Sleep assessment method and system

Abstract

The invention discloses a sleep assessment method and system. The method comprises the following steps: acquiring sleep reference data of a target user; acquiring current sleep monitoring data of the target user; based on the sleep reference data and the current sleep monitoring data, determining a multi-dimensional sleep evaluation result; determining a sleep confidence factor based on an effective sleep data proportion meeting a preset duration threshold in a preset time window; and determining a comprehensive sleep assessment result based on the multi-dimensional sleep assessment result and the sleep confidence factor. According to the method and the device, multi-dimensional sleep evaluation is realized by integrating the multi-dimensional sleep reference data and the current monitoring data, and the result is closer to the real state; a sleep confidence factor is introduced to avoid invalid data interference, and the evaluation reliability is improved; personalized reference is generated based on historical data of the user, different sleep habits are adapted, and the pertinence is higher; professional equipment is not needed, convenience and precision are both considered, and popularization of scenes such as home health management is facilitated.

Description

Technical Field

[0001] This application relates to the field of smart home technology, and in particular to a sleep assessment method and system. Background Technology

[0002] Sleep health has become a key factor affecting human physical and mental well-being. Sleep consistency, as a core indicator reflecting the stability of sleep patterns, is of great significance for accurate assessment in scenarios such as sleep disorder diagnosis and health management. Current sleep consistency assessment methods are mainly divided into two categories: subjective assessment methods and objective assessment methods. However, related technical solutions have significant limitations in practical applications, making it difficult to meet users' needs for accuracy, convenience, and interference resistance in sleep consistency assessment.

[0003] In the field of subjective assessment methods, sleep diaries and various sleep assessment scales are the mainstream approaches. Sleep diaries rely on users recording information such as sleep onset time, number and duration of awakenings, total sleep duration, and daytime functional status. They subjectively judge sleep continuity and consistency based on these parameters. While simple and easy to implement, and long-term recording can reflect long-term sleep patterns to some extent, this method relies heavily on subjective recall, inevitably leading to memory bias and insufficient accuracy. Commonly used sleep assessment scales (such as the Pittsburgh Sleep Quality Index (PSQI) and the Self-Rating Sleep Status Scale (SRSS)) assess sleep through standardized question scoring. For example, the PSQI includes nine questions about the frequency and cause of nighttime awakenings, scored from 0 to 3 based on severity. The SRSS has 10 items related to the number and duration of nighttime awakenings, scored from 1 to 5, with higher total scores indicating poorer sleep continuity. These methods require long-term tracking of scale results or the creation of custom sleep consistency scales. However, similar to sleep diaries, they are significantly influenced by subjective factors, making it difficult to guarantee the objectivity and reliability of the assessment results.

[0004] In the field of objective assessment methods, polysomnography (PSG) is the "gold standard" for sleep staging. It acquires physiological signals and completes sleep staging to assess sleep continuity and consistency. However, this device is complex to operate, requires highly skilled operators, and is usually used in professional medical institutions, making it difficult to meet the convenient assessment needs of everyday home scenarios. In addition, some existing objective methods judge sleep consistency by statistically analyzing the variance of sleep onset time and wake-up time over a period of time. While these methods simplify the assessment, they only focus on a single time dimension and have extremely poor anti-interference capabilities. If a few days within the window contain abnormal data such as sleeping too early or too late, it will directly lead to a significant deviation in the consistency assessment results for the entire window, failing to accurately reflect the stability of the user's true sleep patterns.

[0005] In summary, existing sleep consistency assessment methods generally suffer from problems such as insufficient accuracy of subjective assessment, poor convenience or weak anti-interference ability of objective assessment, and single assessment dimensions, making it difficult to achieve a comprehensive and accurate assessment of sleep consistency. Summary of the Invention

[0006] Therefore, it is necessary to provide a sleep assessment method and system to address the aforementioned technical problems and solve at least one of the problems existing in the prior art.

[0007] Firstly, a sleep assessment method is provided, including: Obtain sleep reference data of the target user, wherein the sleep reference data includes at least one of sleep state reference data, sleep structure reference data, and sleep quality reference data; Obtain the current sleep monitoring data of the target user; Based on the sleep reference data and current sleep monitoring data, a multi-dimensional sleep assessment result is determined. The multi-dimensional sleep assessment result includes at least one of the following: sleep state assessment result, sleep structure assessment result, and sleep quality assessment result. The sleep confidence factor is determined based on the proportion of valid sleep data that meets the preset duration threshold within a preset time window. Based on the multidimensional sleep assessment results and the sleep confidence factor, a comprehensive sleep assessment result is determined.

[0008] In one possible implementation, the sleep state reference data is obtained in the following manner: Obtain the target user's historical sleep data; Based on the historical sleep data, generate multiple sleep state arrays corresponding to preset time periods; At the same preset time point, the sleep state arrays are filtered; The sleep state array after filtering for the previous preset time period is truncated to obtain a sleep state reference sequence, which is used as the sleep state reference data. The sleep structure reference data is obtained in the following manner: Based on the target user’s historical sleep data, determine the duration and percentage of each historical sleep stage. Preprocessing is performed on the historical sleep stage percentage and the historical sleep stage duration along the time dimension to obtain reference values ​​for sleep stage percentage and sleep process duration. Based on the aforementioned reference values ​​for the proportion of sleep stages and the reference values ​​for the duration of sleep processes, a reference value for the duration of sleep stages is determined. The reference values ​​for the percentage of sleep stages and the reference values ​​for the duration of sleep stages are used as reference data for the sleep structure.

[0009] In one possible implementation, the sleep state reference data includes a sleep state reference sequence, and the sleep state assessment results are obtained in the following manner: Based on the current sleep monitoring data, a current sleep state array is obtained; A segment with the same length as the current sleep state array is extracted from the sleep state reference sequence and used as the real-time sleep state reference sequence. Based on the current sleep state array and the real-time sleep state reference sequence, a paired state array is determined; Based on the real-time sleep state reference sequence, the weight distribution of the real-time sleep state is determined; Based on the paired state array and the real-time sleep state weight distribution, a sleep state consistency score is calculated and used as the sleep state evaluation result.

[0010] In one possible implementation, determining the real-time sleep state weight distribution based on the real-time sleep state reference sequence includes: Based on the real-time sleep state reference sequence, the sleep start time point and the sleep end time point are obtained; Based on the sleep start time and sleep end time, determine the sleep midpoint. The real-time sleep state weight distribution is calculated based on a preset weight distribution template, the sleep start time, the sleep end time, and the sleep midpoint.

[0011] In one possible implementation, obtaining the sleep start time point and sleep end time point based on the real-time sleep state reference sequence includes: Based on the real-time sleep state reference sequence, the initial search starting point at the start time of the target sleep interval and the initial search starting point at the end time are determined. The sleep start point and sleep end point are determined by performing a reverse search along the time dimension based on the initial search starting point of the start time point and a forward search along the time dimension based on the initial search starting point of the end time point.

[0012] In one possible implementation, the process of performing a reverse search along the time dimension based on the initial search starting point of the start time point and a forward search along the time dimension based on the initial search starting point of the end time point to determine the sleep start time point and the sleep end time point includes: Starting from the initial search point at the beginning time, the search proceeds backward along the time dimension. If a target sleep segment exists in the preceding sequence and the time interval between the end time of the target sleep segment and the initial search point is less than a preset threshold, the initial search point is updated to the starting point of the preceding segment and the search is repeated. If no preceding target sleep segment meets the conditions, the current search point is determined as the sleep start time. Starting from the initial search point at the end time, search forward along the time dimension. If a target sleep segment exists subsequently and the time interval between the start time of the target sleep segment and the initial search point is less than the preset threshold, then update the initial search point to the end point of the subsequent target sleep segment and repeat the search. If no subsequent target sleep segment meets the conditions, then determine the current search point as the end time of sleep.

[0013] In one possible implementation, determining the initial search starting point for the start time of the target sleep interval and the initial search starting point for the end time based on the real-time sleep state reference sequence includes: If there is a target sleep segment in the real-time sleep state reference sequence that meets the preset duration condition, the start time node of the first target sleep segment is selected as the initial search starting point of the target sleep interval, and the end time node of the last target sleep segment is selected as the initial search starting point of the target sleep interval. If there is no target sleep segment in the real-time sleep state reference sequence that meets the preset duration condition, the start time node and end time node of the sleep segment with the longest sleep duration in the real-time sleep state reference sequence are selected as the initial search starting points for the start and end times of the target sleep interval, respectively.

[0014] In one possible implementation, the sleep structure assessment results are obtained in the following manner: Based on the current sleep monitoring data, determine the duration of each current sleep stage and the percentage of each current sleep stage. Calculate the first relative difference between the duration of each current sleep stage and the corresponding reference value for the duration of the sleep stage, and the second relative difference between the proportion of each current sleep stage and the corresponding reference value for the proportion of the sleep stage. According to the preset sleep structure weighting coefficient and difference attenuation coefficient, the first relative difference and the second relative difference are weighted and fused to obtain the sleep structure consistency score, which is used as the sleep structure evaluation result.

[0015] In one possible implementation, the sleep quality reference data includes sleep efficiency reference values ​​and sleep continuity reference values, and the sleep quality assessment results are obtained in the following manner: Based on the current sleep monitoring data, determine the actual value of the current sleep efficiency and the actual value of the current sleep continuity; Based on the sleep quality reference data, obtain the corresponding sleep efficiency reference value and sleep continuity reference value; Calculate the third relative difference between the current actual value of sleep efficiency and the corresponding reference value of sleep efficiency, and the fourth relative difference between the current actual value of sleep continuity and the corresponding reference value of sleep continuity. The third and fourth relative differences are weighted and fused according to a preset sleep quality weighting coefficient to obtain a sleep quality consistency score, which is used as the sleep quality assessment result.

[0016] Secondly, a sleep assessment system is provided, including: A sleep reference data acquisition unit is used to acquire sleep reference data of a target user, wherein the sleep reference data includes at least one of sleep state reference data, sleep structure reference data, and sleep quality reference data. The current sleep monitoring data acquisition unit is used to acquire the current sleep monitoring data of the target user; A multidimensional sleep assessment result determination unit is used to determine multidimensional sleep assessment results based on the sleep reference data and current sleep monitoring data. The multidimensional sleep assessment results include at least one of sleep state assessment results, sleep structure assessment results, and sleep quality assessment results. The sleep confidence factor determination unit is used to determine the sleep confidence factor based on the proportion of valid sleep data that meets a preset duration threshold within a preset time window. The comprehensive sleep assessment result determination unit is used to determine the comprehensive sleep assessment result based on the multi-dimensional sleep assessment results and the sleep confidence factor.

[0017] The aforementioned sleep assessment method and system, implemented as follows: acquiring sleep reference data of a target user, wherein the sleep reference data includes at least one of sleep state reference data, sleep structure reference data, and sleep quality reference data; acquiring current sleep monitoring data of the target user; determining multi-dimensional sleep assessment results based on the sleep reference data and the current sleep monitoring data, wherein the multi-dimensional sleep assessment results include at least one of sleep state assessment results, sleep structure assessment results, and sleep quality assessment results; determining a sleep confidence factor based on the proportion of valid sleep data meeting a preset duration threshold within a preset time window; and determining a comprehensive sleep assessment result based on the multi-dimensional sleep assessment results and the sleep confidence factor. In this embodiment, by integrating three types of reference data—sleep state, sleep structure, and sleep quality—with current sleep monitoring data, the limitations of existing methods in single-dimensional assessment are overcome. Sleep conditions are depicted from multiple dimensions, including sleep patterns, stage characteristics, and core indicators, resulting in assessment results that more closely reflect the user's actual sleep state. By introducing a sleep confidence factor based on the proportion of effective sleep data within a preset time window, the impact of data validity on the assessment results is dynamically quantified, effectively avoiding assessment distortion caused by invalid or incomplete data, and improving the robustness and credibility of the assessment results. Personalized reference data is generated based on the target user's historical sleep data, abandoning existing general standard assessment models and adapting to the sleep habits and physiological characteristics of different users, resulting in more targeted assessments and a wider range of applications. Furthermore, no professional equipment or operation is required; it can be achieved based on routine sleep monitoring data. Simultaneously, multi-dimensional reference comparison and confidence optimization ensure assessment accuracy, balancing convenience and accuracy, making it suitable for widespread application in scenarios such as home health management and sleep disorder screening. Attached Figure Description

[0018] To more clearly illustrate the technical solutions of the embodiments of this application, the drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a schematic diagram of an application environment for a sleep assessment method according to an embodiment of this application; Figure 2 This is a flowchart illustrating a sleep assessment method in one embodiment of this application; Figure 3 This is a schematic diagram of a user's real-time sleep staging results in one embodiment of this application; Figure 4 This is a schematic diagram of the distribution of historical sleep data in one embodiment of this application; Figure 5This is a schematic diagram of sleep-wake data distribution in one embodiment of this application; Figure 6 This is a schematic diagram of the filtered sleep-wake data distribution in one embodiment of this application; Figure 7 This is a schematic diagram of the structure of a sleep assessment system according to one embodiment of this application; Figure 8 This is a schematic diagram of a computer device according to one embodiment of this application. Detailed Implementation

[0020] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0021] The sleep assessment method provided in this embodiment can be applied to, for example, Figure 1 The application environment shown includes a server S1, a client S2, and a device S3, which can communicate and connect with each other.

[0022] Specifically, the server S1 can be implemented using a standalone server or a server cluster consisting of multiple servers. It can be a local server or a cloud server. It is used for user data management, specifically including user data storage and information distribution. This information includes, but is not limited to, user personal information, historical sleep data, and historical physiological parameter information. User personal information includes, but is not limited to, height, weight, age, and gender; historical sleep data includes, but is not limited to, the user's sleep state information at various points in time, including bedtime, bedtime, sleep onset time, and wakefulness time; the sleep state information can be categorized as: bed-out, awake, light sleep, deep sleep, rapid eye movement (REM), or multiple classification formats such as bed-out, awake, N1, N2, N3, and REM; historical physiological parameter information includes: resting heart rate and resting respiratory rate, which refer to the user's average heart rate and average respiratory rate in a resting state, respectively. The resting state refers to the time segment in which the user remains motionless for more than the threshold TH6 (e.g., 5 minutes) and has not entered sleep. Resting heart rate and resting respiratory rate are updated once daily at a fixed time.

[0023] Client S2 includes, but is not limited to, various personal computers, laptops, smartphones, tablets and portable wearable devices, which are pre-installed with APPs or mini-programs for displaying comprehensive sleep assessment results, entering user personal information, and uploading relevant data to the server. The main carrier of the S3 device can be smart home devices such as smart bed frames or smart mattresses, and it can include five parts: signal acquisition module, signal processing module, sleep staging module, sleep consistency calculation module, and consistency information management module. The responsibilities of each module are as follows: The signal acquisition module is used to collect sensor data. The type of sensor data can be a piezoelectric sensor signal placed directly below the shoulder, or a photoplethysmography signal collected from the fingers, earlobe, or wrist. The signal processing module is used to perform noise filtering and feature extraction on the acquired sensor signals. The noise filtering can be a simple infinite impulse corresponding Butterworth filter, or a combination of filters, or an autoencoder filter based on a deep learning model. The features extracted include the heartbeat interval array (the heartbeat interval is the time difference between adjacent heartbeat peaks), the standard deviation of the heartbeat interval (SDNN), the respiratory interval array (the expiratory interval is the time difference between adjacent respiratory wave peaks), the standard deviation of the respiratory interval, and body movement intensity. The sleep staging module is used to classify the current sleep state based on the feature array composed of features extracted by the signal processing module and a preset sleep staging model. The feature array includes, but is not limited to, the heart interval array, respiratory interval array, standard deviation of heart interval, standard deviation of respiratory interval, body movement intensity, current sleep state, and duration of current sleep state within the current window time. The preset sleep staging model can be a deep learning model or a traditional machine learning model such as random forest. The sleep consistency calculation module is used to calculate sleep consistency parameters, specifically including: based on the real-time sleep status and sleep reference data output by the sleep staging module, calculating the consistency differences of sleep status, sleep structure, and sleep quality, and outputting the corresponding consistency score; The consistency information management module is used for storing, synchronizing, and interacting with sleep consistency data. It can synchronize sleep data with the server and convert sleep consistency results and difference analysis results into visualized data, which can then be pushed to the device display interface or associated user terminals for display.

[0024] The specific implementation process can be as follows: Extract at least 30 days of historical sleep monitoring data from the server, filter out valid data with a sleep duration of ≥3 hours, extract and calculate the user's sleep-wake binary array, the intermediate time point between falling asleep / wake / sleep, the duration and proportion of each sleep stage, sleep efficiency, sleep continuity, and other data, and generate the user's inherent sleep pattern dataset after filtering; then collect the user's current physiological signals through sensors, and obtain feature data such as heart rate interval, respiratory interval, and body movement intensity after noise filtering and feature extraction, classify the real-time sleep state using a preset sleep stage model, and generate the sleep monitoring data for the day; then calculate the consistency scores of sleep state, structure, and quality, and combine them with the sleep confidence factor determined by the proportion of valid sleep data within a preset time window to obtain the user's comprehensive sleep assessment result, which can finally be sent to the client for visualization and updating of the user's historical sleep data.

[0025] In one embodiment, such as Figure 2 As shown, a sleep assessment method is provided, which can be applied to... Figure 1 Taking the device side as an example, the explanation includes the following steps: In step S110, sleep reference data of the target user is obtained, and the sleep reference data includes at least one of sleep state reference data, sleep structure reference data, and sleep quality reference data; Among them, the sleep state reference data is a sleep state sequence generated based on the user's historical sleep data after filtering and processing, which is the sleep-wake array. In this sleep-wake array, the time corresponding to getting out of bed and waking up can be set to 0, and the time corresponding to light sleep, deep sleep, REM, and nap can be set to 1.

[0026] Sleep structure reference data may include sleep stage percentage reference information and sleep stage duration reference information. The sleep stage percentage reference information includes reference values ​​for the proportions of deep sleep, light sleep, REM sleep, and wakefulness throughout the night. The sleep stage duration reference information includes reference values ​​for the durations of deep sleep, light sleep, REM sleep, and wakefulness throughout the night. These reference values ​​are obtained by filtering the acquired historical sleep structure information of the user within a window time period.

[0027] Sleep quality reference data includes sleep efficiency reference values ​​and sleep continuity reference values. Sleep efficiency is calculated as sleep duration / time spent in bed × 100% (sleep duration refers to actual sleep time, and time spent in bed refers to the total time from falling asleep to waking up). Sleep continuity is assessed using sleep stage characteristics and micro-arousal characteristics during the sleep process, evaluated through a preset sleep continuity calculation model. The sleep stage characteristics include the stage of each sleep segment, the duration of that stage, and the sleep cycle number of that stage. The micro-arousal characteristics include micro-arousal intensity, micro-arousal duration, the sleep stage of the micro-arousal, the sleep cycle of the micro-arousal, and the sleep duration before the micro-arousal. The preset sleep continuity calculation model can be a simple linear model or a deep learning model. The above reference values ​​are obtained by filtering the user's historical sleep quality information within a window time.

[0028] In step S120, the current sleep monitoring data of the target user is obtained; Optionally, the current sleep monitoring data acquisition rule is as follows: if the data acquisition time is earlier than 18:00 on the current day, then sleep data from 18:00 on the previous day to the current time are used; otherwise, sleep data from 18:00 on the current day to the current time are used. This current sleep monitoring data may include the user's real-time sleep status data, duration of each phase, percentage of each phase, sleep efficiency, sleep continuity, etc. An example of the user's real-time sleep status data is shown below. Figure 3 As shown, the current acquisition time is approximately 14:00.

[0029] In step S130, based on the sleep reference data and the current sleep monitoring data, a multi-dimensional sleep assessment result is determined. The multi-dimensional sleep assessment result includes at least one of the following: sleep state assessment result, sleep structure assessment result, and sleep quality assessment result. Optionally, the multi-dimensional assessment result includes at least one of the three categories of assessment results: sleep state, sleep structure, and sleep quality. Specifically, a sleep state assessment result is generated by comparing the current sleep state sequence with the inherent sleep-wake array in the sleep state reference data and combining it with a personalized time weight distribution; a sleep structure assessment result is generated by quantifying the difference between the duration and proportion of each current sleep stage and the corresponding reference value in the sleep structure reference data and combining it with a preset weight; and a sleep quality assessment result is generated by calculating the relative differences between the current sleep efficiency and sleep continuity and the corresponding reference values ​​in the sleep quality reference data and combining it with preset indicator weights.

[0030] In step S140, the sleep confidence factor is determined based on the proportion of valid sleep data that meets the preset duration threshold within the preset time window. Optionally, a time window TH1 (e.g., 30 days) and a minimum sleep duration threshold TH5 (e.g., 3 hours) are set. Then, the number of effective sleep days that the user actually obtains and meets the duration requirement within the window is counted. Effective sleep data refers to sleep data with a sleep duration greater than or equal to TH5. The percentage is calculated by dividing the number of effective days by the total number of days in the window TH1. This percentage is the sleep confidence factor, which is used to measure the reliability and stability of the user's sleep data.

[0031] In step S150, a comprehensive sleep assessment result is determined based on the multidimensional sleep assessment results and the sleep confidence factor.

[0032] Optionally, based on the obtained multi-dimensional sleep assessment results (including at least one consistency score such as sleep state, sleep structure, and sleep quality), and combined with the sleep confidence factor, the confidence factor is used as a weight coefficient to quantitatively correct the multi-dimensional assessment results through a preset weighted fusion rule (for example, the higher the confidence factor, the greater the weight given to the multi-dimensional assessment results, and vice versa, the weight is appropriately reduced to avoid the distortion of assessment results due to poor data quality). Finally, a comprehensive sleep assessment result that can comprehensively reflect the level of sleep consistency and data reliability is generated.

[0033] It should be noted that if the multidimensional sleep assessment result only includes a consistency score for one of the sleep state, sleep structure, and sleep quality, then the consistency score of that single dimension is weighted and fused with the sleep confidence factor to directly obtain the comprehensive sleep assessment result. If the multidimensional sleep assessment result includes two or three consistency scores, then the consistency scores of each dimension are first weighted and integrated into a total multidimensional sleep assessment score according to the preset dimension weight rules (for example, 0.6 is assigned to the consistency score of sleep quality, and 0.2 is assigned to sleep state and structure respectively). Then, this total score is weighted and fused with the sleep confidence factor to finally obtain the comprehensive sleep assessment result.

[0034] This application provides a sleep assessment method, comprising: acquiring sleep reference data of a target user, the sleep reference data including at least one of sleep state reference data, sleep structure reference data, and sleep quality reference data; acquiring current sleep monitoring data of the target user; determining a multi-dimensional sleep assessment result based on the sleep reference data and the current sleep monitoring data, the multi-dimensional sleep assessment result including at least one of sleep state assessment result, sleep structure assessment result, and sleep quality assessment result; determining a sleep confidence factor based on the proportion of valid sleep data meeting a preset duration threshold within a preset time window; and determining a comprehensive sleep assessment result based on the multi-dimensional sleep assessment result and the sleep confidence factor. In this embodiment, by integrating three types of reference data—sleep state, sleep structure, and sleep quality—with current sleep monitoring data, the limitations of existing methods in single-dimensional assessment are overcome. Sleep conditions are depicted from multiple dimensions, including sleep patterns, stage characteristics, and core indicators, resulting in assessment results that more closely reflect the user's actual sleep state. By introducing a sleep confidence factor based on the proportion of effective sleep data within a preset time window, the impact of data validity on the assessment results is dynamically quantified, effectively avoiding assessment distortion caused by invalid or incomplete data, and improving the robustness and credibility of the assessment results. Personalized reference data is generated based on the target user's historical sleep data, abandoning existing general standard assessment models and adapting to the sleep habits and physiological characteristics of different users, resulting in more targeted assessments and a wider range of applications. Furthermore, no professional equipment or operation is required; it can be achieved based on routine sleep monitoring data. Simultaneously, multi-dimensional reference comparison and confidence optimization ensure assessment accuracy, balancing convenience and accuracy, making it suitable for widespread application in scenarios such as home health management and sleep disorder screening.

[0035] In one embodiment of this application, the sleep state reference data is obtained in the following manner: Obtain the target user's historical sleep data; Based on the historical sleep data, generate multiple sleep state arrays corresponding to preset time periods; At the same preset time point, the sleep state arrays are filtered; The sleep state array after filtering for the previous preset time period is truncated to obtain a sleep state reference sequence, which is used as the sleep state reference data. Optionally, the historical sleep data of the target user within a preset window time threshold TH1 (e.g., 30 days) can be obtained first. This historical sleep data may include an array of sleep state information corresponding to each moment of the target user, such as bedtime, bedtime, sleep onset time, and wake-up time. An example of the distribution of the sleep state information array at each moment can be found here. Figure 4 , Figure 4The graph displays the results of 75 days of continuous sleep monitoring for a single user. The horizontal axis represents the time sequence; if the horizontal axis is 1, it indicates day 1. The vertical axis represents the time points, from bottom to top, from 18:00 of the previous day to 18:00 of the current day. The sleep states shown in the graph include out of bed, awake, light sleep, deep sleep, REM sleep, and nap. Then, sleep-related data (such as light sleep, deep sleep, REM sleep, and nap) and awake data (such as out of bed and awake) are extracted from historical sleep data. The times corresponding to out of bed and awake are set to 0, and the times corresponding to light sleep, deep sleep, REM sleep, and nap are set to 1. This forms a sleep state array, namely the sleep-awake array.

[0036] The criteria for classifying a sleep segment as a nap are as follows: if the time interval between this sleep segment and the main sleep segment exceeds 3 hours, and the occurrence time of this sleep segment is not within a specified time segment, then this sleep segment is classified as a nap. The main sleep segment refers to the time segment with the longest sleep duration in the sleep period data for that day; the specified time segment can be from 0:00 to 8:00. The time resolution of the sleep-wake array can be 30 seconds, meaning one state value corresponds to every 30 seconds. The daily sleep-wake data length is 2880. An example of the processed sleep-wake distribution is shown below. Figure 5 As shown; Low-pass filtering (such as mean filtering, finite impulse response (FIR) Butterworth low-pass filtering, or infinite impulse response (IIR) Butterworth low-pass filtering, etc.) is applied to the sleep-wake array at the same time every day. Taking Gaussian filtering as an example, the Gaussian filter template is as follows: ; Where i corresponds to the index of the filter template, that is, the i-th value of the template; Pi; The standard deviation is denoted as σ, which can be 20; e is the natural constant; N is the mean of the template length, taking a template length of 60 as an example, where N is 30; the obtained Gaussian filter template is then normalized, and the normalized Gaussian coefficients are... The acquisition process is as follows: ; Where M is the length of the Gaussian template, i.e., 60; the first half of the Gaussian filter template is truncated, and all coefficients are multiplied by 2 to ensure that the sum of the filter coefficients is 1. The coefficient array processed above is used as the sleep-wake filter array. The sleep-wake state distribution after filtering is as follows: Figure 6 As shown in the diagram, brighter areas represent a greater likelihood of being asleep, while darker areas represent a greater likelihood of being awake. Figure 6 In the initial period, due to insufficient data, daily sleep patterns had a significant impact on the filtered results, resulting in greater fluctuations compared to subsequent time periods.

[0037] Finally, the sleep-wake array after filtering the previous preset time period (such as the complete sleep cycle of the previous day) can be truncated for sleep probability at each moment, and the sequence corresponding to the effective sleep period can be extracted as the sleep state reference sequence (such as the sequence corresponding to the effective sleep period with a filtered state value of 1), thus obtaining the sleep state sequence under the user's inherent sleep mode.

[0038] Among them, the sleep probability truncation process refers to setting the values ​​in the filtered sleep-wake array that are greater than the preset threshold TH2 (e.g., 0.5) to 1, which represents the sleep state, and setting the values ​​that are less than the preset threshold TH2 to 0, which represents the wake state.

[0039] In one embodiment of this application, the sleep structure reference data is obtained in the following manner: Based on the target user’s historical sleep data, determine the duration and percentage of each historical sleep stage. Preprocessing is performed on the historical sleep stage percentage and the historical sleep stage duration along the time dimension to obtain reference values ​​for sleep stage percentage and sleep process duration. Based on the aforementioned reference values ​​for the proportion of sleep stages and the reference values ​​for the duration of sleep processes, a reference value for the duration of sleep stages is determined. The reference values ​​for the percentage of sleep stages and the reference values ​​for the duration of sleep stages are used as reference data for the sleep structure.

[0040] The reference values ​​for the proportion of sleep stages include the proportion of deep sleep, light sleep, REM sleep, and wakefulness during the entire sleep cycle (such as the whole night's sleep); the reference values ​​for the duration of sleep stages include the reference values ​​for the duration of deep sleep, light sleep, REM sleep, and wakefulness during the entire night's sleep.

[0041] Optionally, the historical sleep data of the target user within a preset window time threshold TH1 (e.g., 30 days) can be obtained first. This historical sleep data may include an array of sleep state information corresponding to each moment of the target user, such as bedtime, bedtime, sleep onset time, and wakefulness time. Then, the duration and percentage of each phase between the daily sleep onset time and wakefulness time are extracted from the historical sleep data. Each phase includes deep sleep, light sleep, REM sleep, and wakefulness. During the statistics, the bed-out state and the wakefulness state are combined into a "wakefulness" category and uniformly participate in the percentage and duration calculation. The percentage is the ratio of the duration of each phase to the time interval between the sleep onset time and the wakefulness time × 100%.

[0042] Preprocess the daily percentage of each sleep stage: First, perform normalization (i.e., percentage of each stage ÷ sum of percentages of each stage × 100%) to ensure that the sum of the percentages of deep sleep, light sleep, REM sleep, and wakefulness is 100%; then, perform low-pass filtering (such as mean filtering, finite impulse response (FIR) Butterworth low-pass filtering, or infinite impulse response (IIR) Butterworth low-pass filtering, etc.) on the normalized percentage data along the time dimension to obtain reference values ​​for the percentage of sleep stages.

[0043] Similarly, by performing low-pass filtering on the daily sleep duration (the time interval between falling asleep and waking up) along the time dimension, a reference value for the sleep duration is obtained.

[0044] Based on the formula: Sleep stage duration reference value = Sleep stage percentage reference value × Sleep process duration reference value ÷ 100%, the duration reference values ​​for deep sleep, light sleep, REM sleep, and wakefulness are calculated respectively. The sleep stage percentage reference value and the sleep stage duration reference value are used as sleep structure reference data.

[0045] In addition, sleep quality reference information may include: sleep efficiency reference value and sleep continuity reference value. Sleep efficiency is calculated as: Sleep efficiency = (Sleep duration ÷ Bedtime) × 100%. Sleep continuity is assessed using sleep stage characteristics and micro-arousal characteristics during sleep, through a pre-defined sleep continuity calculation model. Sleep stage characteristics include the stage of each sleep segment, the duration of that stage, and the sleep cycle number of that stage. Micro-arousal characteristics include micro-arousal intensity, micro-arousal duration, the sleep stage of the micro-arousal, the sleep cycle of the micro-arousal, and the sleep duration before the micro-arousal. It should be noted that the pre-defined sleep continuity calculation model can be a simple linear model or a deep learning model. The above reference values ​​can be obtained by filtering the user's historical sleep quality information within a window of time.

[0046] In one embodiment of this application, the sleep state reference data includes a sleep state reference sequence, and the sleep state assessment result is obtained in the following manner: Based on the current sleep monitoring data, a current sleep state array is obtained; A segment with the same length as the current sleep state array is extracted from the sleep state reference sequence and used as the real-time sleep state reference sequence. Based on the current sleep state array and the real-time sleep state reference sequence, a paired state array is determined; Based on the real-time sleep state reference sequence, the weight distribution of the real-time sleep state is determined; Based on the paired state array and the real-time sleep state weight distribution, a sleep state consistency score is calculated and used as the sleep state evaluation result.

[0047] Optionally, sleep-related data (such as light sleep, deep sleep, REM sleep, nap, etc.) and wakefulness data (such as getting out of bed, awake, etc.) are extracted from the target user's current sleep monitoring data, and a current sleep state array is generated, that is, the times corresponding to getting out of bed and awakening are set to 0, and the times corresponding to light sleep, deep sleep, REM sleep, and nap are set to 1. It is worth noting that since the real-time sleep state reference sequence for the day is derived from the user's historical data based on the previous day's front window time length TH1, this length matching means that because the user may have acquired data before 18:00 on the same day (because the user's data acquisition time may be earlier than 18:00 on the same day, the length of the sleep-wake array for the day is not fixed), the length of the sleep-wake array for the day is not fixed. Therefore, it is necessary to extract the same length as the sleep-wake array for the day from the beginning of the real-time sleep state reference sequence as the matching real-time sleep state reference sequence. Then, based on the current sleep state array and the real-time sleep state reference sequence, it can be determined whether their states are consistent at the same time point, generating a paired state array. That is, if the value of the current sleep state array at the same array index is consistent with the value of the matching real-time sleep state reference sequence, the value of the paired state array at the corresponding index position is set to 1; otherwise, it is set to 0, thus obtaining the paired state array. Next, based on the sleep state distribution characteristics at each time point in the real-time sleep state reference sequence (such as key periods such as falling asleep, deep sleep, and wakefulness), the real-time sleep state weight distribution is determined according to a preset weighting rule. Finally, the paired state array and the real-time sleep state weight distribution are weighted and summed to obtain a sleep state consistency score, which is used as the sleep state evaluation result.

[0048] In one embodiment of this application, determining the real-time sleep state weight distribution based on the real-time sleep state reference sequence includes: Based on the real-time sleep state reference sequence, the sleep start time point and the sleep end time point are obtained; Based on the sleep start time and sleep end time, determine the sleep midpoint. The real-time sleep state weight distribution is calculated based on a preset weight distribution template, the sleep start time, the sleep end time, and the sleep midpoint.

[0049] Optionally, the sleep onset time point can be identified based on a real-time sleep state reference sequence. With the end of sleep Then, the midpoint of sleep is calculated based on the average of these two time points. Simultaneously, the time interval H between the two time points is calculated, based on the sleep start time point. Sleep end time Mid-sleep time point The time interval H between the start and end of sleep and a preset weight distribution template are used to calculate the real-time sleep state weight distribution. This involves mapping the key parameters (such as mean and standard deviation) of the preset weight distribution template (such as Gaussian distribution) to the start, end, and intermediate points of sleep, and calculating the weights corresponding to each point of sleep using a preset algorithm, thereby generating the real-time sleep state weight distribution.

[0050] The preset weight distribution template is shown below: ; Where t is the current time point, These are the weighting coefficients at different time points. The variance is a Gaussian distribution.

[0051] It should be noted that, and It can be obtained by parameter fitting on a specific dataset 1. The fitting process refers to using the paired state array of the day as input features, based on a preset weight distribution template, and using the sleep state consistency score as the target value to perform fitting, or hyperparameter search.

[0052] Specifically, Specific Dataset 1 refers to a sleep dataset whose continuous sleep data collection duration reaches or exceeds TH1 (e.g., 30 days), during which there is no continuous invalid sleep data (the proportion of records with a single-day sleep duration less than TH5 is less than a preset threshold), and which carries a sleep state consistency score label. Invalid sleep data refers to sleep data with a sleep duration less than TH5 (e.g., 3 hours). The sleep state consistency score label refers to the sleep state consistency score performed manually by multiple (e.g., 3) professionals on the sleep data within each user's window duration TH1. The score is calculated based on the sleep consistency score scale and paired state array filled in by each user. The paired state array refers to the sleep-wake state array of the last day within the window time TH1 and the reference value of the sleep-wake array of the last day within the window time TH1. The average value of the scores from multiple people is used to obtain the final sleep state consistency score label.

[0053] In one embodiment of this application, obtaining the sleep start time point and sleep end time point based on the real-time sleep state reference sequence includes: Based on the real-time sleep state reference sequence, the initial search starting point at the start time of the target sleep interval and the initial search starting point at the end time are determined. The sleep start point and sleep end point are determined by performing a reverse search along the time dimension based on the initial search starting point of the start time point and a forward search along the time dimension based on the initial search starting point of the end time point.

[0054] Optionally, the initial search starting point for the start time of the target sleep interval and the initial search starting point for the end time are determined in the real-time sleep state reference sequence; then, the search proceeds backward along the time dimension from the initial search starting point of the start time to find the first time point from wakefulness to sleep as the sleep start time point, and simultaneously, the search proceeds forward along the time dimension from the initial search starting point of the end time to find the first time point from sleep to wakefulness as the sleep end time point.

[0055] Furthermore, to improve search accuracy, a continuous state verification mechanism can be implemented. For example, during reverse search, after detecting L consecutive time units of sleep, the search should trace back to the first moment when the user transitioned from wakefulness to sleep. Similarly, during forward search, it should be confirmed that L consecutive time units of wakefulness are observed before using the first moment when the user transitioned from sleep to wakefulness as the end point. Additionally, a dynamic threshold adjustment strategy can be introduced, adaptively adjusting the L value based on the user's historical sleep pattern data to accommodate differences in sleep habits among different users.

[0056] Forward search refers to traversing data point by point in the direction of time flow (from morning to night) starting from the initial search point. In sleep detection scenarios, it is often used to start from the initial search point of the target interval and search backward to find the first moment when the sleep state transitions to the awake state, that is, the moment when sleep ends.

[0057] Reverse search refers to traversing data point by point in reverse time (from night to morning) starting from the initial search point. In sleep detection scenarios, it is often used to start from the initial search point of the target interval and look backward to find the first moment when the state of wakefulness transitions to sleep, i.e., the moment when sleep begins.

[0058] In one embodiment of this application, the step of performing a reverse search along the time dimension based on the initial search starting point of the start time point and a forward search along the time dimension based on the initial search starting point of the end time point to determine the sleep start time point and the sleep end time point includes: Starting from the initial search point at the beginning time, the search proceeds backward along the time dimension. If a target sleep segment exists in the preceding sequence and the time interval between the end time of the target sleep segment and the initial search point is less than a preset threshold, the initial search point is updated to the starting point of the preceding segment and the search is repeated. If no preceding target sleep segment meets the conditions, the current search point is determined as the sleep start time. Starting from the initial search point at the end time, search forward along the time dimension. If a target sleep segment exists subsequently and the time interval between the start time of the target sleep segment and the initial search point is less than the preset threshold, then update the initial search point to the end point of the subsequent target sleep segment and repeat the search. If no subsequent target sleep segment meets the conditions, then determine the current search point as the end time of sleep.

[0059] Optionally, the search proceeds backward along the time dimension from the initial search starting point, examining sleep state data prior to that starting point segment by segment. If a preceding target sleep segment (e.g., a sleep segment with a duration greater than a preset threshold TH3 (e.g., 3 hours)) is found, and the time interval between the end point of that target sleep segment and the current initial search starting point is less than a preset threshold TH4 (e.g., 15 minutes), then the two sleep segments are determined to belong to the same complete sleep cycle. The initial search starting point is then updated to the starting point of the preceding target sleep segment, and the backward search continues based on this new starting point. If, after continuous backward search, no preceding target sleep segment meeting the time interval condition is found, then the current search starting point is determined as the final sleep start point. Meanwhile, starting from the initial search point at the end time, a forward search is performed along the time dimension, examining sleep state data after that point segment by segment. If a subsequent target sleep segment is found, and the time interval between the start time of the target sleep segment and the current initial search point is less than the aforementioned preset threshold, then the two sleep segments are determined to belong to the same complete sleep cycle. The initial search point is then updated to the end point of the subsequent target sleep segment, and the forward search continues based on this new starting point. If, after continuous forward search, no subsequent target sleep segment that meets the time interval condition is found, then the current search point is determined as the final sleep end time.

[0060] In one embodiment of this application, determining the initial search starting point for the start time of the target sleep interval and the initial search starting point for the end time based on the real-time sleep state reference sequence includes: If there is a target sleep segment in the real-time sleep state reference sequence that meets the preset duration condition, the start time node of the first target sleep segment is selected as the initial search starting point of the target sleep interval, and the end time node of the last target sleep segment is selected as the initial search starting point of the target sleep interval. If there is no target sleep segment in the real-time sleep state reference sequence that meets the preset duration condition, the start time node and end time node of the sleep segment with the longest sleep duration in the real-time sleep state reference sequence are selected as the initial search starting points for the start and end times of the target sleep interval, respectively.

[0061] Optionally, a comprehensive traversal and screening of the real-time sleep state reference sequence is performed to identify all target sleep segments in the sequence that meet the preset duration condition TH3 (e.g., duration not less than 3 hours). If a target sleep segment that meets the condition is found in the sequence after screening, the start time node of the first target sleep segment is directly selected as the initial search starting point for the target sleep interval, and the end time node of the last target sleep segment is selected as the initial search starting point for the end time of the target sleep interval. If, after traversal and screening, no target sleep segment that meets the preset duration condition is found in the real-time sleep state reference sequence, the screening strategy is adjusted, the duration of all sleep segments in the sequence is quantified and sorted, the sleep segment with the longest duration is selected, and the start time node and end time node of the longest sleep segment are used as the initial search starting points for the start and end times of the target sleep interval, respectively.

[0062] If there are no valid sleep segments in the sequence, the start and end times of the sequence are selected as the initial search starting points for the start and end times of the target sleep interval, respectively.

[0063] In one embodiment of this application, the sleep structure assessment results are obtained in the following manner: Based on the current sleep monitoring data, determine the duration of each current sleep stage and the percentage of each current sleep stage. Calculate the first relative difference between the duration of each current sleep stage and the corresponding reference value for the duration of the sleep stage, and the second relative difference between the proportion of each current sleep stage and the corresponding reference value for the proportion of the sleep stage. According to the preset sleep structure weighting coefficient and difference attenuation coefficient, the first relative difference and the second relative difference are weighted and fused to obtain the sleep structure consistency score, which is used as the sleep structure evaluation result.

[0064] Optionally, based on the target user's current sleep monitoring data, the real-time duration and real-time percentage of each sleep stage, such as deep sleep, light sleep, REM sleep, and wakefulness, are extracted. Then, for each sleep stage, the first relative difference between its current duration and the corresponding sleep stage duration reference value, and the second relative difference between its current percentage and the corresponding sleep stage percentage reference value are calculated. Combining a pre-set sleep structure weighting coefficient (used to distinguish the importance of different stages) and a difference decay coefficient (used to weaken the interference of excessive differences), the first relative difference and the second relative difference of each stage are weighted and fused. The final sleep structure consistency score is the sleep structure assessment result.

[0065] The sleep structure consistency score can be calculated using the following formula: ; Where l is the array index, and Preset weight parameters; For deep sleep duration, For light sleep duration, REM duration Duration of wakefulness; This is a reference value for deep sleep duration. This is a reference value for the duration of light sleep. This is a reference value for REM duration. This is a reference value for the duration of lucidity; For the percentage of deep sleep, The percentage of light sleep, REM percentage, The percentage of those who are sober; This is a reference value for the percentage of deep sleep. This is a reference value for the percentage of light sleep. This is a reference value for the REM ratio. This is a reference value for the percentage of people who are awake.

[0066] It should be noted that, and It can be obtained by fitting a specific dataset 2. The fitting process refers to using the duration of each sleep stage, the proportion of each sleep stage, the reference value of the duration of each sleep stage, and the reference value of the proportion of each sleep stage as input features on the last day within the window time TH1, and using the sleep structure consistency score as the target value to perform fitting or hyperparameter search.

[0067] Among them, specific dataset 2 refers to a sleep dataset in which user sleep data is continuously collected for a duration of TH1 or more (e.g., 30 days), during which there is no continuous invalid sleep data (the proportion of records with a single day sleep duration less than TH5 is less than a preset threshold), and which carries a sleep structure consistency score label. Invalid sleep data refers to the corresponding sleep data with a sleep duration less than TH5 (e.g., 3 hours). The sleep structure consistency score label refers to the sleep structure consistency score of each user's sleep data by multiple (e.g., 3) human sleep structure consistency score within a window of TH1. The score is calculated by averaging the sleep consistency score scale filled in by each user, the duration of each sleep stage on the last day within the window of TH1, the proportion of each sleep stage on the last day within the window of TH1, the reference value of the duration of each sleep stage on the last day within the window of TH1, and the reference value of the proportion of each sleep stage on the last day within the window of TH1, and the final sleep structure consistency score label is obtained by averaging the scores of multiple people.

[0068] In one embodiment of this application, the sleep quality reference data includes sleep efficiency reference values ​​and sleep continuity reference values, and the sleep quality assessment results are obtained in the following manner: Based on the current sleep monitoring data, determine the actual value of the current sleep efficiency and the actual value of the current sleep continuity; Based on the sleep quality reference data, obtain the corresponding sleep efficiency reference value and sleep continuity reference value; Calculate the third relative difference between the current actual value of sleep efficiency and the corresponding reference value of sleep efficiency, and the fourth relative difference between the current actual value of sleep continuity and the corresponding reference value of sleep continuity. The third and fourth relative differences are weighted and fused according to a preset sleep quality weighting coefficient to obtain a sleep quality consistency score, which is used as the sleep quality assessment result.

[0069] Optionally, based on the current sleep monitoring data, the actual values ​​of the target user's sleep efficiency and sleep continuity for the day are extracted; then, the corresponding reference values ​​of sleep efficiency and sleep continuity are retrieved from the sleep quality reference data; subsequently, the third relative difference between the current actual value of sleep efficiency and the reference value of sleep efficiency, and the fourth relative difference between the current actual value of sleep continuity and the reference value of sleep continuity are calculated respectively, in order to quantify the degree of deviation between the data of the day and the reference benchmark; finally, combined with a pre-set sleep quality weighting coefficient (used to distinguish the importance of sleep efficiency and continuity in the assessment), the two relative differences are weighted and fused, and the final sleep quality consistency score is the sleep quality assessment result.

[0070] The sleep quality consistency score can be calculated using the following formula: ; in, , Preset weight parameters; For sleep efficiency; Inherent sleep efficiency; For sleep continuity; This is due to inherent sleep continuity.

[0071] It should be noted that, , It can be obtained by fitting a specific dataset 3. The fitting process refers to using the sleep efficiency, sleep continuity, inherent sleep efficiency, and inherent sleep continuity of the last day within the window time TH1 as input features, and using the sleep quality consistency score as the target value based on the calculation form of the sleep quality consistency score, to perform fitting or hyperparameter search.

[0072] Specifically, Specific Dataset 3 refers to a sleep dataset whose continuous sleep data collection duration reaches or exceeds TH1, during which there is no continuous invalid sleep data (the proportion of records with a single-day sleep duration less than TH5 is less than a preset threshold), and which carries a sleep quality consistency score label. Invalid sleep data refers to the corresponding sleep data with a sleep duration less than TH5 (e.g., 3 hours). The sleep structure consistency score label refers to the sleep structure consistency score of multiple people (e.g., 3 people) on the sleep data within the window time TH11 (e.g., 30 days). The scoring basis is the sleep consistency score scale of the last day within the window time TH1, the sleep efficiency of the last day, the sleep continuity of the last day, the inherent sleep efficiency of the last day (i.e., the sleep efficiency reference value), and the inherent sleep continuity of the last day (i.e., the sleep continuity reference value). The average value of the multiple people's scores is calculated to obtain the final sleep quality consistency score label.

[0073] In this embodiment, by integrating three types of reference data—sleep state, sleep structure, and sleep quality—with current sleep monitoring data, the limitations of existing methods in single-dimensional assessment are overcome. Sleep conditions are depicted from multiple dimensions, including sleep patterns, stage characteristics, and core indicators, resulting in assessment results that more closely reflect the user's actual sleep state. By introducing a sleep confidence factor based on the proportion of effective sleep data within a preset time window, the impact of data validity on the assessment results is dynamically quantified, effectively avoiding assessment distortion caused by invalid or incomplete data, and improving the robustness and credibility of the assessment results. Personalized reference data is generated based on the target user's historical sleep data, abandoning existing general standard assessment models and adapting to the sleep habits and physiological characteristics of different users, resulting in more targeted assessments and a wider range of applications. Furthermore, no professional equipment or operation is required; it can be achieved based on routine sleep monitoring data. Simultaneously, multi-dimensional reference comparison and confidence optimization ensure assessment accuracy, balancing convenience and accuracy, making it suitable for widespread application in scenarios such as home health management and sleep disorder screening.

[0074] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0075] In one embodiment, a sleep assessment system is provided, which corresponds one-to-one with the sleep assessment methods described in the above embodiments. For example... Figure 7 As shown, the sleep assessment system includes a sleep reference data acquisition unit 10, a current sleep monitoring data acquisition unit 20, a multi-dimensional sleep assessment result determination unit 30, a sleep confidence factor determination unit 40, and a comprehensive sleep assessment result determination unit 50. Detailed descriptions of each functional module are as follows: The sleep reference data acquisition unit 10 is used to acquire sleep reference data of the target user, wherein the sleep reference data includes at least one of sleep state reference data, sleep structure reference data, and sleep quality reference data. The current sleep monitoring data acquisition unit 20 is used to acquire the current sleep monitoring data of the target user; The multidimensional sleep assessment result determination unit 30 is used to determine the multidimensional sleep assessment result based on the sleep reference data and the current sleep monitoring data. The multidimensional sleep assessment result includes at least one of the following: sleep state assessment result, sleep structure assessment result, and sleep quality assessment result. The sleep confidence factor determination unit 40 is used to determine the sleep confidence factor based on the proportion of valid sleep data that meets the preset duration threshold within a preset time window. The comprehensive sleep assessment result determination unit 50 is used to determine the comprehensive sleep assessment result based on the multi-dimensional sleep assessment results and the sleep confidence factor.

[0076] In one embodiment of this application, the sleep reference data acquisition unit 10 is further configured to: Obtain the target user's historical sleep data; Based on the historical sleep data, generate multiple sleep state arrays corresponding to preset time periods; At the same preset time point, the sleep state arrays are filtered; The sleep state array after filtering for the previous preset time period is truncated to obtain a sleep state reference sequence, which is used as the sleep state reference data. In one embodiment of this application, the sleep reference data acquisition unit 10 is further configured to: Based on the target user’s historical sleep data, determine the duration and percentage of each historical sleep stage. Preprocessing is performed on the historical sleep stage percentage and the historical sleep stage duration along the time dimension to obtain reference values ​​for sleep stage percentage and sleep process duration. Based on the aforementioned reference values ​​for the proportion of sleep stages and the reference values ​​for the duration of sleep processes, a reference value for the duration of sleep stages is determined. The reference values ​​for the percentage of sleep stages and the reference values ​​for the duration of sleep stages are used as reference data for the sleep structure.

[0077] In one embodiment of this application, the sleep state reference data includes a sleep state reference sequence. The system further includes a sleep state assessment result acquisition unit, used for: Based on the current sleep monitoring data, a current sleep state array is obtained; A segment with the same length as the current sleep state array is extracted from the sleep state reference sequence and used as the real-time sleep state reference sequence. Based on the current sleep state array and the real-time sleep state reference sequence, a paired state array is determined; Based on the real-time sleep state reference sequence, the weight distribution of the real-time sleep state is determined; Based on the paired state array and the real-time sleep state weight distribution, a sleep state consistency score is calculated and used as the sleep state evaluation result.

[0078] In one embodiment of this application, the sleep state assessment result acquisition unit is further used for Based on the real-time sleep state reference sequence, the sleep start time point and the sleep end time point are obtained; Based on the sleep start time and sleep end time, determine the sleep midpoint. The real-time sleep state weight distribution is calculated based on a preset weight distribution template, the sleep start time, the sleep end time, and the sleep midpoint.

[0079] In one embodiment of this application, the sleep state assessment result acquisition unit is further used for Based on the real-time sleep state reference sequence, the initial search starting point at the start time of the target sleep interval and the initial search starting point at the end time are determined. The sleep start point and sleep end point are determined by performing a reverse search along the time dimension based on the initial search starting point of the start time point and a forward search along the time dimension based on the initial search starting point of the end time point.

[0080] In one embodiment of this application, the sleep state assessment result acquisition unit is further used for Starting from the initial search point at the beginning time, the search proceeds backward along the time dimension. If a target sleep segment exists in the preceding sequence and the time interval between the end time of the target sleep segment and the initial search point is less than a preset threshold, the initial search point is updated to the starting point of the preceding segment and the search is repeated. If no preceding target sleep segment meets the conditions, the current search point is determined as the sleep start time. Starting from the initial search point at the end time, search forward along the time dimension. If a target sleep segment exists subsequently and the time interval between the start time of the target sleep segment and the initial search point is less than the preset threshold, then update the initial search point to the end point of the subsequent target sleep segment and repeat the search. If no subsequent target sleep segment meets the conditions, then determine the current search point as the end time of sleep.

[0081] In one embodiment of this application, the sleep state assessment result acquisition unit is further used for If there is a target sleep segment in the real-time sleep state reference sequence that meets the preset duration condition, the start time node of the first target sleep segment is selected as the initial search starting point of the target sleep interval, and the end time node of the last target sleep segment is selected as the initial search starting point of the target sleep interval. If there is no target sleep segment in the real-time sleep state reference sequence that meets the preset duration condition, the start time node and end time node of the sleep segment with the longest sleep duration in the real-time sleep state reference sequence are selected as the initial search starting points for the start and end times of the target sleep interval, respectively.

[0082] In one embodiment of this application, the system further includes a sleep structure assessment result acquisition unit, used for... Based on the current sleep monitoring data, determine the duration of each current sleep stage and the percentage of each current sleep stage. Calculate the first relative difference between the duration of each current sleep stage and the corresponding reference value for the duration of the sleep stage, and the second relative difference between the proportion of each current sleep stage and the corresponding reference value for the proportion of the sleep stage. According to the preset sleep structure weighting coefficient and difference attenuation coefficient, the first relative difference and the second relative difference are weighted and fused to obtain the sleep structure consistency score, which is used as the sleep structure evaluation result.

[0083] In one embodiment of this application, the sleep quality reference data includes sleep efficiency reference values ​​and sleep continuity reference values. The system further includes a sleep quality assessment result acquisition unit, used for: Based on the current sleep monitoring data, determine the actual value of the current sleep efficiency and the actual value of the current sleep continuity; Based on the sleep quality reference data, obtain the corresponding sleep efficiency reference value and sleep continuity reference value; Calculate the third relative difference between the current actual value of sleep efficiency and the corresponding reference value of sleep efficiency, and the fourth relative difference between the current actual value of sleep continuity and the corresponding reference value of sleep continuity. The third and fourth relative differences are weighted and fused according to a preset sleep quality weighting coefficient to obtain a sleep quality consistency score, which is used as the sleep quality assessment result.

[0084] In this embodiment, by integrating three types of reference data—sleep state, sleep structure, and sleep quality—with current sleep monitoring data, the limitations of existing methods in single-dimensional assessment are overcome. Sleep conditions are depicted from multiple dimensions, including sleep patterns, stage characteristics, and core indicators, resulting in assessment results that more closely reflect the user's actual sleep state. By introducing a sleep confidence factor based on the proportion of effective sleep data within a preset time window, the impact of data validity on the assessment results is dynamically quantified, effectively avoiding assessment distortion caused by invalid or incomplete data, and improving the robustness and credibility of the assessment results. Personalized reference data is generated based on the target user's historical sleep data, abandoning existing general standard assessment models and adapting to the sleep habits and physiological characteristics of different users, resulting in more targeted assessments and a wider range of applications. Furthermore, no professional equipment or operation is required; it can be achieved based on routine sleep monitoring data. Simultaneously, multi-dimensional reference comparison and confidence optimization ensure assessment accuracy, balancing convenience and accuracy, making it suitable for widespread application in scenarios such as home health management and sleep disorder screening.

[0085] For specific limitations regarding the sleep assessment system, please refer to the limitations on sleep assessment methods mentioned above, which will not be repeated here. Each module in the aforementioned sleep assessment system can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.

[0086] In one embodiment, a computer device is provided, which may be a terminal device, and its internal structure diagram may be as follows: Figure 8 As shown, the computer device includes a processor, memory, and a network interface connected via a system bus. The processor provides computing and control capabilities. The memory includes a readable storage medium storing computer-readable instructions. The network interface communicates with external terminals via a network connection. When executed by the processor, the computer-readable instructions implement a sleep evaluation method. The readable storage medium provided in this embodiment includes both non-volatile and volatile readable storage media.

[0087] In this application embodiment, a computer device is provided, including a memory, a processor, and computer-readable instructions stored in the memory and executable on the processor. When the processor executes the computer-readable instructions, it implements the steps of the sleep assessment method described above.

[0088] In this embodiment of the application, a readable storage medium is provided, which stores computer-readable instructions. When the computer-readable instructions are executed by a processor, they implement the steps of the sleep assessment method described above.

[0089] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing related hardware with computer-readable instructions. These computer-readable instructions can be stored in a non-volatile readable storage medium or a volatile readable storage medium. When executed, these computer-readable instructions can include the processes of the embodiments of the methods described above. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0090] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the above-described division of functional units and modules is used as an example. In practical applications, the above functions can be assigned to different functional units and modules as needed, that is, the internal structure of the device can be divided into different functional units or modules to complete all or part of the functions described above.

[0091] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A sleep assessment method characterized by, The method comprises: obtaining sleep reference data of a target user, the sleep reference data comprising at least one of sleep state reference data, sleep structure reference data and sleep quality reference data; obtaining current sleep monitoring data of the target user; based on the sleep reference data and the current sleep monitoring data, determining a multi-dimensional sleep evaluation result, the multi-dimensional sleep evaluation result comprising at least one of a sleep state evaluation result, a sleep structure evaluation result and a sleep quality evaluation result; based on a proportion of effective sleep data satisfying a preset time length threshold within a preset time window, determining a sleep confidence factor; based on the multi-dimensional sleep evaluation result and the sleep confidence factor, determining a comprehensive sleep evaluation result.

2. The sleep assessment method of claim 1, wherein, The sleep state reference data is obtained by the following method: obtaining historical sleep data of the target user; based on the historical sleep data, generating sleep state arrays corresponding to a plurality of preset time periods; filtering each sleep state array at the same preset time point; performing truncation processing on the sleep state array of the previous preset time period after filtering to obtain a sleep state reference sequence as the sleep state reference data; The sleep structure reference data is obtained by the following method: based on the historical sleep data of the target user, determining the historical sleep stage length and the historical sleep stage proportion corresponding to each historical sleep stage; preprocessing the historical sleep stage proportion and the historical sleep stage length along the time dimension respectively to obtain sleep stage proportion reference values and sleep process length reference values; based on the sleep stage proportion reference values and the sleep process length reference values, determining sleep stage length reference values; the sleep stage proportion reference values and the sleep stage length reference values are taken as the sleep structure reference data.

3. The sleep assessment method of claim 2, wherein, The sleep state reference data comprises a sleep state reference sequence, and the sleep state evaluation result is obtained by the following method: based on the current sleep monitoring data, obtaining a current sleep state array; from the sleep state reference sequence, a segment with the same length as the current sleep state array is intercepted as a real-time sleep state reference sequence; based on the current sleep state array and the real-time sleep state reference sequence, determining a paired state array; based on the real-time sleep state reference sequence, determining a real-time sleep state weight distribution; based on the paired state array and the real-time sleep state weight distribution, a sleep state consistency score is calculated as the sleep state evaluation result.

4. The sleep assessment method of claim 3, wherein, The real-time sleep state weight distribution is determined based on the real-time sleep state reference sequence, comprising: based on the real-time sleep state reference sequence, obtaining a sleep start time point and a sleep end time point; based on the sleep start time point and the sleep end time point, determining a sleep intermediate time point; based on a preset weight distribution template, a sleep start time point, a sleep end time point and a sleep intermediate time point, the real-time sleep state weight distribution is calculated.

5. The sleep assessment method of claim 4, wherein, The sleep start time point and the sleep end time point are obtained based on the real-time sleep state reference sequence, comprising: determine an initial search starting point of a target sleep interval start time point and an initial search starting point of an end time point based on the real-time sleep state reference sequence; determine a sleep start time point and a sleep end time point based on reverse search along a time dimension from the initial search starting point of the start time point and forward search along the time dimension from the initial search starting point of the end time point.

6. The sleep assessment method of claim 5, wherein, The determination of the sleep start time point and the sleep end time point based on the reverse search along the time dimension from the initial search starting point of the start time point and the forward search along the time dimension from the initial search starting point of the end time point includes: reverse search along the time dimension from the initial search starting point of the start time point, if there is a target sleep segment in the previous sequence and a time interval between an end time point of the target sleep segment and the initial search starting point is less than a preset threshold, update the initial search starting point to a starting point of the previous segment and repeat the search, if there is no target sleep segment in the previous sequence, determine the current search starting point as the sleep start time point; forward search along the time dimension from the initial search starting point of the end time point, if there is a target sleep segment in the subsequent sequence and a time interval between a start time point of the target sleep segment and the initial search starting point is less than the preset threshold, update the initial search starting point to an end point of the subsequent target sleep segment and repeat the search, if there is no target sleep segment in the subsequent sequence, determine the current search starting point as the sleep end time point.

7. The sleep assessment method of claim 5, wherein, The determination of the initial search starting point of the target sleep interval start time point and the initial search starting point of the end time point based on the real-time sleep state reference sequence includes: if there is a target sleep segment in the real-time sleep state reference sequence that meets a preset time length condition, select a starting time node of a first target sleep segment as the initial search starting point of the target sleep interval start time point and select an ending time node of a last target sleep segment as the initial search starting point of the target sleep interval end time point; if there is no target sleep segment in the real-time sleep state reference sequence that meets the preset time length condition, select a starting time node and an ending time node of a sleep segment with the longest sleep duration in the real-time sleep state reference sequence as the initial search starting points of the target sleep interval start time point and the target sleep interval end time point respectively.

8. The sleep assessment method of claim 2, wherein, The sleep structure evaluation result is obtained by the following method: determine a current sleep stage duration and a current sleep stage proportion corresponding to each current sleep stage based on the current sleep monitoring data; calculate a first relative difference between each current sleep stage duration and a corresponding sleep stage duration reference value and a second relative difference between each current sleep stage proportion and a corresponding sleep stage proportion reference value respectively; weight and fuse the first relative difference and the second relative difference according to a preset sleep structure weighting coefficient and a difference attenuation coefficient to obtain a sleep structure consistency score as the sleep structure evaluation result.

9. The sleep assessment method of claim 1, wherein, The sleep quality reference data includes a sleep efficiency reference value and a sleep continuity reference value, and the sleep quality evaluation result is obtained by the following method: determine a current sleep efficiency actual value and a current sleep continuity actual value based on the current sleep monitoring data; obtain a corresponding sleep efficiency reference value and a sleep continuity reference value based on the sleep quality reference data; calculate a third relative difference between the current sleep efficiency actual value and the corresponding sleep efficiency reference value, and a fourth relative difference between the current sleep continuity actual value and the corresponding sleep continuity reference value; weight and fuse the third relative difference and the fourth relative difference according to a preset sleep quality weighting coefficient to obtain a sleep quality consistency score as the sleep quality evaluation result.

10. A sleep assessment system, characterized by The system comprises: a sleep reference data acquisition unit configured to acquire sleep reference data of a target user, the sleep reference data comprising at least one of sleep state reference data, sleep structure reference data and sleep quality reference data; a current sleep monitoring data acquisition unit configured to acquire current sleep monitoring data of the target user; a multi-dimensional sleep evaluation result determination unit configured to determine multi-dimensional sleep evaluation results based on the sleep reference data and the current sleep monitoring data, the multi-dimensional sleep evaluation results comprising at least one of sleep state evaluation results, sleep structure evaluation results and sleep quality evaluation results; a sleep confidence factor determination unit configured to determine a sleep confidence factor based on a proportion of effective sleep data satisfying a preset time length threshold within a preset time window; a comprehensive sleep evaluation result determination unit configured to determine a comprehensive sleep evaluation result based on the multi-dimensional sleep evaluation results and the sleep confidence factor.

Images