Sleep quality improving system based on big data analysis

Abstract

The invention discloses a sleep quality improving system based on big data analysis, and relates to the technical field of big data. According to the system, a multi-dimensional sleep state monitoring matrix is constructed, effective intervals of monitoring points are optimized in combination with dynamic adjustment factors and a deep learning algorithm, and redundant data interference is eliminated; generating a sleep intervention regulation factor based on the environmental influence function and the physiological rhythm constraint factor, and training the sleep intervention function through a genetic algorithm to dynamically adjust the intervention intensity; and generating an intervention instruction by utilizing the reinforcement learning model and combining the real-time physiological data of the user, the environmental parameters and the sleep stage prediction result, and controlling the starting sequence and parameter adjustment of the monitoring points until preset ideal sleep state parameters are met. According to the method, the problems of insufficient monitoring precision, poor environment adaptability, intervention lag and the like in a traditional method are solved, the accuracy and the real-time performance of sleep intervention are improved, and the method is suitable for the fields of smart home, medical health and the like.

Description

Technical Field

[0001] This invention belongs to the technical field, specifically relating to a sleep quality improvement system based on big data analysis. Background Technology

[0002] With the fast pace of modern life and increasing work pressure, sleep quality issues have become a significant factor affecting human health. Traditional sleep monitoring and intervention methods mainly rely on single sensors or fixed threshold rules, which have the following limitations: Insufficient monitoring accuracy: Traditional systems often use a single monitoring device (such as a pressure sensor or heart rate belt) or a fixed monitoring range, which makes it difficult to fully cover the multi-dimensional characteristics of the user's sleep state (such as body movement frequency, breathing pattern, environmental interference, etc.), resulting in incomplete data collection or misjudgment.

[0003] Weak redundancy data processing capability: In scenarios with multiple monitoring points, the monitoring ranges of adjacent devices are prone to data overlap (such as multiple sensors simultaneously collecting body motion data in the same time period). However, existing technologies lack quantitative analysis and dynamic optimization mechanisms for redundant intervals, leading to wasted computing resources and ineffective intervention.

[0004] Environmental regulation lacks personalized adaptation: Traditional intervention measures (such as temperature and humidity regulation) are mostly based on static thresholds or subjective user feedback, without combining real-time physiological rhythms (such as sleep cycles and number of awakenings) and environmental perception data (such as noise and light), making it difficult to accurately match user needs.

[0005] Intervention strategies are slow to respond: Existing systems mostly use preset rules to trigger interventions (such as starting white noise at fixed times), which cannot dynamically adjust the intensity and timing of interventions according to the user's real-time sleep status (such as sudden awakening or a decrease in the proportion of deep sleep), resulting in limited intervention effects.

[0006] To address the above problems, this invention proposes a sleep quality improvement system based on big data analysis. Summary of the Invention

[0007] To overcome the shortcomings and deficiencies of the existing technology, the present invention adopts the following technical solution: A sleep quality improvement system based on big data analysis includes: Sleep state monitoring matrix construction module: used to construct a sleep state monitoring matrix based on the distance between each monitoring point and the target user, the monitoring direction of the monitoring point to the target user, and the monitoring interval within the preset sleep monitoring area; Sleep intervention regulation factor acquisition module: used to obtain sleep intervention regulation factors based on real-time environmental parameters of the sleep monitoring area and a preset environmental influence function; Personalized sleep improvement plan generation module: Based on the sleep state monitoring matrix and the configured personalized sleep improvement model, it generates personalized intervention instructions, controls the start order and execution time of intervention measures for each monitoring point corresponding to the target user, and adjusts the intervention parameters based on the sleep intervention adjustment factor until the preset ideal sleep state parameters are met.

[0008] Preferably, the system workflow includes: Step 1: Construct a sleep state monitoring matrix based on the distance between each monitoring point and the target user, the monitoring direction corresponding to the target user, and the monitoring interval within the preset sleep monitoring area. Preferably, the construction process of the sleep state monitoring matrix includes: constructing a dynamic monitoring model for each monitoring point and the original dynamic interval of the monitoring data in the time dimension based on the monitoring parameters of the monitoring points; obtaining the information redundancy interval of adjacent monitoring points through sleep cycle analysis and deep learning algorithms to form an information redundancy interval set; correcting the time dimension interval of the monitoring points by combining dynamic adjustment factors and data augmentation algorithms to obtain an effective monitoring interval set; constructing a sleep stage transition prediction dataset based on the real-time sleep state, physiological indicator sequence, expected sleep stage time, and monitoring point system delay of the target user within the effective monitoring interval set; inputting the compensated sleep stage transition function constructed by a recurrent neural network to obtain the compensated sleep stage transition trajectory and real-time sleep state points, thereby determining the intervention time point set; constructing a sleep state monitoring matrix based on the intervention time point set using a 0-1 matrix, where the matrix element values ​​are jointly determined by the monitoring intervention factor, the duration of continuous sleep, the time length to enter the lower limit of the interval, and the intervention delay time length. Step 2: Based on the real-time environmental parameters of the sleep monitoring area and the preset environmental influence function, obtain the sleep intervention regulation factor; preferably, the environmental influence function is associated with temperature, humidity, noise and the corresponding optimal parameters, tolerance deviation range and weight coefficients, and the noise term uses a logarithmic function to reflect the nonlinear perception of the human ear; the process of obtaining the sleep intervention regulation factor includes establishing a physiological-time-sleep quality database, calculating physiological rhythm constraint factors, filtering intervention parameters, obtaining the deviation of required sleep indicators, correcting the indicator deviation, adjusting the intervention intensity and training the sleep intervention function through a genetic algorithm; Step 3: Based on the sleep state monitoring matrix and the configured personalized sleep improvement model, generate personalized intervention instructions to control the activation sequence and execution time of intervention measures for each monitoring point corresponding to the target user. While the intervention measures are activated, adjust the parameters corresponding to each activated intervention measure based on the sleep intervention regulation factor until the preset ideal sleep state parameters obtained based on the polysomnography standard and human physiological rhythm in the sleep environment are met, thereby obtaining a personalized sleep improvement plan that meets the needs of the target user.

[0009] Preferably, in the process of constructing the sleep state monitoring matrix in step one, the dynamic monitoring model of the monitoring points is constructed based on the sleep region range, monitoring point distribution, number of monitoring points, and individual monitoring parameters extracted by the sleep monitoring system. This is combined with the monitoring accuracy, monitoring range, data acquisition frequency, maximum effective monitoring distance, and signal attenuation coefficient corresponding to the monitoring parameters of each monitoring point, and is achieved through a multi-dimensional sleep state model combined with an environmental influence function. Preferably, the process of obtaining the effective monitoring interval set is as follows: The initial effective monitoring interval of the second monitoring point is obtained by subtracting the product of the interval length of the information redundancy intervals corresponding to the first and second monitoring points in the same time dimension and the dynamic adjustment factor from the time dimension interval of the monitoring data corresponding to the second monitoring point. This initial effective monitoring interval is then input into a data augmentation algorithm to correct the data quality, resulting in the corrected effective monitoring interval. The above process is repeated to obtain the corrected effective monitoring intervals of all monitoring points in the same time series, forming the effective monitoring interval set.

[0010] Preferably, the signal attenuation coefficient is obtained through simulation experiments by monitoring the characteristic relationship between the power of the monitoring device, the signal strength and the distance; the characteristic relationship is that the signal strength decreases logarithmically with increasing distance, and the rate of attenuation is determined by the signal attenuation coefficient, which reflects the influence of the environment on signal attenuation, and the value of the signal attenuation coefficient is different under different environments.

[0011] Preferably, the acquisition of information redundancy intervals includes the following steps: a. Based on the monitoring accuracy, monitoring range, data acquisition frequency, maximum effective monitoring distance, and signal attenuation coefficient of each monitoring point, construct dynamic monitoring models for two adjacent monitoring points respectively, and define their respective monitoring intervals in the time series; b. Determine the maximum monitoring interval range for each monitoring point based on hardware performance or algorithm presets, which is the theoretically coverable time range of the monitoring point; c. Divide sleep time according to physiological sleep cycles, analyze the change pattern of target user's sleep state in each cycle, align the monitoring data of adjacent monitoring points by cycle, identify the temporal correlation of monitoring data in the same cycle, and determine the overlapping part of monitoring data in the time dimension; d. Based on the aligned time series data, apply deep learning algorithms to extract the temporal features, spatial features, and state features of the monitoring data, identify the overlapping intervals of adjacent monitoring points in the time dimension, and mark the overlapping interval as an information redundancy interval when the length of the overlapping interval exceeds a preset threshold and the data correlation is lower than the threshold; e. Repeat the above steps a and b for all adjacent monitoring point pairs, quantify the redundant time intervals between each pair of adjacent monitoring points, and collect all redundant intervals to obtain the information redundancy interval set.

[0012] Preferably, the dynamic adjustment factor is obtained through the following process: Step 1: Based on sleep medicine research and a large amount of user data, preset the baseline turning frequency threshold corresponding to different sleep stages; Step 2: Collect the target user's body movement data through monitoring points at fixed time windows, count the number of real-time turnings per unit time, and calculate the real-time turning frequency; Step 3: Combine the sleep stages predicted in real time in the sleep state monitoring matrix, and select the baseline turning frequency threshold for the corresponding stage; Step 4: Quantify the deviation between the real-time turning frequency and the baseline turning frequency through a formula to obtain the frequency deviation rate; Step 5: Based on the frequency deviation rate, obtain the dynamic adjustment factor through linear or nonlinear function mapping. When the frequency deviation rate is small, the adjustment factor takes a smaller value; when the frequency deviation rate is large, the adjustment factor takes a larger value, and the adjustment factor is updated in real time with changes in sleep stages.

[0013] Preferably, the process of establishing the physiological-time-sleep quality database in the process of obtaining sleep intervention regulation factors in step two is as follows: sleep time is divided into 90-minute sleep cycles, and the range of changes in the physiological indicators of the target user and the corresponding sleep quality impact data within each sleep cycle are collected and organized, and stored according to user ID, sleep cycle, and physiological indicator type; preferably, the calculation process of the physiological rhythm constraint factor is as follows: the correlation between each physiological indicator and the sleep quality indicator is quantified through a machine learning model, an influence weight is assigned to each physiological indicator based on the correlation coefficient, the average value or range of change of each physiological indicator of the target user within each sleep cycle is extracted, the quantified value of each physiological indicator is multiplied by its influence weight and then summed, and the physiological rhythm constraint factor of each sleep cycle is obtained after normalization.

[0014] Preferably, the process of obtaining sleep intervention regulation factors in step two further includes: filtering intervention parameters: using physiological rhythm constraint factors, filtering intervention parameters according to target sleep state parameters, including environmental control parameters and equipment intervention parameters, filtering out intervention parameters that do not conform to the user's physiological rhythm; obtaining the deviation of demand sleep indicators: based on the polysomnography standard and the actual sleep indicators corresponding to the filtered intervention parameters at real-time monitoring points, obtaining absolute and relative deviations, and integrating them into the deviation of demand sleep indicators; correcting the indicator deviation: based on the deviation of demand sleep indicators, correcting the indicator deviation through an adaptive adjustment function, using linear proportional adjustment for single deviation types and weighted nonlinear model adjustment for multiple deviation types; adjusting the intervention intensity: based on the intervention parameters after indicator correction, constructing a sleep adaptation regulation factor by combining sleep cycle, sleep depth, number of awakenings, and temperature and humidity obtained from the environmental simulation model, and adjusting the intervention intensity; training the sleep intervention function: using a genetic algorithm, combining the constraints constructed by real-time physiological parameters and demand physiological parameters at monitoring points, and the sleep intervention function fitted by physiological rhythm constraint factors, sleep adaptation regulation factors, and indicator deviations, training to obtain the trained sleep intervention function and its corresponding weight parameters, thereby obtaining the sleep intervention regulation factors.

[0015] Preferably, the construction process of the personalized sleep improvement model includes: constructing a personalized sleep improvement model based on reinforcement learning, combining a sleep intervention function, a sleep state monitoring matrix, and preset intervention trigger conditions; training the model using sleep data, physiological index data, environmental parameter data, and lifestyle habit data of target users at different time periods; embedding the trained model into the corresponding sleep monitoring system to achieve real-time sleep intervention control; collecting feedback information from target users through a sleep quality assessment scale, comprehensively evaluating to obtain an intervention effect score, constructing a reinforcement reward function, and feeding it back to the personalized sleep improvement model to achieve continuous model training; preferably, the preset intervention trigger conditions include: when the target user is awake and it is not yet the preset sleep time, all monitoring points are in the off state. In the closed state, when the target user has reached the preset sleep time but has not entered the starting point of the effective monitoring interval, all monitoring points are in standby mode. When all monitoring points are in standby mode and the target user enters the starting point of the effective monitoring interval, the corresponding monitoring point is activated through the sleep state monitoring matrix, and the intervention parameters of the activated monitoring point are synchronously adjusted through the sleep intervention function based on the real-time collected physiological parameters, environmental parameters, and sleep stage parameters. When the target user reaches the state point corresponding to the lower limit of the effective monitoring interval, the corresponding monitoring point is converted to standby mode through the sleep state monitoring matrix. If it has not reached the lower limit, the current monitoring point remains in operation, and the intervention parameters corresponding to the current monitoring point are repeatedly adjusted based on the target user and monitoring point parameters within the effective monitoring interval.

[0016] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention improves the real-time performance and accuracy of monitoring by dynamically identifying and correcting information redundancy intervals between adjacent monitoring points, combined with the user's real-time physiological state (dynamically adjusting the monitoring interval, avoiding invalid data calculation, optimizing the allocation of monitoring resources, and ensuring that only accurate sleep state data within the effective time period is retained).

[0017] 2. This invention generates a physiological rhythm constraint factor based on changes in physiological indicators during the user's sleep cycle, and combines it with an environmental influence function to dynamically adapt intervention parameters, avoiding conflicts between intervention measures and the user's natural physiological rhythm, thereby enhancing comfort and improving intervention effectiveness.

[0018] 3. This invention constructs a personalized sleep improvement model through reinforcement learning, combines real-time physiological data of users, environmental parameters and PSG standard feedback, continuously optimizes intervention strategies, and incorporates subjective feedback from users to form a closed loop of monitoring-intervention-evaluation-optimization, thereby achieving a precise sleep improvement plan that adapts to individual differences in the long term. Attached Figure Description

[0019] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0020] Figure 1 A block diagram of a sleep quality improvement system based on big data analysis according to the present invention is shown; Figure 2 A flowchart of the system operation of the present invention is shown; Figure 3 A flowchart illustrating the process of obtaining the information redundancy interval of the present invention is shown. Detailed Implementation

[0021] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0022] Furthermore, the described features, structures, or characteristics can be combined in any suitable manner in one or more exemplary embodiments. Numerous specific details are provided in the following description to give a full understanding of exemplary embodiments of this disclosure. However, those skilled in the art will recognize that the technical solutions of this disclosure can be practiced with one or more of the specific details omitted, or other methods, components, steps, etc., can be employed. In other instances, well-known structures, methods, implementations, or operations are not shown or described in detail to avoid obscuring various aspects of this disclosure.

[0023] Example 1: See Figure 1 As shown in this embodiment, a sleep quality improvement system based on big data analysis includes: Sleep state monitoring matrix construction module: used to construct a sleep state monitoring matrix based on the distance between each monitoring point and the target user, the monitoring direction of the monitoring point to the target user, and the monitoring interval within the preset sleep monitoring area; Sleep intervention regulation factor acquisition module: used to obtain sleep intervention regulation factors based on real-time environmental parameters of the sleep monitoring area and a preset environmental influence function; Personalized sleep improvement plan generation module: It is used to generate personalized intervention instructions based on the sleep state monitoring matrix and the configured personalized sleep improvement model, control the start order and execution time of intervention measures for each monitoring point corresponding to the target user, and adjust the intervention measure parameters based on the sleep intervention adjustment factor until the preset ideal sleep state parameters are met.

[0024] The beneficial effects of this embodiment are as follows: the system optimizes data acquisition efficiency through a dynamic monitoring matrix, precisely adjusts intervention parameters by combining environmental and physiological rhythms, and achieves personalized adaptation through reinforcement learning. Its beneficial effects include: improving the real-time performance and accuracy of sleep state monitoring and reducing redundant data interference; dynamically optimizing intervention strategies through a closed-loop feedback mechanism to enhance user comfort; and customizing intervention plans based on individual physiological characteristics and sleep cycles to improve sleep quality and meet diverse needs.

[0025] Example 2: See Figure 2 As shown in this embodiment, a sleep quality improvement system based on big data analysis has the following workflow: Step 1: Construct a sleep state monitoring matrix based on the distance between each monitoring point and the target user, the monitoring direction of the monitoring point to the target user, and the monitoring interval within the preset sleep monitoring area. The steps for constructing the sleep state monitoring matrix include: extracting the range of the sleep area, the distribution of monitoring points within the sleep area, the number of monitoring points, and the monitoring parameters of a single monitoring point from the sleep monitoring system; based on the monitoring accuracy, monitoring range and data acquisition frequency, maximum effective monitoring distance, and signal attenuation coefficient corresponding to the monitoring parameters of each monitoring point, modeling the monitoring point data through a multi-dimensional sleep state model combined with an environmental influence function, and obtaining the dynamic monitoring model corresponding to each monitoring point and the original dynamic range of the corresponding monitoring data in the time dimension.

[0026] The signal attenuation coefficient was obtained through simulation experiments by monitoring the characteristic relationship between device power, signal strength, and distance. The characteristic relationship is that signal strength attenuates logarithmically with increasing distance, and the rate of attenuation is determined by the signal attenuation coefficient. Decision. The specific mathematical expression is: Where S(d) is the signal strength at a distance of d, representing the signal strength at the distance monitoring device. At the location of the monitoring point, the actual signal strength received is a key indicator for measuring the remaining energy after signal propagation, used to determine whether the monitoring point can effectively collect data related to the sleep state of the target user; Pt is the transmission power of the monitoring equipment; The distance between the monitoring point and the target user; L(d) is the reference distance, a pre-defined standard distance used as a reference for signal attenuation calculation; L(d) is the distance. Path loss (signal attenuation) at the location. For reference distance; for Path loss at the location (can be directly measured experimentally); The path loss exponent (i.e., signal attenuation coefficient) reflects the impact of the environment on signal attenuation (such as in free space). Complex indoor environment ).

[0027] Based on the dynamic monitoring model corresponding to two adjacent monitoring points in the same time series and the maximum monitoring interval range corresponding to the two monitoring points, information redundancy intervals corresponding to two adjacent monitoring points in the same time series are obtained through sleep cycle analysis and deep learning algorithms. The above process is repeated to obtain the information redundancy intervals corresponding to all adjacent monitoring points, resulting in a set of information redundancy intervals. The effective monitoring interval of the second monitoring point in the time dimension is obtained by subtracting the product of the interval length of the information redundancy intervals corresponding to the first and second monitoring points in the same time dimension and the dynamic adjustment factor from the interval of the monitoring data corresponding to the second monitoring point in the time dimension. The effective monitoring interval of the second monitoring point in the time dimension is input into the data augmentation algorithm to correct the data quality of the subtracted intervals, resulting in the corrected effective monitoring interval of the second monitoring point in the time dimension. The effective monitoring interval refers to the effective time period actually used to monitor the sleep state of the target user after the redundancy interval correction and dynamic adjustment.

[0028] The information redundancy interval refers to the overlapping portion of monitoring data from adjacent monitoring points within the same time series. Because multiple monitoring points (such as pressure sensors and radar) are deployed within a sleep monitoring area, the monitoring ranges of adjacent points may overlap, leading to duplicate data in the monitoring of the target user's sleep state within the same time period. The purpose of obtaining the information redundancy interval is to optimize the efficiency of monitoring data utilization and avoid ineffective calculations by quantifying this overlap.

[0029] See Figure 3 As shown, obtaining the information redundancy interval includes the following steps: a. Construct a dynamic monitoring model for adjacent monitoring points.

[0030] Based on the parameters of each monitoring point (including but not limited to monitoring accuracy, monitoring range, data acquisition frequency, maximum effective monitoring distance, and signal attenuation coefficient), a dynamic monitoring model is constructed for each monitoring point. This model describes the original dynamic interval of the monitoring point in the time dimension (i.e., the effective time range of the monitoring data). For two adjacent monitoring points (monitoring point A and monitoring point B), their dynamic monitoring models define their respective monitoring intervals in the time series. The interval for monitoring point A is... arrive , The interval of monitoring point B is arrive , .

[0031] b. Determine the maximum monitoring interval range.

[0032] The maximum monitoring range of each monitoring point is limited by its hardware performance or algorithm presets. For example: physical limitations: the maximum effective monitoring distance of the monitoring device (beyond this distance, signal attenuation prevents effective data collection); algorithm limitations: a preset monitoring time range based on signal attenuation coefficient and monitoring accuracy (e.g., activating only within 30 minutes before the target user enters the mattress to 10 minutes after waking up). The maximum monitoring range of two adjacent monitoring points defines the theoretically coverable time range, providing boundaries for subsequent analysis of overlapping areas.

[0033] c. Sleep cycle analysis and time series alignment.

[0034] Sleep time is divided into physiological cycles (e.g., 90-minute sleep cycles), and the patterns of sleep state changes in the target user within each cycle (e.g., alternation between light sleep, deep sleep, and wakefulness) are analyzed. Through sleep cycle analysis, monitoring data from adjacent monitoring points are aligned by cycle to identify the temporal correlation of monitoring data within the same cycle. For example, within a certain sleep cycle, monitoring points A and B may both cover the transition stage from light sleep to deep sleep for the target user, and their monitoring data may overlap in the time dimension.

[0035] The overlapping part is ,in, It represents the overlapping portion of monitoring data from two adjacent monitoring points (monitoring point A and monitoring point B) in the time dimension. It is a time interval used to quantify the overlap of the monitoring ranges of adjacent monitoring points. It is the starting time point of the overlapping time interval; it is the starting time of the monitoring interval for monitoring point A. The start time of the monitoring interval at monitoring point B The larger value in, i.e. ; It is the end time of the monitoring interval for monitoring point A. and the end time of the monitoring interval at monitoring point B The smaller value in, i.e. ; d. Deep learning algorithms identify redundant patterns.

[0036] Based on aligned time-series data, deep learning algorithms (such as Convolutional Neural Networks (CNN), Recurrent Neural Networks (RNN), or Long Short-Term Memory Networks (LSTM)) are applied to extract features and recognize patterns from monitoring data at adjacent monitoring points. The core tasks of the algorithm are: Feature Extraction: Extracting temporal features (such as data collection time and duration), spatial features (such as distance and direction between the monitoring point and the target user), and state features (such as sleep stage and physiological indicators) from the monitoring data; Pattern Recognition: By comparing the data features of adjacent monitoring points, identifying the overlapping intervals in the time dimension (i.e., within the same time period, both monitoring points collected sleep state data from the target user, and the data content is highly correlated or redundant). For example, the deep learning model can learn, through training, that when the target user is in a light sleep stage and the frequency of turning over is below a threshold, monitoring points A and B overlap in the time interval... The data collected internally is highly consistent, thus enabling labeling. This is a redundant interval.

[0037] like length If the data exceeds a preset threshold, and the data correlation (such as differences in sleep stages or physiological indicators) is below the threshold, then... It is marked as an information redundancy zone.

[0038] e. Quantify the redundant intervals and aggregate them into an information redundant interval set.

[0039] Repeat the above process for all adjacent monitoring point pairs (e.g., AB, BC, CD, etc.), quantifying the redundant time interval between each pair of adjacent monitoring points (e.g., the redundant interval between monitoring points A and B is...). The redundancy interval between monitoring points B and C is (etc.). Finally, the redundant intervals of all adjacent monitoring point pairs are collected to obtain the information redundancy interval set.

[0040] The dynamic adjustment factor is constructed using the target user's real-time turning frequency and the baseline turning frequency; the process of obtaining the effective monitoring interval of the second monitoring point in the time dimension is repeated to obtain the corrected set of effective monitoring intervals in the same time series; The specific process for obtaining the dynamic adjustment factor is as follows: Step 1: Set the baseline rolling frequency.

[0041] Based on sleep medicine research and extensive user data, baseline turning frequency thresholds are predefined for different sleep stages (such as light sleep, deep sleep, and REM sleep): Light sleep stage: The baseline turning frequency is relatively high (e.g., 2-3 times per minute), because users are prone to turning over frequently due to external disturbances or their own discomfort during this stage; Deep sleep: The baseline turning frequency is low (e.g., 0.5-1 times per minute) because the user is in a deep resting state and has less body movement; REM phase: The baseline rolling frequency is moderate (e.g., 1-2 times per minute) because the muscles are relaxed but the brain is active.

[0042] Step 2: Collect the target user's turning frequency in real time.

[0043] By using monitoring points (such as pressure sensors and radar) within the sleep monitoring area, the body movement data of the target user is collected at fixed time windows (such as every 10 seconds), and the number of real-time turning over per unit time (such as per minute) is counted to calculate the real-time turning over frequency (unit: times / minute).

[0044] Step 3: Match the current sleep stage with the baseline frequency.

[0045] By combining the real-time predicted sleep stages in the sleep state monitoring matrix (such as sleep stages compensated by a recurrent neural network), a baseline turning frequency threshold is selected for the corresponding stage. For example, if the current prediction is light sleep, the baseline frequency is 2 times / minute; if the current prediction is deep sleep, the baseline frequency is 0.5 times / minute.

[0046] Step 4: Calculate the difference between the real-time and reference frequencies.

[0047] The deviation between the real-time frequency and the reference frequency is quantified using a formula: ;in, This refers to the real-time turning frequency; This is the baseline frequency corresponding to the current sleep stage; This is the frequency deviation rate (range 0-1).

[0048] Step 5: Map to a dynamic adjustment factor.

[0049] According to frequency deviation rate It is mapped to a dynamic adjustment factor a (range 0-1) through a linear or nonlinear function: like Smaller (i.e.) This indicates that the user's state is stable, and a smaller value for 'a' (such as 0.2-0.5) indicates a smaller correction range. like Larger (i.e.) This indicates that the user is in an active state (such as light sleep or wakefulness). A larger value for 'a' (such as 0.8-1.0) indicates a larger correction range.

[0050] As sleep stages change (such as transitioning from light sleep to deep sleep), the baseline frequency threshold is updated in real time and recalculated. And 'a', to ensure that the adjustment factor always adapts to the user's current state.

[0051] The effective monitoring interval is defined as: based on the original dynamic interval of the monitoring point, after eliminating the information redundancy interval of adjacent monitoring points (identified by sleep cycle analysis and deep learning algorithms), and after dynamic adjustment and data enhancement correction based on the user's real-time status (such as the frequency of turning over), the effective time period is actually used to accurately monitor the sleep status of the target user.

[0052] The system acquires the target user's real-time sleep state, physiological index sequence, estimated time of the next sleep stage, and system delay corresponding to each monitoring point within the effective monitoring interval set, constructing a target user sleep stage transition prediction dataset. This dataset is then input into a compensating sleep stage transition function constructed using a recurrent neural network. Based on the actual sleep state changes of the target user and the state deviation corresponding to the monitoring point delay, the system obtains the compensated sleep stage transition trajectory and real-time sleep state points corresponding to the target user. Based on the compensated sleep stage transition trajectory and real-time sleep state points corresponding to the target user, and the corrected effective monitoring interval set on the same time series, the system obtains the set of intervention time points corresponding to the target user entering the lower limit of each corrected effective monitoring interval.

[0053] Based on the set of intervention time points, a sleep state monitoring matrix M is constructed using a 0-1 matrix. ;in, This represents the state of the j-th monitoring point at time t (0 indicates not started, 1 indicates started).

[0054] Each element The value is determined by the following parameters: monitoring intervention factors Continuous sleep duration Length of time to enter the lower limit of the interval Intervention delay time length .

[0055] Formula for calculating elements:

[0056] That is, when a monitoring point participates in intervention, and time t falls within the effective intervention period after the user enters the interval, the element is 1; otherwise, it is 0. To monitor intervention factors; This refers to the duration of continuous sleep. The time length for entering the lower limit of the interval; The length of the intervention delay time; Monitoring intervention factors This indicates whether the j-th monitoring point participates in the current intervention (1 indicates participation, 0 indicates non-participation). If a monitoring point is determined to require intervention (e.g., a user enters its effective monitoring range), then... ;otherwise, .

[0057] Continuous sleep duration This represents the duration (in units of time points) of continuous sleep of the target user within the corrected effective monitoring interval at the j-th monitoring point.

[0058] Length of time to reach the lower limit of the interval This represents the time point (unit: time point) at which the target user enters the lower limit (starting point) of the effective monitoring interval after the correction at the j-th monitoring point.

[0059] Intervention delay time length This represents the delay time (in time points) from the receipt of the intervention instruction to the actual execution of the intervention at the j-th monitoring point.

[0060] Step 2: Obtain sleep intervention regulation factors based on real-time environmental parameters of the sleep monitoring area and a preset environmental influence function.

[0061] Environmental impact function: ;in, For temperature; Humidity; For noise; It is the real-time humidity of the sleep monitoring area at time t; It is the real-time temperature of the sleep monitoring area at time t; The optimal temperature; To achieve the optimal humidity; , Temperature tolerance deviation range, humidity tolerance deviation range (e.g., ±3℃, ±15%RH); noise item A logarithmic function is used to reflect the nonlinear perception of noise by the human ear (the subjective loudness doubles for every 10 dB increase in decibels). , , These are the weighting coefficients.

[0062] The process of obtaining sleep intervention regulatory factors includes: S21. Divide sleep time into 90-minute sleep cycles, collect and organize the range of changes in all relevant physiological indicators of the target user and the corresponding impact data on sleep quality within each sleep cycle based on human physiological rhythms, classify and store the collected range of changes in physiological indicators, and establish a physiological-time-sleep quality database.

[0063] Relevant physiological indicators include, but are not limited to, body movement / turning frequency, heart rate, respiratory rate, and blood oxygen saturation.

[0064] Range of changes in body movement / rolling frequency: Light sleep stage: 2-5 times / minute (frequent turning over due to external disturbances or discomfort) Deep sleep stage: 0-1 times / minute (minimal body movement, muscle relaxation) Rapid eye movement (REM) phase: 1-3 times / minute (muscles are relaxed but the brain is active; there may be slight body movements). Heart rate variation range: Arousal level: 60-100 bpm (large individual differences, such as athletes may have lower levels). Light sleep: 5-15% lower than when awake (e.g., 70 → 60 bpm) Deep sleep: 15-30% lower than when awake (e.g., 70 → 50 bpm) REM sleep: close to light sleep level (60-75 bpm) Range of respiratory rate variation: Awake state: 12-20 times / minute (at rest) Sleep period: 10-18 breaths / minute (breathing is even during deep sleep, and may be slightly faster during REM sleep). Range of blood oxygen saturation variation: Normal sleep: 95-100% (deep sleep may be slightly lower, but is usually >92%) Abnormal value: <90% indicates hypoxemia, requiring emergency intervention (such as adjusting the oxygen concentration in the bedroom).

[0065] The corresponding sleep quality impact data include body movement frequency and sleep continuity, heart rate and sleep depth, respiratory rate and sleep apnea, and blood oxygen saturation and hypoxemia.

[0066] Categorized storage: Store the range of changes and corresponding sleep quality data by user ID, sleep cycle (e.g., cycle 1, cycle 2), and physiological indicator type (body movement, heart rate, etc.).

[0067] S22. Based on the monitoring data of the corresponding monitoring points and the physiological-time-sleep quality database, obtain the influence weight of each physiological indicator, and based on the influence weight of each physiological indicator, obtain the physiological rhythm constraint factor corresponding to each sleep cycle.

[0068] Specifically, machine learning models are used to quantify the correlation between various physiological indicators and sleep quality indicators. For example, body movement frequency is positively correlated with the number of awakenings (correlation coefficient r=0.72), meaning that the higher the body movement frequency, the more awakenings. Heart rate variability is negatively correlated with the proportion of deep sleep (r=-0.65), meaning that the greater the heart rate fluctuation, the shorter the deep sleep time.

[0069] Based on the magnitude of the correlation coefficient, an initial weight is assigned to each physiological indicator. For example: High-impact weights (such as body velocity, heart rate variability).

[0070] : The weight of the influence (such as respiratory rate stability).

[0071] Low impact weight (e.g., fluctuations in blood oxygen saturation, unless there is an abnormality).

[0072] The circadian rhythm constraint factor is a quantitative constraint parameter generated by combining the range of changes in the target user's physiological indicators during the sleep cycle and their weighted impact on sleep quality. It reflects the adaptability requirements of the user's circadian rhythm (such as sleep stage and fluctuations in physiological indicators) to sleep intervention. It is used to adjust the intensity and timing of intervention parameters (such as ambient temperature, humidity, and noise) to ensure that the intervention measures conform to the user's natural physiological rhythm.

[0073] The steps for generating physiological rhythm constraint factors are as follows: For each sleep cycle (e.g., cycle 1, cycle 2), extract the average or range of variation of various physiological indicators for the target user within that cycle (e.g., average body movement frequency, heart rate fluctuation range). Example: Cycle 1: Average body movement frequency 3 beats / minute, average heart rate 65 bpm, average respiratory rate 14 breaths / minute. Cycle 2: Average body movement frequency 1 beat / minute, average heart rate 55 bpm, average respiratory rate 12 breaths / minute.

[0074] The quantitative value of each physiological indicator is multiplied by its corresponding influence weight, and the results are summed to generate the physiological rhythm constraint factor for that cycle. : ;in, The physiological rhythm constraint factor represents a certain sleep cycle (such as the first cycle or the second cycle). It is used to quantify the comprehensive result of the changes in the target user's physiological indicators and their weights on sleep quality within the sleep cycle, and to reflect the adaptability requirements of the user's physiological rhythm to sleep intervention. For the number of physiological indicators; This represents the quantified value of the i-th physiological indicator; This represents the influence weight corresponding to the i-th physiological indicator; Normalize to the [0,1] interval: ;in, and For all periods The maximum and minimum values.

[0075] S23. Using the acquired physiological rhythm constraint factors, the intervention parameters corresponding to each monitoring point are filtered according to the target sleep state parameters to obtain the filtered intervention parameters.

[0076] Intervention parameters are the specific variables that need to be adjusted and controlled during sleep intervention, including environmental control parameters and equipment intervention parameters.

[0077] Environmental control parameters are a core component of intervention parameters, used to adjust environmental conditions within the sleep monitoring area to suit the physiological needs of the target user. Specifically, they include: Temperature: The optimal temperature for the target user (e.g., 22℃) and the adjustment step size (e.g., 0.5℃ each time).

[0078] Humidity: The optimal humidity for the target user (e.g., 50%RH) and the adjustment range (e.g., ±5%RH).

[0079] Noise: Noise masking intensity (e.g., decibel value, adjusted by a white noise machine or ambient sound) or noise cancellation level (e.g., parameters of active noise cancellation equipment).

[0080] Light intensity (such as the brightness of a nightlight) and color temperature (such as warm light 2700K) are used to regulate sleep rhythm.

[0081] Device intervention parameters refer to the specific settings of devices (such as smart mattresses and breathing assist devices) controlled by a sleep monitoring system, used to directly improve the user's sleep. These include: Mattress firmness: Adjust the mattress support according to the user's body movement frequency and sleep stage (such as switching between firmness modes).

[0082] Respiratory support equipment: oxygen concentrator flow rate (e.g., 2L / min) or CPAP (continuous positive airway pressure) equipment pressure parameters (e.g., 10cmH2O).

[0083] Music / White Noise: Playback content (e.g., alpha wave music, natural rain sounds) and volume (e.g., 30dB).

[0084] Filter out intervention parameters that do not conform to the user's physiological rhythm. For example, if the user is in a deep sleep ( If the blood oxygen saturation is low, then high-intensity noise masking parameters will be filtered out (to avoid disturbing deep sleep). If the user's blood oxygen saturation is <90%, then the oxygen concentrator intervention parameters will be activated first (such as adjusting the flow rate to 3L / minute).

[0085] S24. Based on the polysomnography (PSG) standard under sleep conditions and the actual sleep indicators corresponding to the filtered intervention parameters of the real-time monitoring points, obtain the deviation of the required sleep indicators.

[0086] Polysomnography (PSG) is the gold standard in clinical sleep medicine. It defines core parameters of sleep quality by simultaneously recording electroencephalography (EEG), eye movement (EOG), electromyography (EMG), respiration, and blood oxygenation. In this embodiment, the PSG standard is converted into quantitative indicators, serving as an ideal reference for the sleep quality of the target user.

[0087] The deviations of the sleep demand index are divided into two categories: absolute deviation (the direct difference between the index value and the standard) and relative deviation (the difference between the rate of change of the index and the expectation). These deviations are integrated into a sleep demand index deviation matrix.

[0088] S25. Based on the deviation of the sleep demand index, the index deviation is corrected through an adaptive adjustment function, and the intervention parameters after the index correction is completed are obtained.

[0089] The deviation of the demand sleep index is a vector. : ;in This represents the deviation of the i-th indicator (such as deviation in the number of awakenings, heart rate deviation, etc.). The output of the adaptive adjustment function f is the adjustment amount of the intervention parameter. : ,Right now: .

[0090] Based on the deviation between actual sleep indicators and required sleep indicators (such as excessive number of awakenings), intervention parameters are corrected through an adaptive adjustment function.

[0091] For a single type of deviation (such as arousal frequency deviation), linear proportional adjustment is used: ;in, This is the proportionality coefficient (which controls the adjustment range of the current deviation). Integral coefficient (to eliminate static bias); The differential coefficient (to suppress overshoot); The time interval is used when calculating the integral term. Time is the smallest unit for discretizing time, used to accumulate deviation information at different times.

[0092] For multiple bias types (such as high heart rate + low blood oxygen saturation), a weighted nonlinear model is used: ;in, This represents the adjustment amount for the j-th intervention parameter, used to refine the precise adjustment of different intervention parameters; This indicates the number of deviation types, that is, the number of types of sleep indicator deviations that currently exist; The weight of the i-th deviation (determined by the physiological rhythm constraint factor); For non-linear activation functions (such as the Sigmoid function, which limits the adjustment range); the sleep stage factor is a weight that is adjusted according to the current sleep stage (such as reducing the noise adjustment range during deep sleep). denoted as the deviation of the i-th sleep index.

[0093] S26. Based on the intervention parameters after indicator correction, construct sleep adaptation regulation factors through sleep cycle, sleep depth, number of awakenings and temperature and humidity obtained from the environmental simulation model, adjust the intervention intensity of the intervention parameters after indicator correction, and obtain the intervention parameters after intervention intensity adjustment. S27. Based on the obtained physiological rhythm constraint factors, sleep adaptation regulation factors, and index correction process, the sleep intervention function fitted by the constraint conditions constructed by combining the real-time physiological parameters and demand physiological parameters of the monitoring points, as well as the physiological rhythm constraint factors, sleep adaptation regulation factors, and index deviation, is trained using a genetic algorithm to obtain the trained sleep intervention function and its corresponding weight parameters, thereby obtaining the sleep intervention regulation factor.

[0094] The sleep intervention function F is used to quantify the correlation between intervention parameters (such as temperature, humidity, and noise) and sleep quality indicators (such as the proportion of deep sleep and the number of awakenings). Its mathematical expression is: in, The weight of the k-th intervention parameter (determined jointly by the physiological rhythm constraint factor and the sleep adaptation regulation factor); For the nonlinear influence function of the k-th intervention parameter (such as the Sigmoid effect of temperature on the proportion of deep sleep); λ is the penalty coefficient for the deviation of the i-th sleep index (determined by the importance of the index, such as a higher λ for the deviation of the number of awakenings). The deviation of the i-th sleep index (such as heart rate deviation, blood oxygen deviation); This indicates the number of deviation types, that is, the number of types of sleep index deviations that currently exist.

[0095] Step 3: Based on the sleep state monitoring matrix and the configured personalized sleep improvement model, generate personalized intervention instructions to control the activation sequence and execution time of intervention measures for each monitoring point corresponding to the target user. At the same time as the intervention measures are activated, the parameters corresponding to each activated intervention measure are adjusted based on the sleep intervention regulation factor until the preset ideal sleep state parameters obtained based on the polysomnography (PSG) standard and human physiological rhythm in the sleep environment are met, so as to obtain a personalized sleep improvement plan that meets the needs of the target user.

[0096] The construction process of the personalized sleep improvement model includes: constructing a personalized sleep improvement model based on reinforcement learning, combining a sleep intervention function, a sleep state monitoring matrix, and preset intervention trigger conditions; training the model using sleep data, physiological indicators, environmental parameters, and lifestyle data of target users at different time periods; embedding the trained model into the corresponding sleep monitoring system for real-time sleep intervention control; collecting feedback information from target users through a sleep quality assessment scale; obtaining an intervention effect score through comprehensive evaluation; and constructing a reinforcement reward function to feed back into the personalized sleep improvement model for continuous training.

[0097] The preset intervention triggering conditions include: when the target user is awake and has not reached the preset sleep time, all monitoring points are in a closed state; otherwise, all monitoring points are in a standby state. When all monitoring points are in a standby state, when the target user enters the starting state point of the effective monitoring interval, the corresponding monitoring point is activated through the sleep state monitoring matrix, and the intervention parameters of the activated monitoring point are synchronously adjusted through the sleep intervention function based on the real-time collected physiological parameters, environmental parameters, and sleep stage parameters. When the target user reaches the state point corresponding to the lower limit of the effective monitoring interval, the corresponding monitoring point is converted to a standby state through the sleep state monitoring matrix. Otherwise, the current monitoring point is in a running state, and the above process is repeated to adjust the intervention parameters corresponding to the current monitoring point in real time based on the target user and monitoring point parameters within the corresponding effective monitoring interval.

[0098] The beneficial effects of this embodiment are as follows: by dynamically redundancy interval correction and physiological rhythm-driven intervention regulation, the utilization rate of monitoring data is optimized and invalid calculations are reduced; by combining environmental influence functions and PSG standards, parameters such as temperature, humidity, and noise are adjusted to meet the physiological needs of users; by using reinforcement learning to build personalized models, intervention strategies are continuously optimized to achieve a closed loop from monitoring to intervention, thereby improving sleep quality and user comfort.

[0099] All formulas in this invention are dimensionless and calculated numerically. The preset parameters in the formulas can be set by those skilled in the art according to the actual situation.

[0100] The weighting coefficients of this invention are used to measure the degree of influence of different factors or variables on a certain outcome or decision. The weighting coefficient is defined as the numerical value assigned to each factor when comparing and evaluating multiple factors, reflecting their importance or priority. These weighting coefficients can be determined according to specific circumstances and needs, and are usually jointly formulated and confirmed by professionals or relevant stakeholders. By reasonably setting the weighting coefficients, programs or systems can be helped to make decisions or predictions more accurately.

[0101] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0102] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to specific implementations. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A sleep quality improvement system based on big data analysis, characterized in that, The system includes: Sleep state monitoring matrix construction module: used to construct a sleep state monitoring matrix based on the distance between each monitoring point and the target user, the monitoring direction of the monitoring point to the target user, and the monitoring interval within the preset sleep monitoring area; Sleep intervention regulation factor acquisition module: used to obtain sleep intervention regulation factors based on real-time environmental parameters of the sleep monitoring area and a preset environmental influence function; Personalized sleep improvement plan generation module: It is used to generate personalized intervention instructions based on the sleep state monitoring matrix and the configured personalized sleep improvement model, control the start order and execution time of intervention measures for each monitoring point corresponding to the target user, and adjust the intervention measure parameters based on the sleep intervention adjustment factor until the preset ideal sleep state parameters are met.

2. The sleep quality improvement system based on big data analysis according to claim 1, characterized in that, The system workflow includes: Step 1: Construct a sleep state monitoring matrix based on the distance between each monitoring point and the target user, the monitoring direction corresponding to the target user, and the monitoring interval within the preset sleep monitoring area. The construction process of the sleep state monitoring matrix includes: constructing a dynamic monitoring model for each monitoring point and the original dynamic interval of the monitoring data in the time dimension based on the monitoring parameters of the monitoring points; obtaining the information redundancy interval of adjacent monitoring points through sleep cycle analysis and deep learning algorithms to form an information redundancy interval set; correcting the time dimension interval of the monitoring points by combining dynamic adjustment factors and data augmentation algorithms to obtain an effective monitoring interval set; constructing a sleep stage transition prediction dataset based on the real-time sleep state, physiological indicator sequence, expected sleep stage time, and monitoring point system delay of the target user within the effective monitoring interval set; inputting the compensated sleep stage transition function constructed by a recurrent neural network to obtain the compensated sleep stage transition trajectory and real-time sleep state points, thereby determining the intervention time point set; constructing a sleep state monitoring matrix based on the intervention time point set using a 0-1 matrix, where the matrix element values ​​are jointly determined by the monitoring intervention factor, the duration of continuous sleep, the time length to enter the lower limit of the interval, and the intervention delay time length. Step 2: Based on the real-time environmental parameters of the sleep monitoring area and the preset environmental influence function, obtain the sleep intervention regulation factor; the environmental influence function is associated with temperature, humidity, noise and the corresponding optimal parameters, tolerance deviation range and weight coefficients, and the noise term uses a logarithmic function to reflect the nonlinear perception of the human ear; the process of obtaining the sleep intervention regulation factor includes establishing a physiological-time-sleep quality database, calculating physiological rhythm constraint factors, filtering intervention parameters, obtaining the deviation of required sleep indicators, correcting the indicator deviation, adjusting the intervention intensity and training the sleep intervention function through a genetic algorithm; Step 3: Based on the sleep state monitoring matrix and the configured personalized sleep improvement model, generate personalized intervention instructions to control the activation sequence and execution time of intervention measures for each monitoring point corresponding to the target user. While the intervention measures are activated, adjust the parameters corresponding to each activated intervention measure based on the sleep intervention regulation factor until the preset ideal sleep state parameters obtained based on the polysomnography standard and human physiological rhythm in the sleep environment are met, thereby obtaining a personalized sleep improvement plan that meets the needs of the target user.

3. The sleep quality improvement system based on big data analysis according to claim 2, characterized in that, In the process of constructing the sleep state monitoring matrix in step one, the dynamic monitoring model of the monitoring points is constructed based on the sleep region range, monitoring point distribution, number of monitoring points, and individual monitoring parameters extracted by the sleep monitoring system. It combines the monitoring accuracy, monitoring range, data acquisition frequency, maximum effective monitoring distance, and signal attenuation coefficient corresponding to the monitoring parameters of each monitoring point, and is realized through a multi-dimensional sleep state model combined with an environmental influence function. The process of obtaining the effective monitoring interval set is as follows: the initial effective monitoring interval of the second monitoring point is obtained by subtracting the product of the interval length of the information redundancy interval corresponding to the first and second monitoring points in the same time dimension and the dynamic adjustment factor from the monitoring data time dimension interval corresponding to the second monitoring point. After the data quality is corrected by inputting the initial effective monitoring interval into the data augmentation algorithm, the corrected effective monitoring interval is obtained. Repeat the above process to obtain the corrected effective monitoring intervals for all monitoring points on the same time series, forming a set of effective monitoring intervals.

4. The sleep quality improvement system based on big data analysis according to claim 3, characterized in that, The signal attenuation coefficient was obtained through simulation experiments by monitoring the characteristic relationship between the power of the monitoring device, the signal strength, and the distance. The characteristic relationship is that the signal strength decreases logarithmically with increasing distance, and the rate of attenuation is determined by the signal attenuation coefficient. This coefficient reflects the influence of the environment on signal attenuation, and the value of the signal attenuation coefficient is different under different environments.

5. A sleep quality improvement system based on big data analysis according to claim 2, characterized in that, The acquisition of the information redundancy interval includes the following steps: a. Based on the monitoring accuracy, monitoring range, data acquisition frequency, maximum effective monitoring distance, and signal attenuation coefficient of each monitoring point, construct dynamic monitoring models for two adjacent monitoring points respectively, and define their respective monitoring intervals in the time series; b. Determine the maximum monitoring interval range of each monitoring point based on hardware performance or algorithm presets, which is the theoretically coverable time range of the monitoring point; c. Divide sleep time according to physiological sleep cycles, analyze the change pattern of target user's sleep state in each cycle, align the monitoring data of adjacent monitoring points by cycle, identify the temporal correlation of monitoring data in the same cycle, and determine the overlapping part of monitoring data in the time dimension; d. Based on the aligned time series data, apply deep learning algorithms to extract the temporal features, spatial features, and state features of the monitoring data, identify the overlapping intervals of adjacent monitoring points in the time dimension, and mark the overlapping interval as an information redundancy interval when the length of the overlapping interval exceeds a preset threshold and the data correlation is lower than the threshold; e. Repeat the above steps a and b for all adjacent monitoring point pairs, quantify the redundant time intervals between each pair of adjacent monitoring points, and collect all redundant intervals to obtain the information redundancy interval set.

6. The sleep quality improvement system based on big data analysis according to claim 2, characterized in that, The dynamic adjustment factor is obtained through the following process: Step 1: Based on sleep medicine research and a large amount of user data, preset the baseline turning frequency threshold corresponding to different sleep stages; Step 2: Collect target user's body movement data through monitoring points at fixed time windows, count the number of real-time turnings per unit time, and calculate the real-time turning frequency; Step 3: Combine the real-time predicted sleep stages in the sleep state monitoring matrix, and select the baseline turning frequency threshold for the corresponding stage; Step 4: Quantify the deviation between the real-time turning frequency and the baseline turning frequency using a formula to obtain the frequency deviation rate; Step 5: Based on the frequency deviation rate, obtain the dynamic adjustment factor through linear or nonlinear function mapping. When the frequency deviation rate is small, the adjustment factor takes a smaller value; when the frequency deviation rate is large, the adjustment factor takes a larger value, and the adjustment factor is updated in real time with changes in sleep stages.

7. The sleep quality improvement system based on big data analysis according to claim 2, characterized in that, The process of establishing the physiological-time-sleep quality database in step two, which involves acquiring sleep intervention regulation factors, is as follows: sleep time is divided into 90-minute sleep cycles; the range of changes in the physiological indicators of the target user and the corresponding sleep quality impact data within each sleep cycle are collected and organized, and stored according to user ID, sleep cycle, and physiological indicator type; the calculation process of the physiological rhythm constraint factor is as follows: the correlation between each physiological indicator and the sleep quality indicator is quantified through a machine learning model; an influence weight is assigned to each physiological indicator based on the correlation coefficient; the average value or range of change of each physiological indicator of the target user within each sleep cycle is extracted; the quantified value of each physiological indicator is multiplied by its influence weight and summed; and the physiological rhythm constraint factor for each sleep cycle is obtained after normalization.

8. A sleep quality improvement system based on big data analysis according to claim 2, characterized in that, The process of obtaining sleep intervention regulation factors in step two further includes: filtering intervention parameters: using physiological rhythm constraint factors, filtering intervention parameters according to target sleep state parameters, including environmental control parameters and equipment intervention parameters, filtering out intervention parameters that do not conform to the user's physiological rhythm; obtaining the required sleep index deviation: based on the polysomnography standard and the actual sleep index corresponding to the filtered intervention parameters at real-time monitoring points, obtaining absolute and relative deviations, and integrating them into the required sleep index deviation; correcting the index deviation: based on the required sleep index deviation, correcting the index deviation through an adaptive adjustment function, using linear proportional adjustment for single deviation types and weighted nonlinear model adjustment for multiple deviation types; adjusting the intervention intensity: based on the intervention parameters after index correction, constructing a sleep adaptive regulation factor by combining sleep cycle, sleep depth, number of awakenings, and temperature and humidity obtained from the environmental simulation model, and adjusting the intervention intensity; training the sleep intervention function: using a genetic algorithm, combining the constraints constructed by real-time physiological parameters and required physiological parameters at monitoring points, and the sleep intervention function fitted by the physiological rhythm constraint factor, sleep adaptive regulation factor, and index deviation, training to obtain the trained sleep intervention function and its corresponding weight parameters, thereby obtaining the sleep intervention regulation factor.

9. A sleep quality improvement system based on big data analysis according to claim 2, characterized in that, The construction process of the personalized sleep improvement model includes: building a personalized sleep improvement model based on reinforcement learning, combining a sleep intervention function, a sleep state monitoring matrix, and preset intervention trigger conditions; training the model using sleep data, physiological index data, environmental parameter data, and lifestyle habit data of target users at different time periods; embedding the trained model into the corresponding sleep monitoring system to achieve real-time sleep intervention control; collecting feedback information from target users through a sleep quality assessment scale, comprehensively evaluating to obtain an intervention effect score, constructing a reinforcement reward function, and feeding it back into the personalized sleep improvement model to achieve continuous model training; the preset intervention trigger conditions include: when the target user is awake and it is not yet the preset sleep time, all monitoring points are in a closed state. The system operates as follows: When the target user has reached the preset sleep time but has not yet entered the starting point of the effective monitoring interval, all monitoring points are in standby mode. When all monitoring points are in standby mode and the target user enters the starting point of the effective monitoring interval, the corresponding monitoring point is activated through the sleep state monitoring matrix. Based on the real-time collected physiological parameters, environmental parameters, and sleep stage parameters, the intervention parameters of the activated monitoring point are synchronously adjusted through the sleep intervention function. When the target user reaches the state point corresponding to the lower limit of the effective monitoring interval, the corresponding monitoring point is converted to standby mode through the sleep state monitoring matrix. If the target user has not reached the lower limit, the current monitoring point remains operational, and the intervention parameters corresponding to the current monitoring point are repeatedly adjusted based on the target user and monitoring point parameters within the effective monitoring interval.

Images