HRV-based sleep monitoring analysis method and system

Abstract

The invention discloses a sleep monitoring analysis method and system based on HRV, and relates to the technical field of sleep monitoring, and the method comprises the following steps: at the beginning of monitoring, continuously recording a continuous heartbeat time interval of a sleeper at a preset initial heartbeat interval acquisition frequency for evaluating the autonomic nervous system state of the sleeper at night; the respiration data of the sleeper is acquired in real time while the heart rate data is acquired. By combining real-time respiration data and an intelligent increment control algorithm, when irregular respiration is detected, the heartbeat interval acquisition frequency is dynamically adjusted, so that the sampling precision of heart rate change is improved, fine fluctuation of a sleeper when apnea or irregular respiration occurs can be captured, and the sleep quality of the sleeper can be improved. In addition, the detection accuracy of the apnea event can be improved under high sensitivity, potential health risks can be recognized in time, more reliable data support is provided for improvement of sleep quality and early intervention, and therefore health damage caused by missing detection is effectively avoided.

Description

Technical Field

[0001] The present invention relates to the technical field of sleep monitoring, and in particular to a sleep monitoring and analysis method and system based on HRV. Background Art

[0002] Sleep monitoring and analysis based on HRV (Heart Rate Variability) is a technology that evaluates the body's recovery by analyzing changes in the autonomic nervous system (ANS) during sleep. Specifically, HRV monitoring can determine whether the body is in a state of relaxation and recovery by detecting changes in heart rate at night. Changes in the activity of the autonomic nervous system at night directly affect the body's recovery and health status. Therefore, HRV can be used as an important indicator of sleep quality. In sleep monitoring combined with HRV analysis, other functions can also be added, such as skin temperature monitoring to prevent the danger of outdoor hypothermia, and sleep apnea monitoring through PPG sensors (photoelectric volume pulse wave sensors) Real-time collection of heart rate, RRI (RR interval), blood oxygen data, etc., calculates respiratory rate, timely identifies whether there is apnea, and provides sound and light prompts for intervention. In addition, through the built-in 6-axis sensor, the device can also detect changes in the user's posture during sleep or exercise, help users improve their body shape and posture during exercise, and perform post-exercise analysis; at the same time, the pace and distance during running can also be accurately recorded and analyzed, and even in the absence of a watch or APP, the data can be stored offline for subsequent analysis and optimization. This comprehensive sleep and exercise monitoring solution not only provides users with in-depth analysis of sleep quality, but also helps optimize athletic performance and protect user health, especially in detecting and intervening in physical recovery and apnea.

[0003] HRV-based sleep monitoring and analysis reflects the regulation state of the autonomic nervous system (ANS) during sleep at night by detecting the indicator of heart rate variability (HRV). Its core working principle is to collect the user's heartbeat interval (RR interval) data using photoplethysmography (PPG) or electrocardiogram (ECG) technology. By calculating the time fluctuations of these intervals, HRV parameters such as standard deviation (SDNN), high-frequency power (HF) and low-frequency power (LF) are further extracted for time domain, frequency domain and nonlinear analysis. High-frequency components are mainly related to parasympathetic nerve activity, while low-frequency components reflect the balance state of sympathetic and parasympathetic nerves. In sleep monitoring, the system analyzes these parameters to evaluate whether the body is in a relaxed (parasympathetic nerve dominant) or tense (sympathetic nerve dominant) state at night, thereby judging the user's recovery quality and providing personalized health analysis and suggestions in combination with sleep stage data.

[0004] The prior art has the following deficiencies: When performing sleep monitoring and analysis based on HRV, existing technologies usually use a preset fixed heartbeat interval acquisition frequency, and the acquisition frequency is not set too high, so as to effectively balance data accuracy and energy consumption in wearable devices, extend battery life, simplify data processing, and ensure stable operation of the device throughout sleep. However, when the sleeper has irregular breathing, this detection frequency is difficult to capture subtle fluctuations in heart rate changes. Irregular breathing may be an early sign of sleep apnea. If these tiny fluctuations cannot be accurately captured, serious apnea events may be missed, leading to persistent hypoxemia (a significant drop in blood oxygen levels), causing irreversible damage to the cardiovascular and cerebrovascular systems, including serious consequences such as hypertension, arrhythmias, and stroke.

[0005] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0006] The purpose of the present invention is to provide a sleep monitoring and analysis method and system based on HRV. By combining real-time respiratory data and an intelligent incremental control algorithm, the heart rate interval acquisition frequency is dynamically adjusted when irregular breathing is detected, thereby improving the sampling accuracy of heart rate changes. It can not only capture the subtle fluctuations of the sleeper when apnea or irregular breathing occurs, but also improve the detection accuracy of apnea events under high sensitivity, and timely identify potential health risks such as hypoxemia, arrhythmia, etc., thereby providing more reliable data support for improving sleep quality and early intervention, thereby effectively avoiding health damage caused by missed detection, so as to solve the problems in the above-mentioned background technology.

[0007] In order to achieve the above object, the present invention provides the following technical solution: a sleep monitoring and analysis method based on HRV, comprising the following steps: At the beginning of monitoring, the continuous heartbeat time intervals of the sleeper are continuously recorded at a preset initial heartbeat interval collection frequency to evaluate the sleeper's autonomic nervous system state at night; Acquire the sleeper's breathing data in real time while collecting heart rate data, and pre-process the acquired raw breathing data to ensure the accuracy and consistency of the data; Extract key features reflecting irregular breathing from the pre-processed breathing data, analyze and quantify the extracted key breathing features within a predetermined monitoring window, and preliminarily evaluate the degree of irregular breathing of the sleeper within each monitoring window; The quantified irregular breathing data is input into the pre-trained long short-term memory network for recognition to determine whether the sleeper has irregular breathing; Once the long short-term memory network identifies that the sleeper has irregular breathing, the incremental control algorithm will be used to dynamically adjust the heart rate interval collection frequency based on the prediction results. Specifically, the frequency of heart rate interval collection of the sleeper will be increased to capture subtle fluctuations in heart rate changes and obtain more detailed HRV data at a higher collection frequency.

[0008] Preferably, the setting of the initial heartbeat interval acquisition frequency strikes a balance between data accuracy and energy efficiency, and is specifically set to 1 Hz to 4 Hz, which not only meets the requirements of sleep state assessment, but also reduces device power consumption and prolongs battery life.

[0009] Preferably, key features reflecting irregular breathing are extracted from the preprocessed respiratory data, wherein the extracted features include the variation amplitude of the respiratory rhythm and the fluctuation of the respiratory depth. The extracted variation amplitude of the respiratory rhythm and the fluctuation of the respiratory depth are analyzed and quantified within a predetermined monitoring window, and a respiratory rhythm instability quantification value and a respiratory depth variation quantification value are generated respectively. The respiratory rhythm instability quantification value and the respiratory depth variation quantification value are used to preliminarily evaluate the degree of irregular breathing of the sleeper within each monitoring window.

[0010] Preferably, the respiratory rhythm instability quantification value and the respiratory depth variation quantification value generated after quantifying the variation amplitude of the respiratory rhythm and the fluctuation of the respiratory depth are input into a pre-trained long short-term memory network, and the respiratory evaluation coefficient generated by the long short-term memory network is used to determine whether the sleeper has irregular breathing.

[0011] Preferably, the breathing assessment coefficient generated when predicting the irregular breathing of the sleeper through the pre-trained long short-term memory network is compared and analyzed with the pre-set breathing assessment coefficient reference threshold to determine whether the sleeper has irregular breathing. The specific steps are as follows: If the breathing assessment coefficient is greater than the breathing assessment coefficient reference threshold, a breathing irregularity signal is generated, and it is determined that the sleeper has irregular breathing; If the breathing assessment coefficient is less than or equal to the breathing assessment coefficient reference threshold, a normal breathing signal is generated, and it is determined that the sleeper does not have irregular breathing, which indicates that the sleeper is breathing normally.

[0012] Preferably, within a predetermined monitoring window, the specific steps of analyzing and quantifying the variation amplitude of the extracted respiratory rhythm to generate a quantitative value of respiratory rhythm instability are as follows: First, extract the respiratory cycle sequence within each monitoring window from the preprocessed respiratory data , ,in Indicates iThe duration of the breathing cycle is used to calculate the instantaneous fluctuation rate between each two consecutive cycles. The calculation expression is: , in: For the The duration of the breathing cycle, is a normalization factor used to eliminate the influence of absolute period differences, is the weighting factor, and the weight increases with the time distance from the monitoring window center The distance changes exponentially. To adjust the weight sensitivity parameter, It is i The instantaneous fluctuation rate of each breath, n Indicates the total number of breathing cycles in the current monitoring window; Constructing instantaneous volatility series through instantaneous volatility , , the quantitative value of respiratory rhythm instability is calculated based on the instantaneous fluctuation rate sequence, and the calculation expression is: , in: It is a quantitative value for the instability of respiratory rhythm. is a dynamic adjustment factor that emphasizes the contribution of fluctuations at the tail of the monitoring window. To monitor the window position weight parameter, adjust the impact of the tail data, is the cumulative value of the fluctuation rate of the respiratory cycle at the beginning and end of the monitoring window, is the cumulative value of the respiratory cycle fluctuation rate in the central area of ​​the monitoring window, is the edge weight adjustment factor, controlling the contribution of the head and tail fluctuations, A small constant to prevent the denominator from being zero.

[0013] Preferably, within a predetermined monitoring window, the specific steps of analyzing and quantifying the fluctuation of the extracted breathing depth to generate a quantitative value of breathing depth variation are as follows: In the predetermined monitoring window, the breathing depth of each breathing cycle is extracted from the preprocessed breathing data to represent the amount of gas inhaled in each breath. The time-weighted mean is calculated using the breathing depth. The calculation expression is: , in: It is a time-weighted mean that smoothes fluctuations in breathing depth, eliminates interference from short-term noise, and provides a stable benchmark value for subsequent calculations. is the length of the monitoring window, is the breathing depth at time t, is a time-weighted function that reflects the importance of breathing depth over time; Based on the extracted respiratory depth data and time-weighted mean, the depth deviation at each time point is further calculated, and the degree of variation is quantified to generate a quantitative value of respiratory depth variation. The generated expression is: , in: is the deviation between the breathing depth and the time-weighted mean, reflecting the instantaneous fluctuation of the breathing depth. is the normalized deviation, which is used to eliminate the influence of different users or breathing depth baseline differences. Quantify the variability of breathing depth.

[0014] Preferably, once the long short-term memory network identifies that the sleeper has irregular breathing, an incremental control algorithm is used to dynamically adjust the acquisition frequency of the heartbeat interval according to the prediction result. The specific steps are as follows: Once the LSTM network identifies that the sleeper has irregular breathing, the adjustment factor of the acquisition frequency is first calculated based on the deviation between the breathing assessment coefficient and the reference threshold of the breathing assessment coefficient. The adjustment factor is used to indicate the degree to which the acquisition frequency needs to be increased. The calculation expression is: , in, is the frequency increment proportional constant, which is used to control the overall amplitude of the adjustment. It is a breathing evaluation coefficient generated by the long short-term memory network, reflecting the current degree of breathing abnormality. It is the reference threshold of the breathing assessment coefficient and defines the dividing line between normal breathing and abnormal breathing; After calculating the adjustment factor, the heartbeat interval acquisition frequency is updated to increase the frequency to capture subtle changes in heart rate. The expression for updating the heartbeat interval acquisition frequency is: , in: The updated heartbeat interval acquisition frequency is used to enhance the sensitivity of data acquisition. is the initial heartbeat interval acquisition frequency, indicating the acquisition frequency under normal conditions; In order to avoid excessive frequency adjustment resulting in a sharp increase in energy consumption or data redundancy, a dynamic frequency range limit is set to ensure that the updated heartbeat interval collection frequency is adjusted within a controllable range. The limit formula is: , in, The final adjusted heartbeat interval acquisition frequency is used as the actual application frequency value. The upper limit of the heartbeat interval acquisition frequency is to ensure that it does not run overloaded. It is the lower limit of the heartbeat interval collection frequency to ensure that the data accuracy meets the monitoring requirements.

[0015] The sleep monitoring and analysis system based on HRV includes an initial heartbeat interval acquisition module, a respiratory data acquisition module, a respiratory data preprocessing module, a respiratory irregularity feature extraction and analysis module, an irregular respiratory recognition module, and an acquisition frequency adjustment module; The initial heartbeat interval acquisition module continuously records the time intervals of the sleeper's continuous heartbeats at a preset initial heartbeat interval acquisition frequency at the beginning of monitoring, so as to evaluate the state of the sleeper's autonomic nervous system at night; The breathing data collection module acquires the sleeping person's breathing data in real time while collecting heart rate data; The respiratory data preprocessing module preprocesses the acquired raw respiratory data to ensure the accuracy and consistency of the data; The breathing irregularity feature extraction and analysis module extracts key features reflecting breathing irregularity from the preprocessed breathing data, analyzes and quantifies the extracted key breathing features within a predetermined monitoring window, and preliminarily evaluates the degree of irregular breathing of the sleeper within each monitoring window; The irregular breathing recognition module inputs the quantified irregular breathing data into the pre-trained long short-term memory network for recognition to determine whether the sleeper has irregular breathing; The acquisition frequency adjustment module, once the long short-term memory network recognizes that the sleeper has irregular breathing, will use the incremental control algorithm to dynamically adjust the acquisition frequency of the heartbeat intervals based on the prediction results. Specifically, it will increase the acquisition frequency of the sleeper's heartbeat intervals to capture subtle fluctuations in heart rate changes and obtain more detailed HRV data at a higher acquisition frequency.

[0016] In the above technical solution, the technical effects and advantages provided by the present invention are: The sleep monitoring method based on HRV of the present invention combines real-time respiratory data and an intelligent incremental control algorithm. When irregular breathing is detected, the heart rate interval collection frequency is dynamically adjusted, thereby improving the sampling accuracy of heart rate changes. It can not only capture subtle fluctuations when the sleeper has apnea or irregular breathing, but also improve the detection accuracy of apnea events under high sensitivity, and timely identify potential health risks such as hypoxemia and arrhythmia, thereby providing more reliable data support for improving sleep quality and early intervention, thereby effectively avoiding health damage caused by missed detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0018] Figure 1 The figure is a flow chart of the sleep monitoring and analysis method based on HRV of the present invention.

[0019] Figure 2 It is a module schematic diagram of the HRV-based sleep monitoring and analysis system of the present invention. DETAILED DESCRIPTION

[0020] Example embodiments will now be described more fully with reference to the accompanying drawings. However, example embodiments can be implemented in a variety of forms and should not be construed as limited to the examples set forth herein; rather, these example embodiments are provided so that the description of the present disclosure will be more comprehensive and complete, and the concept of the example embodiments will be fully conveyed to those skilled in the art.

[0021] The present invention provides Figure 1 The HRV-based sleep monitoring and analysis method shown includes the following steps: At the beginning of monitoring, the sleeper's continuous heartbeat time interval (RR interval) is continuously recorded at a preset initial heartbeat interval collection frequency to evaluate the sleeper's autonomic nervous system (ANS) status at night; This stage is mainly used to collect basic heart rate data and evaluate the state of the sleeper's autonomic nervous system (ANS) at night. By analyzing this data, it is possible to preliminarily determine whether the body is in a state of relaxation (parasympathetic nerve dominance) or tension (sympathetic nerve dominance), thereby evaluating the user's overall recovery quality.

[0022] The setting of the initial heartbeat interval acquisition frequency should find a balance between data accuracy and energy efficiency. Specifically, the frequency needs to be able to capture the main characteristics of heart rate variability under normal conditions (such as time domain and frequency domain parameters), but it does not have to achieve too high a sampling density. It is generally recommended to set it in the range of **1Hz to 4Hz (collecting 1 to 4 heartbeat intervals per second)**, which can not only meet the basic requirements of sleep state assessment (such as dynamic balance analysis of parasympathetic and sympathetic nerves), but also significantly reduce device power consumption and extend battery life. In addition, the selection of this frequency should also take into account sensor performance and data processing capabilities to ensure that the system can operate stably in most user scenarios (such as normal sleep, low activity), while leaving room for dynamic adjustments in subsequent high-risk scenarios.

[0023] Acquire the sleeper's breathing data in real time while collecting heart rate data; Respiratory rate and rhythm are usually monitored through built-in sensors such as accelerometers, airflow sensors, or PPG sensors. Monitoring of respiratory rate and rhythm is usually achieved through built-in high-sensitivity sensors, including accelerometers, airflow sensors, and PPG sensors (photoplethysmography sensors). Accelerometers can capture the respiratory cycle by sensing the tiny ups and downs of the chest, while airflow sensors can directly measure the flow rate and volume of gas in and out of the respiratory tract, providing the most intuitive respiratory signal. PPG sensors use optical signals to detect periodic changes in blood flow, which is related to the fluctuations in blood oxygenation during breathing, and indirectly reflects the rhythm and frequency of breathing. These sensors work together to not only accurately capture real-time respiratory data, but also have the characteristics of high sensitivity and non-invasiveness, making them suitable for long-term monitoring of breathing patterns during sleep.

[0024] Pre-process the acquired raw respiratory data to ensure the accuracy and consistency of the data; The acquired raw respiratory data usually contains noise and interference, so preprocessing is required. The preprocessing steps include noise removal, filtering, smoothing, and signal correction to ensure the accuracy and consistency of the data. Through these processes, clear respiratory signals can be extracted, errors and deviations can be reduced, and the reliability of subsequent feature extraction and analysis can be ensured. This step is crucial to improving the accuracy of the overall detection system.

[0025] Extract key features reflecting irregular breathing from the pre-processed breathing data, analyze and quantify the extracted key breathing features within a predetermined monitoring window, and preliminarily evaluate the degree of irregular breathing of the sleeper within each monitoring window; This step aims to transform complex respiratory data into quantifiable indicators and provide structured data for the input of the long short-term memory network.

[0026] Key features reflecting irregular breathing are extracted from the preprocessed respiratory data, wherein the extracted features include the variation amplitude of respiratory rhythm and the fluctuation of respiratory depth. The extracted variation amplitude of respiratory rhythm and the fluctuation of respiratory depth are analyzed and quantified within a predetermined monitoring window, and a respiratory rhythm instability quantification value and a respiratory depth variation quantification value are generated respectively. The degree of irregular breathing of the sleeper within each monitoring window is preliminarily evaluated through the respiratory rhythm instability quantification value and the respiratory depth variation quantification value.

[0027] When the amplitude of the breathing rhythm of a sleeper presents an unstable dynamic pattern, their breathing is generally considered irregular. This instability is manifested by significant fluctuations in the time intervals of the breathing cycle, which may occur periodically. Acceleration or deceleration, sudden rhythm interruptions, or abnormal changes in depth and frequency. This dynamic pattern is usually caused by a variety of reasons, such as sleep apnea syndrome (SAS), partial airway obstruction, abnormal regulation of the nervous system, or other physiological problems. Irregular breathing rhythm can lead to fluctuations in blood oxygen levels, overactivation of the autonomic nervous system, and decreased sleep quality, and may also pose a potential threat to cardiovascular and cerebrovascular health. Therefore, identifying this dynamic unstable pattern is important for early detection and intervention of breathing abnormalities.

[0028] The specific steps of analyzing and quantifying the variation amplitude of the extracted respiratory rhythm within the predetermined monitoring window to generate the respiratory rhythm instability quantification value are as follows: First, extract the respiratory cycle sequence within each monitoring window from the preprocessed respiratory data , ,in Indicates i The duration of the breathing cycle is used to calculate the instantaneous fluctuation rate between each two consecutive cycles. The calculation expression is: , in: For the The duration of the breathing cycle, is a normalization factor used to eliminate the influence of absolute period differences, is the weighting factor, and the weight increases with the time distance from the monitoring window center The distance changes exponentially. To adjust the weight sensitivity parameter, It is i The instantaneous fluctuation rate of each breath, n Indicates the total number of breathing cycles in the current monitoring window; The instantaneous fluctuation rate can capture the amplitude of the respiratory cycle at each time step, and give higher weight to the cycle changes close to the center of the monitoring window. This weighting method can highlight the most critical abnormal respiratory rhythm signals in the window and provide highly sensitive dynamic information for the generation of quantitative values ​​of respiratory rhythm instability.

[0029] Constructing instantaneous volatility series through instantaneous volatility , , the quantitative value of respiratory rhythm instability is calculated based on the instantaneous fluctuation rate sequence, and the calculation expression is: , in: It is a quantitative value for the instability of respiratory rhythm. is a dynamic adjustment factor that emphasizes the contribution of fluctuations at the tail of the monitoring window. To monitor the window position weight parameter, adjust the impact of the tail data, is the cumulative value of the respiratory cycle fluctuation rate at the beginning and end (edge ​​area) of the monitoring window, is the cumulative value of the respiratory cycle fluctuation rate in the central area of ​​the monitoring window, is the edge weight adjustment factor, which controls the contribution of the head and tail fluctuations to the overall index. To prevent a small constant with a zero denominator; The head and tail areas of the monitoring window refer to the starting and ending parts of the time window, which usually occupy a small part of the entire window (such as the first 20% and the last 20%). The cumulative value of the respiratory cycle fluctuation rate in the head and tail areas is the sum of all instantaneous fluctuation rates in these edge areas after accumulation or weighting, reflecting the dynamic change characteristics of the respiratory rhythm at the edge of the monitoring window. Since the irregularity of the respiratory rhythm is often more significant at the beginning or end stage, the cumulative fluctuation rate in this area can serve as an important basis for identifying abnormal breathing. For example, when a sleeper transitions from one breathing state to another (such as from stable breathing to apnea), the fluctuation rate in the head and tail areas usually increases significantly.

[0030] The central area of ​​the monitoring window refers to the middle part of the time window, usually the remaining part after removing the first and last edge areas (for example, the middle 60%). The cumulative value of the respiratory cycle fluctuation rate in the central area is the sum calculated by accumulating or weighting the instantaneous fluctuation rate in this middle area, which reflects the changing characteristics of the respiratory rhythm during the relatively stable stage within the window. The cumulative fluctuation rate in the central area is usually used to measure the overall stability of the respiratory rhythm. If the cumulative value of the fluctuation rate in the central area is high, it may indicate that there is a persistent irregularity in the respiratory rhythm, such as periodic respiratory fluctuations or long-term abnormal breathing patterns. Compared with the edge area, the cumulative value of the central area is more representative of the respiratory rhythm characteristics of the entire period of time.

[0031] The head and tail areas of the monitoring window and the center area of ​​the monitoring window are not specifically limited here, and can be comprehensively set according to actual scenarios and needs.

[0032] The respiratory rhythm instability quantification value comprehensively considers the instantaneous fluctuation of the respiratory rhythm in the monitoring window and the overall dynamic characteristics. The edge weight adjustment factor assigns different contribution weights to the edge and center areas, so that the respiratory rhythm instability quantification value can sensitively reflect the degree of irregular breathing in the entire window.

[0033] It can be seen from the respiratory rhythm instability quantification value that within the predetermined monitoring window, the larger the performance value of the respiratory rhythm instability quantification value generated by analyzing and quantifying the variation amplitude of the extracted respiratory rhythm, the more obvious the irregular breathing of the sleeper. This is because the respiratory rhythm instability quantification value combines the instantaneous fluctuation rate of the respiratory cycle and the dynamic characteristics of the center and edge areas in the monitoring window. A larger respiratory rhythm instability quantification value usually means that there are significant fluctuations, instability or abnormal patterns in the respiratory rhythm (such as apnea, shallow breathing or periodic breathing, etc.). Therefore, the respiratory rhythm instability quantification value can be used as an important indicator to measure the severity of irregular breathing. The higher the value, the more prominent the abnormality of breathing, which requires further attention or intervention.

[0034] When the sleeper's breathing depth fluctuates greatly, it usually indicates that there is irregularity in their breathing. This is because normal breathing depth fluctuations are generally relatively stable and maintained within a certain range, while significant depth fluctuations may reflect a disorder in the breathing control mechanism, such as alternating deep and shallow breathing (such as Cheyne-Stokes breathing) or irregular ventilation caused by airway obstruction. This irregularity may further cause physiological problems such as intermittent hypoxia and carbon dioxide accumulation, which will have a negative impact on sleep quality and physical recovery.

[0035] The specific steps for analyzing and quantifying the fluctuation of the extracted breathing depth within the predetermined monitoring window to generate the breathing depth variation quantitative value are as follows: In the predetermined monitoring window, the breathing depth of each breathing cycle is extracted from the preprocessed breathing data to represent the amount of gas inhaled in each breath. The time-weighted mean is calculated using the breathing depth. The calculation expression is: , in: It is a time-weighted mean that smoothes fluctuations in breathing depth, eliminates interference from short-term noise, and provides a stable benchmark value for subsequent calculations. The length of the monitoring window, usually 1 minute or 5 minutes. is the breathing depth at time t, is a time-weighted function that reflects the importance of breathing depth over time. A decreasing weight function is often used (e.g. , so that the influence of earlier data on the current monitoring window gradually weakens, and the parameter Control weight decay speed); The amount of gas inhaled during each breath can be calculated by an airflow sensor or the amplitude of chest movement, which will not be described in detail here.

[0036] Based on the extracted respiratory depth data and time-weighted mean, the depth deviation at each time point is further calculated, and the degree of variation is quantified to generate a quantitative value of respiratory depth variation. The generated expression is: , in: is the deviation between the breathing depth and the time-weighted mean, reflecting the instantaneous fluctuation of the breathing depth. is the normalized deviation, which is used to eliminate the influence of different users or breathing depth baseline differences. Quantify the variation of breathing depth; The overall fluctuation of breathing depth can be quantified by calculating the breathing depth variation quantification value. The larger the breathing depth variation quantification value is, the more significant the fluctuation of breathing depth within the monitoring window is, reflecting a higher degree of irregular breathing.

[0037] It can be seen from the respiratory depth variation quantification value that within the predetermined monitoring window, the larger the performance value of the respiratory depth variation quantification value generated by analyzing and quantifying the fluctuation of the extracted respiratory depth, the more significant the fluctuation of the sleeper's respiratory depth within the predetermined monitoring window, which in turn reflects the more obvious irregular breathing. This is because a larger respiratory depth variation quantification value means that the deviation of the respiratory depth from its mean is large and the fluctuation is frequent, which may cause alternation of deep and shallow breathing, apnea or respiratory rhythm disorders. These characteristics are closely related to imbalance in the regulation of the autonomic nervous system or airway obstruction, and are often early signs of abnormal sleep breathing. By quantifying this fluctuation, the respiratory depth variation quantification value provides an intuitive and sensitive indicator for evaluating respiratory regularity, which can be used to quickly identify potential respiratory abnormalities and conduct further analysis or intervention.

[0038] The quantified irregular breathing data is input into the pre-trained long short-term memory network for recognition to determine whether the sleeper has irregular breathing; The respiratory rhythm instability quantification value and the respiratory depth variation quantification value generated by quantifying the amplitude of the respiratory rhythm change and the fluctuation of the respiratory depth are input into the pre-trained long short-term memory network, and the respiratory evaluation coefficient generated by the long short-term memory network is used to determine whether the sleeper has irregular breathing.

[0039] The pre-trained long short-term memory network (LSTM) refers to a neural network model that has been trained and optimized with a large amount of breathing-related data, and is specifically used to analyze and identify complex patterns in time series. Due to its unique memory unit and gating mechanism, the LSTM network can capture long-term and short-term dependencies in time series, and is very suitable for processing dynamic changes in breathing rhythm and depth. When training the model, it is first necessary to build a high-quality dataset that contains a wealth of breathing samples, including normal breathing and irregular breathing patterns (such as apnea, alternating deep and shallow breathing, etc.). These data are annotated to clarify the respiratory state corresponding to each sample to provide supervision signals. In addition, the input features include quantified values ​​of respiratory rhythm instability and quantified values ​​of respiratory depth variation. These quantified features can accurately reflect the core characteristics of the breathing pattern and provide rich information for the learning of the LSTM model.

[0040] The training process usually uses batch input data, combined with time step design, to organize feature data into a time series form. LSTM gradually learns the time correlation in the data through its memory unit, while using the Forget Gate to filter out irrelevant information, using the Input Gate to update important information, and generating prediction results for each step through the Output Gate. In order to further optimize the performance of the model, the Cross-Entropy Loss function or the Mean Squared Error Loss function is used during training, and the model parameters are adjusted in combination with the Backpropagation Through Time (BPTT) algorithm. Through multiple rounds of iterations, the model can effectively capture the difference between normal and abnormal breathing patterns.

[0041] After sufficient training and verification, the network parameters of the pre-trained LSTM model have been fixed and can be directly used to analyze the input respiratory feature data in real time. In actual applications, the quantified values ​​of respiratory rhythm instability and respiratory depth variation generated in the monitoring window are input into the LSTM model, and the model dynamically analyzes and outputs a respiratory evaluation coefficient based on its memorized respiratory pattern. The respiratory evaluation coefficient can quantitatively reflect the overall health status of the user's breathing.

[0042] The long short-term memory network is not specifically limited here, and can achieve the quantification of the unstable respiratory rhythm and respiratory depth variability quantification Comprehensive analysis to generate respiratory assessment coefficients In order to realize the technical solution of the present invention, the present invention provides a specific implementation method; the respiratory evaluation coefficient The resulting calculation formula is: , In the formula, , Respiratory rhythm instability quantification value and respiratory depth variability quantification The preset scaling factor of , Both are greater than 0.

[0043] It can be seen from the breathing assessment coefficient that, within the predetermined monitoring window, the greater the performance value of the breathing rhythm instability quantization value generated by analyzing and quantifying the variation amplitude of the extracted breathing rhythm, the greater the performance value of the breathing depth variation quantization value generated by analyzing and quantifying the fluctuation of the extracted breathing depth, that is, the greater the performance value of the breathing assessment coefficient generated when predicting the irregular breathing of the sleeper through the pre-trained long short-term memory network, the more obvious the degree of irregular breathing of the sleeper within the predetermined monitoring window, and vice versa.

[0044] The breathing assessment coefficient generated when predicting the irregular breathing of the sleeper through the pre-trained long short-term memory network is compared and analyzed with the pre-set breathing assessment coefficient reference threshold to determine whether the sleeper has irregular breathing. The specific steps are as follows: If the breathing assessment coefficient is greater than the breathing assessment coefficient reference threshold, a breathing irregularity signal is generated, and it is determined that the sleeper has irregular breathing; If the breathing assessment coefficient is less than or equal to the breathing assessment coefficient reference threshold, a normal breathing signal is generated, and it is determined that the sleeper does not have irregular breathing, indicating that the sleeper is breathing normally.

[0045] Once the LSTM network identifies that the sleeper has irregular breathing, it will use an incremental control algorithm to dynamically adjust the heartbeat interval acquisition frequency based on the prediction results. Specifically, it will increase the heartbeat interval acquisition frequency of the sleeper to capture subtle fluctuations in heart rate changes, obtain more detailed HRV data at a higher acquisition frequency, and improve the sensitivity and accuracy of detecting apnea events. Once the LSTM network identifies that the sleeper has irregular breathing, the incremental control algorithm will be used to dynamically adjust the heartbeat interval acquisition frequency based on the prediction results. The specific steps are as follows: Once the LSTM network identifies that the sleeper has irregular breathing, the adjustment factor of the acquisition frequency is first calculated based on the deviation between the breathing assessment coefficient and the reference threshold of the breathing assessment coefficient. The adjustment factor is used to indicate the degree to which the acquisition frequency needs to be increased. The calculation expression is: , in, is the frequency increment proportional constant, which is used to control the overall amplitude of the adjustment. It is a breathing evaluation coefficient generated by the long short-term memory network, reflecting the current degree of breathing abnormality. It is the reference threshold of the breathing assessment coefficient and defines the dividing line between normal breathing and abnormal breathing; The purpose of this step is to quantify the impact of respiratory abnormalities on the acquisition frequency and to amplify the frequency adjustment requirements of severe abnormalities through a nonlinear formula.

[0046] After calculating the adjustment factor, the heartbeat interval acquisition frequency is updated to increase the frequency to capture subtle changes in heart rate. The expression for updating the heartbeat interval acquisition frequency is: , in: The updated heartbeat interval acquisition frequency is used to enhance the sensitivity of data acquisition. is the initial heartbeat interval acquisition frequency, indicating the acquisition frequency under normal conditions; This step dynamically increases the acquisition frequency by superimposing an adjustment factor onto the initial frequency, allowing the device to record heart rate interval data at a higher resolution, thereby more accurately assessing subtle fluctuations in heart rate variability (HRV).

[0047] In order to avoid excessive frequency adjustment resulting in a sharp increase in energy consumption or data redundancy, a dynamic frequency range limit is set to ensure that the updated heartbeat interval collection frequency is adjusted within a controllable range. The limit formula is: , in, The final adjusted heartbeat interval acquisition frequency is used as the actual application frequency value. The upper limit of the heartbeat interval acquisition frequency is to ensure that the device will not be overloaded. It is the lower limit of the heartbeat interval acquisition frequency to ensure that the data accuracy meets the monitoring requirements; The purpose of this step is to ensure that the dynamic adjustment of the collection frequency can not only meet the needs of anomaly detection, but also avoid resource waste and equipment performance problems caused by excessively high frequency.

[0048] By dynamically increasing the frequency of heart rate interval acquisition, the device's ability to capture heart rate changes is enhanced, thereby more accurately recording and analyzing subtle fluctuations in heart rate variability (HRV). Especially when the long short-term memory network (LSTM) identifies irregular breathing in the sleeper, increasing the acquisition frequency can ensure that the device responds quickly to abnormalities and obtains high-resolution real-time data. This high-frequency data can not only significantly improve the sensitivity and accuracy of detecting potential health risks such as apnea events, but also provide more reliable support for subsequent evaluation and intervention, ultimately helping to protect the health of the sleeper and optimize the overall performance of the monitoring system.

[0049] The above scheme can effectively solve the problem in the prior art that it is difficult to capture tiny heart rate fluctuations due to the fixed acquisition frequency. Specifically, through the HRV-based sleep monitoring method, combined with real-time respiratory data and intelligent incremental control algorithm, when irregular breathing is detected, the heart rate interval acquisition frequency is dynamically adjusted to improve the sampling accuracy of heart rate changes. This method can not only capture subtle fluctuations when a sleeper has apnea or irregular breathing, but also improve the detection accuracy of apnea events with high sensitivity, and timely identify potential health risks such as hypoxemia, arrhythmia, etc., thereby providing more reliable data support for improving sleep quality and early intervention, thereby effectively avoiding health damage caused by missed detection.

[0050] The present invention provides Figure 2 The HRV-based sleep monitoring and analysis system shown includes an initial heartbeat interval acquisition module, a respiratory data acquisition module, a respiratory data preprocessing module, a respiratory irregularity feature extraction and analysis module, an irregular breathing recognition module, and an acquisition frequency adjustment module; The initial heartbeat interval acquisition module continuously records the time intervals of the sleeper's continuous heartbeats at a preset initial heartbeat interval acquisition frequency at the beginning of monitoring, so as to evaluate the state of the sleeper's autonomic nervous system at night; The breathing data collection module acquires the sleeping person's breathing data in real time while collecting heart rate data; The respiratory data preprocessing module preprocesses the acquired raw respiratory data to ensure the accuracy and consistency of the data; The breathing irregularity feature extraction and analysis module extracts key features reflecting breathing irregularity from the preprocessed breathing data, analyzes and quantifies the extracted key breathing features within a predetermined monitoring window, and preliminarily evaluates the degree of irregular breathing of the sleeper within each monitoring window; The irregular breathing recognition module inputs the quantified irregular breathing data into the pre-trained long short-term memory network for recognition to determine whether the sleeper has irregular breathing; The acquisition frequency adjustment module, once the long short-term memory network identifies that the sleeper has irregular breathing, will use the incremental control algorithm to dynamically adjust the acquisition frequency of the heartbeat interval according to the prediction results. Specifically, it increases the acquisition frequency of the sleeper's heartbeat interval to capture subtle fluctuations in heart rate changes and obtain more detailed HRV data at a higher acquisition frequency; The HRV-based sleep monitoring and analysis method provided in the embodiment of the present invention is implemented by the above-mentioned HRV-based sleep monitoring and analysis system. The specific method and process of the HRV-based sleep monitoring and analysis system are detailed in the embodiment of the above-mentioned HRV-based sleep monitoring and analysis method, which will not be repeated here.

[0051] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0052] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0053] The above description is only by way of illustration of certain exemplary embodiments of the present invention. It is undoubted that those skilled in the art can modify the described embodiments in various ways without departing from the spirit and scope of the present invention. Therefore, the above drawings and descriptions are illustrative in nature and should not be construed as limiting the scope of protection of the claims of the present invention.

Claims

1. A sleep monitoring and analysis method based on HRV, characterized in that: The following steps are involved: At the beginning of monitoring, the continuous heartbeat time intervals of the sleeper are continuously recorded at a preset initial heartbeat interval collection frequency to evaluate the sleeper's autonomic nervous system state at night; Acquire the sleeper's breathing data in real time while collecting heart rate data, and pre-process the acquired raw breathing data to ensure the accuracy and consistency of the data; Extract key features reflecting irregular breathing from the preprocessed breathing data, analyze and quantify the extracted key breathing features within a predetermined monitoring window, and preliminarily evaluate the degree of irregular breathing of the sleeper within each monitoring window; The quantified irregular breathing data is input into the pre-trained long short-term memory network for recognition to determine whether the sleeper has irregular breathing; Once the long short-term memory network identifies that the sleeper has irregular breathing, the incremental control algorithm will be used to dynamically adjust the heart rate interval collection frequency based on the prediction results. Specifically, the frequency of heart rate interval collection of the sleeper will be increased to capture subtle fluctuations in heart rate changes and obtain more detailed HRV data at a higher collection frequency.

2. The HRV-based sleep monitoring and analysis method according to claim 1, characterized in that: The setting of the initial heartbeat interval collection frequency strikes a balance between data accuracy and energy efficiency. The specific setting is 1Hz to 4Hz, which not only meets the requirements of sleep state assessment, but also reduces device power consumption and prolongs battery life.

3. The sleep monitoring and analysis method based on HRV according to claim 1, characterized in that: Key features reflecting irregular breathing are extracted from the preprocessed respiratory data, wherein the extracted features include the variation amplitude of respiratory rhythm and the fluctuation of respiratory depth. The extracted variation amplitude of respiratory rhythm and the fluctuation of respiratory depth are analyzed and quantified within a predetermined monitoring window, and a respiratory rhythm instability quantification value and a respiratory depth variation quantification value are generated respectively. The degree of irregular breathing of the sleeper within each monitoring window is preliminarily evaluated through the respiratory rhythm instability quantification value and the respiratory depth variation quantification value.

4. The HRV-based sleep monitoring and analysis method according to claim 3, characterized in that: The respiratory rhythm instability quantification value and the respiratory depth variation quantification value generated by quantifying the amplitude of the respiratory rhythm change and the fluctuation of the respiratory depth are input into the pre-trained long short-term memory network, and the respiratory evaluation coefficient generated by the long short-term memory network is used to determine whether the sleeper has irregular breathing.

5. The HRV-based sleep monitoring and analysis method according to claim 4, characterized in that: The breathing assessment coefficient generated when predicting the irregular breathing of the sleeper through the pre-trained long short-term memory network is compared and analyzed with the pre-set breathing assessment coefficient reference threshold to determine whether the sleeper has irregular breathing. The specific steps are as follows: If the breathing assessment coefficient is greater than the breathing assessment coefficient reference threshold, a breathing irregularity signal is generated, and it is determined that the sleeper has irregular breathing; If the breathing assessment coefficient is less than or equal to the breathing assessment coefficient reference threshold, a normal breathing signal is generated, and it is determined that the sleeper does not have irregular breathing, which indicates that the sleeper is breathing normally.

6. The HRV-based sleep monitoring and analysis method according to claim 3, characterized in that: The specific steps of analyzing and quantifying the variation amplitude of the extracted respiratory rhythm within the predetermined monitoring window to generate the respiratory rhythm instability quantification value are as follows: First, extract the respiratory cycle sequence within each monitoring window from the preprocessed respiratory data , ,in Indicates i The duration of the breathing cycle is used to calculate the instantaneous fluctuation rate between each two consecutive cycles. The calculation expression is: ， in: For the The duration of the breathing cycle, is a normalization factor used to eliminate the influence of absolute period differences, is the weighting factor, and the weight increases with the time distance from the monitoring window center The distance changes exponentially. To adjust the weight sensitivity parameter, It is i The instantaneous fluctuation rate of each breath, n Indicates the total number of breathing cycles in the current monitoring window; Constructing instantaneous volatility series through instantaneous volatility , , the quantitative value of respiratory rhythm instability is calculated based on the instantaneous fluctuation rate sequence, and the calculation expression is: ， in: It is a quantitative value for the instability of respiratory rhythm. is a dynamic adjustment factor that emphasizes the contribution of fluctuations at the tail of the monitoring window. To monitor the window position weight parameter, adjust the impact of the tail data, is the cumulative value of the fluctuation rate of the respiratory cycle at the beginning and end of the monitoring window, is the cumulative value of the respiratory cycle fluctuation rate in the central area of ​​the monitoring window, is the edge weight adjustment factor, controlling the contribution of the head and tail fluctuations, A small constant to prevent the denominator from being zero.

7. The sleep monitoring and analysis method based on HRV according to claim 3, characterized in that: The specific steps for analyzing and quantifying the fluctuation of the extracted breathing depth within the predetermined monitoring window to generate the breathing depth variation quantitative value are as follows: In the predetermined monitoring window, the breathing depth of each breathing cycle is extracted from the preprocessed breathing data to represent the amount of gas inhaled in each breath. The time-weighted mean is calculated using the breathing depth. The calculation expression is: ， in: It is a time-weighted mean that smoothes fluctuations in breathing depth, eliminates interference from short-term noise, and provides a stable benchmark value for subsequent calculations. is the length of the monitoring window, is the breathing depth at time t, is a time-weighted function that reflects the importance of breathing depth over time; Based on the extracted respiratory depth data and time-weighted mean, the depth deviation at each time point is further calculated, and the degree of variation is quantified to generate a quantitative value of respiratory depth variation. The generated expression is: ， in: is the deviation between the breathing depth and the time-weighted mean, reflecting the instantaneous fluctuation of the breathing depth. is the normalized deviation, which is used to eliminate the influence of different users or breathing depth baseline differences. Quantify the variability of breathing depth.

8. The sleep monitoring and analysis method based on HRV according to claim 5, characterized in that: Once the LSTM network identifies that the sleeper has irregular breathing, the incremental control algorithm will be used to dynamically adjust the heartbeat interval acquisition frequency based on the prediction results. The specific steps are as follows: Once the LSTM network identifies that the sleeper has irregular breathing, the adjustment factor of the acquisition frequency is first calculated based on the deviation between the breathing assessment coefficient and the reference threshold of the breathing assessment coefficient. The adjustment factor is used to indicate the degree to which the acquisition frequency needs to be increased. The calculation expression is: ， in, is the frequency increment proportional constant, which is used to control the overall amplitude of the adjustment. It is a breathing evaluation coefficient generated by the long short-term memory network, reflecting the current degree of breathing abnormality. It is the reference threshold of the breathing assessment coefficient and defines the dividing line between normal breathing and abnormal breathing; After calculating the adjustment factor, the heartbeat interval acquisition frequency is updated to increase the frequency to capture subtle changes in heart rate. The expression for updating the heartbeat interval acquisition frequency is: ， in: The updated heartbeat interval acquisition frequency is used to enhance the sensitivity of data acquisition. is the initial heartbeat interval acquisition frequency, indicating the acquisition frequency under normal conditions; In order to avoid excessive frequency adjustment resulting in a sharp increase in energy consumption or data redundancy, a dynamic frequency range limit is set to ensure that the updated heartbeat interval collection frequency is adjusted within a controllable range. The limit formula is: ， in, The final adjusted heartbeat interval acquisition frequency is used as the actual application frequency value. The upper limit of the heartbeat interval acquisition frequency is to ensure that it does not run overloaded. It is the lower limit of the heartbeat interval collection frequency to ensure that the data accuracy meets the monitoring requirements.

9. A sleep monitoring and analysis system based on HRV, used to implement the sleep monitoring and analysis method based on HRV as described in any one of claims 1 to 8, characterized in that: It includes an initial heartbeat interval acquisition module, a respiratory data acquisition module, a respiratory data preprocessing module, a respiratory irregularity feature extraction and analysis module, an irregular respiratory recognition module, and an acquisition frequency adjustment module; The initial heartbeat interval acquisition module continuously records the time intervals of the sleeper's continuous heartbeats at a preset initial heartbeat interval acquisition frequency at the beginning of monitoring, so as to evaluate the state of the sleeper's autonomic nervous system at night; The breathing data collection module acquires the sleeping person's breathing data in real time while collecting heart rate data; The respiratory data preprocessing module preprocesses the acquired raw respiratory data to ensure the accuracy and consistency of the data; The breathing irregularity feature extraction and analysis module extracts key features reflecting breathing irregularity from the preprocessed breathing data, analyzes and quantifies the extracted key breathing features within a predetermined monitoring window, and preliminarily evaluates the degree of irregular breathing of the sleeper within each monitoring window; The irregular breathing recognition module inputs the quantified irregular breathing data into the pre-trained long short-term memory network for recognition to determine whether the sleeper has irregular breathing; The acquisition frequency adjustment module, once the long short-term memory network recognizes that the sleeper has irregular breathing, will use the incremental control algorithm to dynamically adjust the acquisition frequency of the heartbeat intervals based on the prediction results. Specifically, it will increase the acquisition frequency of the sleeper's heartbeat intervals to capture subtle fluctuations in heart rate changes and obtain more detailed HRV data at a higher acquisition frequency.

Images