first", "second", etc. can refer to different or the same objects. Other definitions, whether explicit or implicit, may be included below. Unless clearly specified in the context, the definition of a term is consistent throughout the specification.

[0021] The technical solution provided by the present invention targets sleep time points, including the falling asleep time point and the waking up time point, which respectively correspond to the falling asleep time and the waking up time of the user in the total time period from going to bed to getting out of bed.

[0022] In the present invention, the non-contact sensor is mainly used to collect vibration characteristic signals, which can be implemented using a piezoelectric sensor, a resistive sensor, or an acoustic wave sensor. Characteristics such as body movement (turning over, moving limbs, or shifting), snoring, breathing, and heart beating of the human body will all generate vibrations, and there are differences in the vibration characteristics corresponding to different behavioral characteristics. Taking the piezoelectric sensor as an example, when it is installed in an auxiliary medium (such as the mattress of a sleep bed), it is used to collect all vibration information and convert the vibration information into a piezoelectric signal (in the form of voltage) based on the piezoelectric principle. As Figure 3 shown, since in the piezoelectric signal, different voltage characteristics are distributed in different frequency bands, existing algorithms can, based on physical parameters such as frequency and amplitude, and in combination with means such as confidence evaluation, separate different voltage characteristics from the obtained piezoelectric signal, and perform operations such as data conversion and data cleaning on these voltage characteristics to obtain different discriminant characteristics. The discriminant characteristics are composed of various types of sub-characteristics, such as body movement characteristics, breathing characteristics, heartbeat characteristics, and sleep talking characteristics, etc. For different users, different discriminant characteristic weights can be used to determine the sleep time points. In addition, the non-contact sensor can process all the original piezoelectric signals collected based on the existing homologous judgment technology to solve the problem of data mixing caused by one auxiliary medium coming into contact with multiple users.

[0023] As Figure 1 shown, an embodiment of the present invention provides a method for detecting sleep time points based on a non-contact sensor. The method includes: Obtaining historical data and detection data collected by the non-contact sensor, and preprocessing the detection data to extract discriminant characteristics; the discriminant characteristics include multiple sub-characteristics, each sub-characteristic includes multiple sub-characteristic values, and each sub-characteristic value corresponds to a time point; the time interval length between any two adjacent time points is the same, and this time interval length is set as the calculation interval T.

[0024] In this embodiment, taking the piezoelectric film sensor as a possible implementation manner of the non-contact sensor, and arranging the piezoelectric film sensor in the mattress of the intelligent bed to further explain this method.

[0025] First, the piezoelectric film sensor arranged in the intelligent bed mattress is in a 24-hour standby state to obtain all vibration information of the user on the intelligent bed mattress. The piezoelectric film sensor can send data to the analysis device in the prior art. The analysis device processes the overall voltage information by executing the existing voltage feature separation algorithm to obtain multiple discriminant features. The discriminant features are the features used by the present invention for sleep time point detection, and they include multiple different sub-features, such as body movement sub-features, breathing sub-features, and heartbeat sub-features, etc. Each sub-feature is composed of multiple sub-feature values. The sub-feature value is the numerical value of the feature behavior. Each sub-feature value corresponds to a time point, and the time interval between any two time points is the calculation interval T, that is, the sub-feature value is the numerical value of the occurrence of the feature behavior in the calculation interval T.

[0026] For example, let the calculation interval T be 10s. The processor sorts out and records the numerical value of the occurrence of the feature behavior collected by the piezoelectric film sensor in the previous 10s time period every 10s; taking the body movement sub-feature as an example, if the first body movement sub-feature value is 0, the corresponding time point is 22:00:00, the second body movement sub-feature value is 1, the corresponding time point is 22:00:10, and the third body movement sub-feature value is 3, the corresponding time point is 22:00:20. It can be seen that within the 10s from 22:00:00 to 22:00:10, the user performed 1 body movement behavior, and within the 10s from 22:00:10 to 22:00:20, the user performed 3 body movement behaviors. Similarly, it can be known that every time the calculation interval T passes, the processor makes a sorting record, and the corresponding discriminant features obtained include 3 sub-feature values corresponding to each of the body movement sub-feature, breathing sub-feature, and heartbeat sub-feature.

[0027] The historical data includes all discriminant feature data in the effective sleep data of each day in the past z days. The effective sleep data refers to the user's time in bed. For example, if the bedtime on the 3rd is 22:00:00 and the out-of-bed time on the 4th is 06:00:00, the actual time length corresponding to the effective sleep data is 28800s. If the calculation interval T is set to 10s, in the effective sleep data on the 4th, each sub-feature includes 2880 sub-feature values; the detection data includes all discriminant feature data in the latest set of effective sleep data.

[0028] In addition, the method further includes sorting the sub-feature values extracted from the detection data based on the time sequence to form a candidate time series corresponding to each sub-feature; Since each sub-eigenvalue corresponds to a time point, sorting the sub-eigenvalues based on the time order can form candidate time series corresponding to each sub-feature. Taking the body movement sub-feature, the breathing sub-feature, and the heart rate sub-feature as examples, three candidate time series can be formed: the body movement candidate time series st1, the breathing candidate time series st2, and the heart rate candidate time series st3.

[0029] Next, calculate the feature weights corresponding to each sub-feature based on historical data and detection data.

[0030] Let the body movement weight be w1, the breathing weight be w2, and the heart rate weight be w3. Then, the weight calculation can be performed according to Equations (4), (5), and (6): (4) (5) (6) where: w tw is the comprehensive index of the body movement data weight, w br is the comprehensive index of the breathing data weight, w he is the comprehensive index of the breathing data weight.

[0031] (7) (8) (9) where, Q tw represents the quality score of the body movement index, Q br is the quality score of the breathing index, Q he is the quality score of the heart rate index, V tw is the fluctuation value of the body movement index, V br is the fluctuation value of the breathing index, V he is the fluctuation value of the heart rate index; α and β are weighting coefficients, and α + β = 1. The weighting coefficients reflect the trade-off between the quality score and the fluctuation situation; the value range of α is [0, 1], indicating the proportion of the importance of the corresponding quality score when calculating the weight index; the value range of β is [0, 1], indicating the proportion of the importance of the corresponding index fluctuation when calculating the weight index; α and β can be preset and assigned manually with reference to historical experience or historical data.

[0032] In this embodiment, the calculation methods of the quality score Q tw of the body movement index, the quality score Q br of the breathing index, and the quality score Q he of the heart rate index are shown in Equations (1), (10), and (11) respectively: (1) (10) As shown in formula (1), sum_tw mean represents the mean value of the total body movements in the past z days; specifically, by summing all the body movement sub-feature values in the effective sleep data for each day in the past z days of historical data (time corresponding: from the detected bed-in state to the detected bed-out state), and then dividing by the total number of days z, the total number of body movements per day on average can be obtained; sum_tw represents the total number of detected body movements; specifically, the total number of detected body movements represents the sum of all the body movement sub-feature values in the complete effective sleep data on the latest day, indicating how many times the user has had body movement behaviors during the process from getting into bed to getting out of bed; sum_tw i represents the total number of body movements on the i-th day in the past z days; ε is a correction coefficient, which is actually a very small positive number used to avoid a zero denominator; tw _history represents the maximum value of the difference between the total number of body movements on a single day and the mean value of the total number of body movements in the past z days; |sum_tw - sum_tw mean | represents the absolute numerical difference between the total number of detected body movements and the mean value of the total number of body movements; Among them, the total number of detected body movements refers to the total number of body movements in the most recent detection (all the data obtained during the process of the user getting into bed to getting out of bed). In formula (1), both the numerator and the denominator represent the difference in the number of body movements based on the mean value of the total number of body movements, and both are positive values (taking the absolute value). If |sum_tw - sum_tw mean | is numerically less than tw _history, then the value of Q tw is not 0, which means that in the most recent detection corresponding to the total number of detected body movements, the body movement index conforms to the historical performance; if |sum_tw - sum_tw mean | is numerically greater than tw _history, then the value of Q tw is 0, which means that in the most recent detection corresponding to the total number of detected body movements, the body movement index has deviated from the "most deviated" body movement index in history, and the body movement characteristics are no longer referenceable, representing that in subsequent calculations, the participation degree of the weight of the body movement characteristics is adjusted to 0; based on this, it reflects the credibility of the detected body movement characteristics in historical data and reflects the importance of the detected body movement characteristics.

[0033] (11) As shown in formula (11), Q br is the respiratory index quality score; T is the calculation interval; N is the total number of times the processor sorts and records the occurrence quantity values of the features collected by the piezoelectric film sensor in a set of valid sleep data (which is also the number of sub-feature values corresponding to each sub-feature), and its value is equal to the number of calculation intervals T. For example, when T = 10s, if the time length of the valid sleep data on a certain day is 8 hours (28800s), then it can be known that N = 2880; cont of br in[x1,x2] represents the number of respiratory sub-feature values whose magnitudes fall within the range of [x1,x2] in the detected valid sleep data; [x1,x2] is set based on the standard breathing interval of human sleep. For the users with bound information, their attribute portraits can be clearly known. Based on this, in combination with the authoritative medical research results or the historical data sets in relevant medical databases, the [x1,x2] interval can be selected specifically for this user (including but not limited to the official website of the National Health Commission, the National Center for Cardiovascular Diseases, the National Standard Full-text Public System, the World Health Organization WHO, etc.).

[0034] For example, for an elderly user with bound information over 60 years old, based on the above method, the breathing rate interval per minute is manually set to [12,18]. If the calculation interval T = 10s, then it can be known that cont of br in[2,3].

[0035] As can be seen from Equation (11), by dividing the number of respiratory sub-feature values whose magnitudes fall within the range of [2,3] by N, the proportion of the time when the elderly user is within the normal breathing frequency range during sleep to the total time can be obtained. The higher this proportion, the br higher the quality score Q of the breathing index, indicating that the contribution degree of the respiratory sub-feature is greater.

[0036] (12) As can be seen from Equation (12), Q he is the quality score of the heart rate index; T is the calculation interval; In the same calculation method as Equation (11), the larger Q he is, the greater the contribution degree of the heart rate sub-feature.

[0037] (i = 1,2,...,N) (13) (i = 1,2,...,N) (14) (i = 1,2,...,N) (15) Among them, based on the variance formula, in Equations (13), (14) and (15), V tw is the body movement index fluctuation value, V br is the breathing index fluctuation value, Vhe is the fluctuation value of the heartbeat index; As can be seen from the previous steps, the three sub - eigenvalue values are sorted based on the time sequence, forming a body movement candidate time series st1, a breathing candidate time series st2, and a heartbeat candidate time series st3. Each series contains N sub - eigenvalue values; It should be noted that based on the normalization operation in the prior art, all sub - eigenvalue values are normalized so that the numerical size of each sub - eigenvalue value is within the range of [0, 1], making V tw 、V br and V he unified in data form for facilitating the calculation of the weighted comprehensive index by substituting into equations (4), (5), and (6); where, tw_i norm represents the value of the i - th body movement sub - eigenvalue in the body movement candidate time series after normalization; tw norm_mean represents the mean value of all body movement sub - eigenvalue values in the body movement candidate time series after normalization; br_i norm represents the value of the i - th breathing sub - eigenvalue in the breathing candidate time series after normalization; br_i norm_mean represents the mean value of all breathing sub - eigenvalue values in the breathing candidate time series after normalization; he_i norm represents the value of the i - th heartbeat sub - eigenvalue in the heartbeat candidate time series after normalization; he_i norm_mean represents the mean value of all heartbeat sub - eigenvalue values in the heartbeat candidate time series after normalization; Based on the above equations (7), (8), and (9), the weighted comprehensive index is calculated, and the relative weights of the quality score and the fluctuation situation are adaptively adjusted through the weighting coefficients α and β; when there is less historical data, the β coefficient can be increased and the α coefficient can be decreased to reduce the weight influence of the quality score; when there is more historical data, the β coefficient can be decreased and the α coefficient can be increased to increase the weight influence of the quality score.

[0038] After obtaining the weighted comprehensive index w tw of the body movement data, the weighted comprehensive index w br of the breathing data, and the weighted comprehensive index w he of the breathing data, substitute them into the above equations (4), (5), and (6) to calculate and obtain the body movement weight w1, the breathing weight w2, and the heartbeat weight w3.

[0039] Then, the candidate time series corresponding to each sub - eigenvalue are merged to obtain the total candidate time series, and the total candidate time series is input into the score model to obtain the score - candidate time series table.

[0040] The above-mentioned body motion candidate time series st1, breathing candidate time series st2 and heartbeat candidate time series st3 are merged to form a total candidate time series ST (the repeated time points are not merged and deleted).

[0041] Furthermore, the total candidate time series ST is input into the score model to obtain a score-candidate time series table; in this embodiment, examples of three score models are provided: a body movement score model, a breathing score model, and a heartbeat score model. The total candidate time series ST will be input into the above three score models respectively, so that each time point corresponds to three score results.

[0042] Specifically, the body movement score model (falling asleep) is first shown: (16) (1) Set the score value range to [0,1]; (2) As can be seen from the above, each body motion sub-feature value corresponds to a time point, namely, a candidate time point. The body motion sub-feature value corresponding to each candidate time point is extracted in turn, and the score of the candidate time point is calculated based on the above model; Among them, tw seep_score represents the sleeping score of the candidate time point for falling asleep with body movement, before_sum represents the sum of the number of body movements in a period of time before the candidate time point for falling asleep with body movement, and after_sum represents the sum of the number of body movements in a period of time after the candidate time point for falling asleep with body movement; the larger the difference between before_sum and after_sum, the higher the sleeping score; the "period of time" here can be set to m calculation intervals T, and m>0.

[0043] The following shows the breathing and heart rate score model (falling asleep): (17) (1) Set the score value range to [0,1]; (2) bh sleep_score represents the sleep score at the candidate time point of breathing / heartbeat falling asleep, and e represents a natural constant; =before_mean-after_mean; =before_mean-mean; before_mean represents the average number of breaths / heartbeats in a period of time before the candidate time point of falling asleep by breathing / heartbeat; after_mean represents the average number of breaths or heartbeats in a period of time after the candidate time point of falling asleep by breathing / heartbeat; mean represents the average value during the entire sleep process of breathing / heartbeat (i.e., corresponding to the effective sleep data), and numerically it is equal to the average value of all sub-feature values of breathing / heartbeat in the effective sleep data. It can be seen from the formula that the greater the difference between before_mean and after_mean, the higher the falling asleep score.

[0044] k1 and k2 are slope adjustable parameters used to control the slope of the function and can be adjusted according to actual needs; for example, if it is desired to increase or the impact on the score, k1 or k2 can be increased; α1 and α2 are offset adjustable parameters used to control the offset of the function and adjust the baseline of the difference; for example, if it is desired or to reach a certain value to significantly affect the score, α1 or α2 can be adjusted.

[0045] Since both k and α are adjustable parameters, their specific values can be adjusted to make it applicable to the breathing / heartbeat model function and different users.

[0046] Secondly, the calculation method of the falling asleep score is described in the above embodiments. Similarly, the calculation method of the waking up score under the method of the present invention is introduced below: Show the body movement score model (waking up): (18) (1) Set the score value range as [0, 1]; (2) Based on the previous steps, each body movement sub-feature value corresponds to a time point, that is, the candidate time point. The body movement sub-feature value corresponding to each candidate time point is extracted in sequence, and the score of the waking up candidate time point is calculated based on the waking up model shown in formula (18); where tw wake_score represents the waking up score of the body movement waking up candidate time point, before_sum represents the sum of the number of body movements in a period of time before the body movement waking up candidate time point, after_sum represents the sum of the number of body movements in a period of time after the body movement waking up candidate time point; the greater the difference between after_sum and before_sum, the higher the waking up score; the "period of time" here can be set as m calculation intervals T, and m > 0.

[0047] The following shows the breathing and heartbeat score models (waking up): (19) (1)Set the score value range to [0, 1]; (2)bh wake_score represents the arousal score at the candidate time point of respiration / heartbeat arousal, and e represents the natural constant; =before_mean - after_mean; =after_mean - mean; before_mean represents the average number of breaths / heartbeats in a period of time before the candidate time point of respiration / heartbeat arousal; after_mean represents the average number of breaths or heartbeats in a period of time after the candidate arousal time point of respiration / heartbeat; mean represents the average value during the entire sleep process of respiration / heartbeat (i.e., corresponding to the effective sleep data), and numerically it is equal to the average value of all sub - feature values of respiration / heartbeat in the effective sleep data. From the formula, it can be seen that the greater the difference between before_mean and after_mean, the higher the arousal score.

[0048] k3 and k4 are slope - adjustable parameters used to control the slope of the function and can be adjusted according to actual needs; for example, if it is desired to increase or the impact on the score, k3 or k4 can be increased; α3 and α4 are offset - adjustable parameters used to control the offset of the function and adjust the baseline of the difference; for example, if it is desired or to reach a certain value to significantly affect the score, α3 or α4 can be adjusted.

[0049] Since both k and α are adjustable parameters, their specific values can be adjusted to make it applicable to the respiration / heartbeat model function and different users.

[0050] Based on the score calculation of the total candidate time series ST using the above - mentioned score model, each time point corresponds to three score results. Then, the above - mentioned body movement weight w1, respiration weight w2, and heartbeat weight w3 are substituted into the score - candidate time series table for weighted summation calculation, so that each time point finally corresponds to a total score value, and finally the target prediction sequence table is obtained. Based on the above calculation operations, it can be seen that the total score value comprehensively considers the contribution degrees of different sub - features, and the contribution degrees of sub - features in turn reflect the biases of different user groups in specific types of feature data, making this method applicable to all user groups.

[0051] Finally, from the target prediction sequence list, select the point with the largest total score value as the final target sleep time point. If the sleep onset score model is used above, this target sleep time point corresponds to the target sleep onset time point; if the wake-up score model is used above, this target sleep time point corresponds to the target wake-up time point.

[0052] In another possible implementation, different from the above embodiment, when selecting the point with the largest total score value from the target prediction sequence list as the final target sleep time point, a threshold verification step needs to be performed: extract the historical sleep time points from the historical data; add threshold time periods in the forward time direction and backward time direction of the historical sleep time points respectively to form a threshold verification interval; determine whether the time point with the largest total score value in the target prediction sequence list falls into the threshold verification interval; if it successfully falls in, output this time point as the final target sleep time point; if it fails to fall in, do not output, and continue to determine and analyze whether the time point with the second largest total score value falls into the threshold verification interval until the final target sleep time point is successfully output.

[0053] Specifically, taking the acquisition of the target sleep onset time point as an example, first, it is necessary to extract the historical sleep onset time points from the historical data, and the historical sleep onset time points can be determined based on the sleep onset time points that have been determined in the previous few days. Taking 21:00:00 as an example, add threshold time periods in the forward time direction and backward time direction of 21:00:00 respectively. For example, if the threshold time period is set to 15 minutes, the finally formed threshold verification interval is from 20:45:00 to 21:15:00. At this time, when outputting the time point with the largest total score value in the target prediction sequence list, it is necessary to first determine whether this time point falls into the above threshold verification interval.

[0054] For example, if the time point with the highest score in the target prediction sequence list is 21:10:40, then it can be directly determined that this time point is the final target sleep time point; if the time point with the highest score in the target prediction sequence list is 20:35:20, then this time point is not output; then select the time point with the second largest total score value in the target prediction sequence list. For example, if the time point with the second largest total score value is 21:06:20, then this time point is used as the final target sleep time point, and so on. If the time point with the second largest total score value still does not fall into the threshold verification interval, continue to find the time point with the third largest total score value until the final target sleep time point is determined.

[0055] Thus, between several adjacent time points with relatively large total score values, there may be a situation where the time point interval is abnormal, which is usually caused by individual factors. At this time, through the threshold verification step provided in this embodiment, several possible time points with similar scores can be distinguished with the help of historical data, ensuring that the final target sleep time point reaches the highest credibility.

[0056] In another possible embodiment, for the body movement feature, when forming the candidate time series corresponding to the body movement feature, the adjacent window iteration method is also used to select time points.

[0057] By using the adjacent window iteration method, the time points corresponding to each body movement feature value in the body movement candidate time series can be made more scientific. The specific operation is as follows: First, arrange all body movement feature values in chronological order. Then, set a time window, and the time length of each time window is an integer multiple of the calculation interval T, which makes each time window contain multiple time points. Then, set a sliding step, which is also an integer multiple of the calculation interval T but needs to be less than the time length of the time window. Each time window corresponds to a window body movement feature value, and its value is obtained by summing the body movement feature values corresponding to multiple time points under this window. Then, set a window body movement threshold TWP and a body movement proportion threshold TWRP. Extract a time window, and then compare the window body movement feature value with the set window body movement threshold TWP. If the former is greater than the latter, calculate the ratio of the window body movement feature value of the current time window to the window body movement feature value of the next adjacent window (the window obtained by sliding the current time window once according to the sliding step) (referred to as the "body movement proportion between two adjacent windows"), and compare this ratio with the body movement proportion threshold TWRP. If the former is greater than the latter, select the time point in the middle of the current window. By sliding the time window, all time points in the original total time series are selected, and based on these time points, a candidate time series corresponding to the body movement feature is formed. Compared with the sequence before the selection step is not executed, this candidate time series removes some time points with low credibility (the adjacent window iteration method can utilize the local continuity feature of the time series data and the statistical correlation between adjacent windows, and eliminate time points with low credibility by dynamically adjusting the confidence of the data within the window), reducing the data volume of the candidate time series.

[0058] In another possible embodiment, for the respiration or heartbeat sub-feature, when forming the candidate time series corresponding to the respiration or heartbeat sub-feature, a time series change point detection algorithm in the prior art (such as the Dynp algorithm) can be used to detect change points for the time points corresponding to all respiration sub-feature values and heartbeat sub-feature values, and finally, a respiration or heartbeat candidate time series is formed based on all the successfully detected time points. In addition, it can also combine the screening steps in the above embodiments for double screening to further reduce the data volume and improve the scientific nature of the data.

[0059] On the other hand, this embodiment also provides a sleep time point detection system based on a non-contact sensor. The system includes: A feature extraction module, configured to obtain discriminant features based on historical data and detection data collected by a non-contact sensor; the discriminant features include multiple sub-features, each sub-feature includes multiple sub-feature values, and each sub-feature value corresponds to a time point; A candidate time series generation module, configured to sort each sub-feature value based on time sequence to form a candidate time series corresponding to each sub-feature; A weight calculation module, configured to calculate the feature weights corresponding to each sub-feature based on historical data and detection data; A sequence list generation module, configured to merge the candidate time series corresponding to each sub-feature to obtain a total candidate time series, input the total candidate time series into a score model to obtain a score-candidate time series list; the score-candidate time series list includes multiple candidate time points and multiple score values corresponding to each candidate time point, and the multiple score values include the score values corresponding to multiple sub-features respectively; A calculation module, configured to substitute the feature weights corresponding to each sub-feature into the score-candidate time series list and calculate to obtain a target prediction sequence list, the target prediction sequence list includes multiple final target candidate time points and multiple total score values; each final target candidate time point corresponds to a total score value; the total score value is determined based on the score values corresponding to each sub-feature in the score-candidate time series list and the feature weights corresponding to the sub-features; An output module, configured to output the final target candidate time point with the largest total score value in the target prediction sequence list as the final target sleep time point.

[0060] On the other hand, the present invention further provides an electronic device, including a processor and a memory; The processor is connected to the memory; The memory is used to store executable program codes; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the sleep time point detection method.

[0061] Wherein, the computer-readable storage medium may include, but is not limited to, any type of disk, including floppy disks, optical disks, DVDs, CD-ROMs, microdrives, and magneto-optical disks, ROMs, RAMs, EPROMs, EEPROMs, DRAMs, VRAMs, flash memory devices, magnetic cards or optical cards, nanosystems (including molecular memory ICs), or any type of medium or device suitable for storing instructions and / or data.

[0062] It should be noted that, for the foregoing method embodiments, for the sake of simple description, they are all expressed as a series of action combinations. However, those skilled in the art should know that this application is not limited by the described action sequence, because according to this application, certain steps can be performed in other sequences or simultaneously. Secondly, those skilled in the art should also know that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.

[0063] In the above embodiments, the descriptions of the respective embodiments have their own emphases. For the parts not detailed in a certain embodiment, reference can be made to the relevant descriptions of other embodiments. In the several embodiments provided in this application, it should be understood that the disclosed device can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only a logical function division. In actual implementation, there can be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed coupling or direct coupling or communication connection to each other can be through some service interfaces. The indirect coupling or communication connection of the device or unit can be in an electrical or other form.

[0064] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0065] In addition, in each embodiment of this application, the functional units can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0066] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable memory. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of this application. The aforementioned memory includes: various media such as USB flash drives, read-only memory (ROM), random access memory (RAM), external hard drives, magnetic disks, or optical discs that can store program codes.

[0067] Those of ordinary skill in the art can understand that all or part of the steps in the various methods of the above embodiments can be completed by instructing relevant hardware through a program. This program can be stored in a computer-readable memory, and the memory can include: flash drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical discs, etc.

[0068] The specific embodiments of this specification have been described above. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims can be executed in a different order from that in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A method for detecting sleep time points based on a non-contact sensor, characterized in that, The method includes: Obtaining discriminant features based on historical data and detection data collected by a non-contact sensor; The discriminant features include multiple sub-features, each sub-feature includes multiple sub-feature values, and each sub-feature value corresponds to a time point; the time interval length between any two adjacent time points is the same; Sorting the sub-feature values extracted from the detection data based on the time order to form candidate time series corresponding to each sub-feature; Calculating the feature weights corresponding to each sub-feature based on the discriminant features obtained from the historical data and the detection data respectively; Merging the candidate time series corresponding to each sub-feature to obtain a total candidate time series, and inputting the total candidate time series into a score model to obtain a score-candidate time series table; the score-candidate time series table includes multiple candidate time points and multiple score values corresponding to each candidate time point, and the multiple score values include the score values corresponding to multiple sub-features respectively; Substituting the feature weights corresponding to each sub-feature into the score-candidate time series table, and calculating to obtain a target prediction sequence table, the target prediction sequence table includes multiple final target candidate time points and multiple total score values; each final target candidate time point corresponds to a total score value; the total score value is determined based on the score values corresponding to each sub-feature in the score-candidate time series table and the feature weights corresponding to the sub-features; Selecting the final target candidate time point with the largest total score value in the target prediction sequence table as the final target sleep time point.

2. The sleep time point detection method based on a non-contact sensor according to claim 1, characterized in that, The specific calculation method of the feature weight corresponding to the sub-feature is: Obtaining the sub-feature quality score and the sub-feature fluctuation value based on the historical data and the data of the current day; Calculating a comprehensive weight index for each sub-feature based on the sub-feature quality score and the sub-feature fluctuation value; Calculating the feature weight of the target sub-feature based on the comprehensive weight index of the sub-feature.

3. The sleep time point detection method based on a non-contact sensor according to claim 2, wherein, One of the multiple sub-features is a body movement feature, and the specific calculation method of the quality score of the body movement feature is: Among them, Q tw represents the mass fraction of the body movement index, sum_tw mean represents the mean value of the total number of body movements in the past z days. sum_tw represents the total number of detected body movements, sum_tw i represents the total number of body movements on the i-th day in the past z days. ε is a correction coefficient, and tw_history represents the maximum value of the difference between the total number of body movements on a single day and the mean value of the total number of body movements in z days in the past z days.

4. The sleep time point detection method based on a non-contact sensor according to claim 2, wherein One of the multiple sub-features is a breathing feature, and the specific calculation method of the quality score of the breathing feature is: Among them, Q br is the mass fraction of the respiration index. The cont of br in[x1,x2] represents the number of respiration sub-feature values whose numerical magnitudes fall within the range of [x1,x2] in the detected valid sleep data; N is the total number of respiration sub-feature values in the detected valid sleep data; [x1,x2] is set based on the standard respiration interval of human sleep.

5. The method for detecting sleep time points based on a non-contact sensor according to claim 2, wherein One of the multiple sub-features is a heart rate feature, and the specific calculation method of the quality score of the heart rate feature is: Among them, Q he is the quality score of the heart rate index. The cont of br in [x3, x4] represents the number of heart rate sub-feature values whose numerical magnitudes fall within the range of [x3, x4] in the detected valid sleep data. N is the total number of heart rate sub-feature values in the detected valid sleep data. [x3, x4] is set based on the standard heart rate range for human sleep.

6. The sleep time point detection method based on a non-contact sensor according to claim 1, characterized in that, Before outputting the time point with the highest score, it also includes a threshold verification step: Extracting historical sleep time points from the historical data; Adding threshold periods in the pre-order time direction and the post-order time direction of the historical sleep time point to form a threshold verification interval; Judging whether the time point with the largest total score value in the target prediction sequence table falls into the threshold verification interval; if it successfully falls, output this time point as the final target sleep time point; if it fails to fall, do not output, and continue to judge whether the time point with the second largest total score value falls into the threshold verification interval until the final target sleep time point is successfully output.

7. A method for detecting sleep time points based on a non-contact sensor according to claim 1, characterized in that, The discriminant features include body movement sub-features, the body movement sub-features include multiple body movement sub-feature values, and when forming the candidate time series corresponding to the body movement sub-features, the adjacent window iteration method is used for time point selection.

8. A sleep time point detection system based on a non-contact sensor, characterized in that, The system includes: A feature extraction module, configured to obtain discriminant features based on historical data and detection data collected by a non-contact sensor; the discriminant features include multiple sub-features, each sub-feature includes multiple sub-feature values, and each sub-feature value corresponds to a time point; A candidate time series generation module, configured to sort each sub-feature value based on time order to form a candidate time series corresponding to each sub-feature; A weight calculation module, configured to calculate the feature weights corresponding to each sub-feature based on historical data and detection data; A sequence list generation module, configured to merge the candidate time series corresponding to each sub-feature to obtain a total candidate time series, and input the total candidate time series into a score model to obtain a score-candidate time series list; the score-candidate time series list includes multiple candidate time points and multiple score values corresponding to each candidate time point, and the multiple score values include the score values corresponding to each of the multiple sub-features; A calculation module, configured to substitute the feature weights corresponding to each sub-feature into the score-candidate time series list, and calculate to obtain a target prediction sequence list, the target prediction sequence list includes multiple final target candidate time points and multiple total score values; each final target candidate time point corresponds to a total score value; the total score value is determined based on the score values corresponding to each sub-feature in the score-candidate time series list and the feature weights corresponding to the sub-features; An output module, configured to output the final target candidate time point with the largest total score value in the target prediction sequence list as the final target sleep time point.

9. An electronic device, characterized in that, It includes a processor and a memory; The processor is connected to the memory; The memory is used to store executable program codes; The processor runs a program corresponding to the executable program code by reading the executable program code stored in the memory, so as to execute the method according to any one of claims 1 to 7.

Images