A multi-device multi-sensor sleep monitoring system

Abstract

The present invention discloses a multi-device multi-sensor sleep monitoring system, including: a data acquisition module and a data processing module, wherein the data acquisition module includes a bracelet acquisition unit, a radar acquisition unit and a sound acquisition unit, wherein the bracelet acquisition unit collects first data of a target including first body movement data, blood oxygen data, heart rate data and first respiratory data in real time. The multi-device multi-sensor sleep monitoring system disclosed by the present invention fully utilizes the advantages of various devices and various sensors, extracts relatively accurate original data detected by various devices and sensors for sleep analysis, evaluates the accuracy of the same type of data, assigns different weights to the same type of data, and uses different types of data to analyze and judge the sleep state according to their respective characteristics, thereby increasing the accuracy of the original data while increasing the amount of the original detection data, thereby improving the accuracy of the sleep detection analysis.

Description

Technical Field

[0001] The present invention belongs to the technical field of sleep monitoring, and in particular relates to a multi-device multi-sensor sleep monitoring system. Background Art

[0002] The current daily sleep monitoring relies on a single monitoring device. For example, wristband monitoring uses the built-in body motion sensor, optical heart rate sensor and optical blood oxygen sensor to record the body motion information and vital signs information throughout the night and analyze the sleep status; mobile phone sleep monitoring apps usually use their built-in body motion sensors and microphones to record the body motion and sound conditions of the whole night's sleep, and analyze the above information to obtain the whole night's sleep status analysis; radar sleep monitoring can monitor human body movement, breathing and heart rate without contact and at a distance, and use this information to analyze sleep. This detection method avoids the discomfort caused by contact between the detection equipment and the body and does not affect people's daily sleep; in addition, there are also smart mattresses, cameras and other sleep monitoring devices available for families.

[0003] The above sleep monitoring devices all use a variety of data during the monitoring process, and conduct a comprehensive analysis to obtain the sleep status analysis results. However, it is difficult for a single monitoring device to accurately monitor all kinds of data. Limited by issues such as usage, installation location, and customer experience, daily sleep monitoring equipment cannot be like professional sleep monitoring equipment, using the most suitable sensors and the most suitable usage methods to comprehensively collect various physiological signals, such as polysomnography systems. Currently, common sleep monitoring devices have a limited number of sensors, a small number of types of original signals collected, and varying signal quality.

[0004] The sleep analysis algorithm of a bracelet usually uses information such as body movement, breathing, and heart rate. Hand movement and heart rate can be directly acquired by the built-in body movement sensor and PPG sensor of the bracelet. Because they are worn close to the body, these two types of data are generally considered to be more accurate. The breathing signal is generally derived indirectly from the PPG signal or body movement signal, and its accuracy is greatly affected by the detection conditions and algorithms, thus affecting the analysis of sleep status.

[0005] Mobile sleep monitoring apps usually require the phone to be placed close to the person being tested, infer human activity by detecting the vibration of the mattress, and record sleep sounds through a microphone. However, the vibration of the mattress is related to the body shape of the person being tested and the structural material of the mattress, so the body movement information obtained by this method is inaccurate. Common radar sleep monitoring equipment also uses body movement, breathing and heart rate information to analyze sleep status. The principle of detecting breathing and heart rate is based on the tiny movements of the body driven by human breathing and heartbeat. The radar captures these tiny movements and performs further analysis and processing to extract the person's breathing and heartbeat information. The movement of the body driven by the heartbeat is very weak, and the radar irradiation position will have a greater impact on the detection results. In actual sleep, the position and posture of a person are random, so it is difficult to guarantee the accuracy of the heart rate data.

[0006] Other sleep monitoring devices also have similar problems, some data can be detected more accurately, while other data detection accuracy is relatively low. These low-accuracy raw data have become the shortcomings of sleep analysis, limiting the accuracy of sleep analysis.

[0007] Therefore, further improvements are made to the above problems. Summary of the invention

[0008] The main purpose of the present invention is to provide a multi-device multi-sensor sleep monitoring system, which fully utilizes the advantages of various devices and various sensors, extracts relatively accurate original data detected by each device and sensor for sleep analysis, and assigns different weights to the same type of data based on its accuracy evaluation, and different types of data are used to analyze and judge the sleep state according to their respective characteristics. In this way, while increasing the amount of original detection data, the accuracy of the original data is also improved, thereby improving the accuracy of sleep detection and analysis.

[0009] To achieve the above objectives, the present invention provides a multi-device multi-sensor sleep monitoring system for comprehensively analyzing the sleep status of a target (person), including: a data acquisition module and a data processing module, wherein the data acquisition module includes a bracelet acquisition unit, a radar acquisition unit and a sound acquisition unit, wherein:

[0010] The wristband collection unit collects first data of the target including first body motion data, blood oxygen data, heart rate data and first breathing data in real time, and transmits the obtained first data to the data processing module in real time;

[0011] The radar acquisition unit acquires second data of the target including second body motion data and second breathing data in real time, and transmits the acquired second data to the data processing module in real time;

[0012] The sound collection unit collects the sound data of the target in real time and transmits the obtained sound data to the data processing module in real time;

[0013] The data processing module performs instant processing and comprehensive processing on the data after obtaining the first data, the second data and the third data respectively, thereby outputting an instant sleep analysis report and a comprehensive sleep analysis report respectively.

[0014] As a further preferred technical solution of the above technical solution, for real-time processing, the data processing module performs real-time analysis on the first data, the second data and the third data, thereby making a timely response display or alarm, wherein:

[0015] For the blood oxygen data, the data processing module analyzes the blood oxygen data of the first data. If the current blood oxygen data is within the normal blood oxygen range, the data processing module outputs a target normal blood oxygen signal. If the current blood oxygen data is lower than the first blood oxygen threshold (preferably lower than 95%), the data processing module outputs a target hypoxia signal. If the current blood oxygen data is lower than the second blood oxygen threshold (preferably lower than 80%), the data processing module outputs a target hypoxia danger signal and outputs a blood oxygen alarm signal to the target (through sound, light or vibration alarm, etc.) or a third party (relatives, etc.);

[0016] For the heart rate data, the data processing module analyzes the heart rate data of the first data. If the current heart rate data is within the normal heart rate range, the data processing module outputs a target normal heart rate signal. If the current heart rate data is lower than the first heart rate threshold (preferably lower than 30 bps), the data processing module outputs a target heart rate bradycardia or stop danger signal and outputs a heart rate alarm signal to the target person (through sound, light or vibration alarm, etc.) or a third party (relatives, etc.);

[0017] For the body motion data, the data processing module analyzes the first body motion data of the first data and the second body motion data of the second data to obtain various sleep states of the target, wherein:

[0018] For bedtime, there is a target within the monitoring range of the radar acquisition unit;

[0019] For the sleep start time, starting from the time of going to bed, if the number of micro-movements and large movements detected in a unit time (e.g., within 10 minutes) is lower than a certain threshold nth1, then sleep is considered to have started;

[0020] For the end time of sleep, starting from the start time of sleep, if the number of micro-movements or large movements detected in a unit time (such as within 10 minutes) is higher than a certain threshold nth2, then a sleep period is considered to have ended;

[0021] For the time of leaving the bed, whether there is a target within the monitoring range of the radar acquisition unit;

[0022] During the sleep from the start time to the end time of sleep, the number of subtle movements per unit time detected by the wristband collection unit and the radar collection unit, n1 and n2, are counted. When n1*α+n2*(1-α) is lower than a certain threshold nth3, the current sleep state is considered to be deep sleep and marked (otherwise it is shallow sleep), where α is the weight of the first body motion data in signal evaluation, ranging from greater than 0 to less than 1. The larger the value, the greater the proportion of the first body motion data, and the greater the impact on the judgment of sleep depth;

[0023] For the breathing data, the data processing module analyzes the first breathing data of the first data and the second breathing data of the second data to obtain various breathing states of the target, wherein:

[0024] The data processing module gives priority to the second breathing data, checks the signal quality of the first breathing data and the second breathing data in real time, switches the breathing data in a timely manner, takes the breathing data with high signal quality as analysis, takes the second breathing data within a unit time, performs frequency domain transformation to obtain spectrum information, the frequency corresponding to the peak point of the spectrum is recorded as f0, the spectrum energy within (f0-delat_f, f0+delta_f) is recorded as P0, and the sum of all energies of the entire spectrum is recorded as Ps. When P0 / (Ps-P0) is lower than a certain threshold, it is determined that the second breathing data collected by the radar acquisition unit is of poor quality, and the data is switched to the first breathing data collected by the bracelet acquisition unit. If the signal peak of the first breathing data collected by the bracelet acquisition unit is lower than a certain threshold, it is determined that the first breathing data is too weak and the data reliability is low, thereby switching to the second breathing data;

[0025] For the sound data, the data processing module analyzes the sound data of the third data, and calculates the frequency, intensity and peak value change of the sound data per unit time in real time. If the sound fluctuation is uniform, the corresponding peak is periodic and the intensity change is small, it indicates that the target is just snoring, and the snoring is marked at the corresponding time. If the sound fluctuation is large, the corresponding peak interval difference is large, and the peak value difference is large, the snoring needs to be marked and a snoring alarm signal is output to the target person.

[0026] Through various data, the data processing module outputs an instant sleep analysis report.

[0027] As a further preferred technical solution of the above technical solution, for comprehensive processing, the data processing module performs comprehensive analysis on the first data, the second data and the third data, thereby making a breathing quality assessment and a sleep quality analysis, wherein:

[0028] For respiratory quality assessment, when the respiratory waveform shows obvious fluctuations, the frequency or intensity shows a large change compared with the historical data, indicating that there is respiratory disorder or apnea. Through the whole night data monitoring, the number of apnea is counted. The more apnea and respiratory disorder occur, the worse the respiratory quality is. The respiratory quality score is set as Soc_resp, with a full score of 100 points. Each time there is abnormal breathing, a certain number of points will be deducted. The severity of apnea is comprehensively evaluated by sound waveform characteristics and blood oxygen data. At the same time, the sound data is compared, and the time period when apnea and snoring marks appear at the same time is checked and marked, and the number of times is counted;

[0029] For sleep quality analysis, sleep quality is related to the proportion of deep sleep, the number of sleep interruptions and the number of tossing and turning. The comprehensive score of sleep quality is calculated. The data used are as follows:

[0030] Deep sleep ratio a1: Count the sleep data of the whole night and calculate the ratio of deep sleep time to total sleep time, which is recorded as deep sleep ratio a1. a1 is a value greater than 0 and less than 1.

[0031] Turning over number a2: Turning over is a large action that is continuous in a short time. The turning over action is accurately judged and marked through the first body motion data and the second body motion data;

[0032] Awakening times a3: The more awakening times you have at night, the worse your sleep is;

[0033] Breathing quality score a4: If the breathing quality assessment score is low, it indicates that the breathing quality during sleep is poor;

[0034] Based on the above four data, the sleep score Soc_sleep is calculated as follows:

[0035] Soc_sleep=SocA1+SocA2+SocA3+SocA4;

[0036] SocA1=100*α1*(0.5+a1);

[0037] SocA2=100*α2*cur(a2);

[0038] SocA3=100*α3*(1-0.2*a3);

[0039] SocA4=100*α4*a4;

[0040] Among them, cur is a centrally symmetrical distribution curve (the center is at x=25, and the peak is 1). The greater the deviation from the center, the smaller the value of cur(x). α1, α2, α3 and α4 are the weights of each parameter in sleep assessment. The larger the weight, the greater the impact of this parameter on sleep quality assessment. Finally, the respiratory quality assessment and sleep quality analysis are combined to output a comprehensive sleep analysis report.

[0041] As a further preferred technical solution of the above technical solution, before monitoring, the data acquisition module and the data processing module are self-checked respectively to ensure that the data acquisition module and the data processing module work normally within the monitoring period.

[0042] As a further preferred technical solution of the above technical solution, the data processing module transmits the timely sleep analysis report and the comprehensive sleep analysis report to the device terminal (including mobile phones, etc.). BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a structural schematic diagram of a multi-device multi-sensor sleep monitoring system of the present invention.

[0044] Figure 2 It is a schematic diagram of the cur(x) parameter in the sleep quality analysis of the multi-device multi-sensor sleep monitoring system of the present invention. DETAILED DESCRIPTION

[0045] The following description is used to disclose the present invention so that those skilled in the art can implement the present invention. The preferred embodiments described below are only examples, and those skilled in the art can think of other obvious variations. The basic principles of the present invention defined in the following description can be applied to other embodiments, variations, improvements, equivalents, and other technical solutions that do not deviate from the spirit and scope of the present invention.

[0046] In the preferred embodiments of the present invention, those skilled in the art should note that the objects and the like involved in the present invention may be regarded as prior art.

[0047] Preferred embodiments.

[0048] The present invention discloses a multi-device multi-sensor sleep monitoring system for comprehensively analyzing the sleep status of a target (person), including: a data acquisition module and a data processing module, wherein the data acquisition module includes a bracelet acquisition unit, a radar acquisition unit and a sound acquisition unit, wherein:

[0049] The wristband collection unit collects first data of the target including first body motion data, blood oxygen data, heart rate data and first breathing data in real time, and transmits the obtained first data to the data processing module in real time;

[0050] The radar acquisition unit acquires second data of the target including second body motion data and second breathing data in real time, and transmits the acquired second data to the data processing module in real time;

[0051] The sound collection unit collects the sound data of the target in real time and transmits the obtained sound data to the data processing module in real time;

[0052] The data processing module performs instant processing and comprehensive processing on the data after obtaining the first data, the second data and the third data respectively, thereby outputting an instant sleep analysis report and a comprehensive sleep analysis report respectively.

[0053] Specifically, for real-time processing, the data processing module performs real-time analysis on the first data, the second data, and the third data, thereby making a timely response display or alarm, wherein:

[0054] For the blood oxygen data, the data processing module analyzes the blood oxygen data of the first data. If the current blood oxygen data is in the normal blood oxygen range (95% to 100%), the data processing module outputs a target normal blood oxygen signal. If the current blood oxygen data is lower than the first blood oxygen threshold (preferably lower than 95%), the data processing module outputs a target hypoxia signal. If the current blood oxygen data is lower than the second blood oxygen threshold (preferably lower than 80%), the data processing module outputs a target hypoxia danger signal and outputs a blood oxygen alarm signal to the target (through sound, light or vibration alarm, etc.) or a third party (relatives, etc.);

[0055] For the heart rate data, the data processing module analyzes the heart rate data of the first data. If the current heart rate data is within the normal heart rate range (60-100 bps), the data processing module outputs a target normal heart rate signal. If the current heart rate data is lower than the first heart rate threshold (preferably lower than 30 bps), the data processing module outputs a target heart rate bradycardia or stop danger signal and outputs a heart rate alarm signal to the target person (through sound, light or vibration alarm, etc.) or a third party (relatives, etc.);

[0056] The blood oxygen and heart rate data are both derived from the bracelet collection unit. However, the detection effect of the bracelet collection unit is related to whether it fits the wristband well. If the heart rate and blood oxygen data uploaded by the bracelet are too low or zero, and the radar collection unit indicates that someone is present and the breathing data is normal, it is very likely that the bracelet detection has failed, and its uploaded data can no longer be used for sleep analysis.

[0057] For the body motion data, the data processing module analyzes the first body motion data of the first data and the second body motion data of the second data to obtain various sleep states of the target, wherein:

[0058] It shows in real time whether the target is in bed and whether he is asleep. The first body motion data in the bracelet can only be used to detect the movements of the hands, while the radar has a wider detection range and can detect the movements of the whole body. Therefore, it can more accurately reflect when the tester goes to bed, gets out of bed, and starts and ends sleep.

[0059] For bedtime, there is a target within the monitoring range of the radar acquisition unit;

[0060] For the sleep start time, starting from the time of going to bed, if the number of micro-movements and large movements detected in a unit time (e.g., within 10 minutes) is lower than a certain threshold nth1, then sleep is considered to have started;

[0061] For the end time of sleep, starting from the start time of sleep, if the number of micro-movements or large movements detected in a unit time (such as within 10 minutes) is higher than a certain threshold nth2, then a sleep period is considered to have ended;

[0062] For the time of leaving the bed, whether there is a target within the monitoring range of the radar acquisition unit;

[0063] During the sleep from the start time to the end time of sleep, the number of subtle movements per unit time detected by the wristband collection unit and the radar collection unit, n1 and n2, are counted. When n1*α+n2*(1-α) is lower than a certain threshold nth3, the current sleep state is considered to be deep sleep and marked (otherwise it is shallow sleep), where α is the weight of the first body motion data in signal evaluation, ranging from greater than 0 to less than 1. The larger the value, the greater the proportion of the first body motion data, and the greater the impact on the judgment of sleep depth;

[0064] For the breathing data, the data processing module analyzes the first breathing data of the first data (the PPG signal collected by the bracelet collection unit can indirectly derive the breathing data, and its signal quality is related to the original signal quality of the PPG) and the second breathing data of the second data to obtain various breathing states of the target, where:

[0065] The respiratory data collected by the radar is closely related to the position of the human body relative to the radar. When the human chest is facing the radar detection surface, clear respiratory data with good signal-to-noise ratio can be obtained. When the human chest is far away from the radar detection surface or facing away from the radar, the respiratory data quality is poor.

[0066] The data processing module gives priority to the second breathing data, checks the signal quality of the first breathing data and the second breathing data in real time, switches the breathing data in a timely manner, takes the breathing data with high signal quality as analysis, takes the second breathing data within a unit time, performs frequency domain transformation to obtain spectrum information, the frequency corresponding to the peak point of the spectrum is recorded as f0, the spectrum energy within (f0-delat_f, f0+delta_f) is recorded as P0, and the sum of all energies of the entire spectrum is recorded as Ps. When P0 / (Ps-P0) is lower than a certain threshold, it is determined that the second breathing data collected by the radar acquisition unit is of poor quality, and the data is switched to the first breathing data collected by the bracelet acquisition unit. If the signal peak of the first breathing data collected by the bracelet acquisition unit is lower than a certain threshold, it is determined that the first breathing data is too weak and the data reliability is low, thereby switching to the second breathing data;

[0067] For the sound data, the data processing module analyzes the sound data of the third data, and calculates the frequency, intensity and peak value change of the sound data per unit time in real time. If the sound fluctuation is uniform, the corresponding peak is periodic and the intensity change is small, it indicates that the target is just snoring, and the snoring is marked at the corresponding time. If the sound fluctuation is large, the corresponding peak interval difference is large, and the peak value difference is large, the snoring needs to be marked and a snoring alarm signal is output to the target person.

[0068] Through various data, the data processing module outputs an instant sleep analysis report.

[0069] Furthermore, for comprehensive processing, the data processing module performs comprehensive analysis on the first data, the second data and the third data, thereby making a breathing quality assessment and a sleep quality analysis, wherein:

[0070] For respiratory quality assessment, when the respiratory waveform shows obvious fluctuations, the frequency or intensity shows a large change compared with the historical data, indicating that there is respiratory disorder or apnea. Through the whole night data monitoring, the number of apnea is counted. The more apnea and respiratory disorder occur, the worse the respiratory quality is. The respiratory quality score is set as Soc_resp, with a full score of 100 points. Each time there is abnormal breathing, a certain number of points will be deducted. The severity of apnea is comprehensively evaluated by sound waveform characteristics and blood oxygen data. At the same time, the sound data is compared, and the time period when apnea and snoring marks appear at the same time is checked and marked, and the number of times is counted;

[0071] For sleep quality analysis, sleep quality is related to the proportion of deep sleep, the number of sleep interruptions and the number of tossing and turning. The comprehensive score of sleep quality is calculated. The data used are as follows:

[0072] Deep sleep ratio a1: Count the sleep data of the whole night and calculate the ratio of deep sleep time to total sleep time, which is recorded as deep sleep ratio a1. a1 is a value greater than 0 and less than 1.

[0073] Number of turning over a2: Turning over is a large continuous movement in a short period of time. The turning over movement is accurately judged and marked through the first body motion data and the second body motion data. During normal sleep, people will turn over, usually about 20 to 30 times. Too many or too few turning over times indicate poor sleep quality.

[0074] Awakening times a3: The more awakening times you have at night, the worse your sleep is;

[0075] Breathing quality score a4: If the breathing quality assessment score is low, it indicates that the breathing quality during sleep is poor;

[0076] Based on the above four data, the sleep score Soc_sleep is calculated as follows:

[0077] Soc_sleep=SocA1+SocA2+SocA3+SocA4;

[0078] SocA1=100*α1*(0.5+a1);

[0079] SocA2=100*α2*cur(a2);

[0080] SocA3=100*α3*(1-0.2*a3);

[0081] SocA4=100*α4*a4;

[0082] Where cur is a centrosymmetric distribution curve (centered at x=25, with a peak value of 1), such as Figure 2 As shown, the greater the deviation from the center, the smaller the value of cur(x). α1, α2, α3 and α4 are the weights of each parameter in the sleep assessment. The larger the weight, the greater the impact of this parameter on the sleep quality assessment. Finally, the breathing quality assessment and sleep quality analysis are combined to output a comprehensive sleep analysis report.

[0083] Furthermore, before monitoring, the data acquisition module and the data processing module are respectively self-checked to ensure that the data acquisition module and the data processing module work normally within the monitoring period.

[0084] Furthermore, the data processing module transmits the timely sleep analysis report and the comprehensive sleep analysis report to the device terminal (including mobile phones, etc.).

[0085] Preferably, the present invention is not only applicable to the case of single-person detection, but also can be used in the case of multiple-person detection. The specific implementation methods can refer to the following two:

[0086] One is to increase the number of sensors, arrange them reasonably within the detection range, limit the detection range, and distinguish multi-person scenes spatially.

[0087] Another method is to use existing sensors to directly detect the information of multiple people, and then match the signals detected by different devices with the person to be tested (target) through signal correlation analysis, and then analyze the sleep state according to the above sleep detection and analysis method (data processing module for data processing and analysis). When the bracelet acquisition unit is used, the tester (target) needs to wear it close to the body for testing, so the bracelet detection data (first data) has a clear one-to-one relationship with the tester, which can be used as a benchmark to distinguish different testers; the millimeter wave radar (radar acquisition unit) itself can achieve high-precision, high-resolution, multi-target detection, and can effectively distinguish the movement and breathing conditions of multiple targets in the detection area; the microphone (sound acquisition unit) can simultaneously collect the sound information of multiple people in the environment, and separate the sound information of different testers through voice recognition technology such as voiceprint recognition. For the same tester, the breathing data obtained by the bracelet acquisition unit and the radar acquisition unit, and the snoring signal (sound data) analyzed by the sound acquisition unit should have a high correlation. Therefore, through signal correlation analysis, the detection signals of different testers separated by each acquisition unit can be matched and grouped to obtain the detection data set of each tester. By performing the above sleep analysis processing on each data set, the sleep analysis situation of each tester can be obtained.

[0088] It is worth mentioning that the technical features such as the objectives involved in the patent application of this invention should be regarded as the prior art. The specific structure, working principle and possible control method and spatial layout method of these technical features can be selected by conventional methods in the field, and should not be regarded as the inventive point of this patent. This patent will not be further elaborated.

[0089] For those skilled in the art, it is still possible to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. A multi-device multi-sensor sleep monitoring system for comprehensively analyzing the target's sleep status, characterized in that: include: Data collection module and data processing module, the data collection module includes a bracelet collection unit, a radar collection unit and a sound collection unit, wherein: The wristband collection unit collects first data of the target including first body motion data, blood oxygen data, heart rate data and first breathing data in real time, and transmits the obtained first data to the data processing module in real time; The radar acquisition unit acquires second data of the target including second body motion data and second breathing data in real time, and transmits the acquired second data to the data processing module in real time; The sound collection unit collects sound data of the target in real time and transmits the obtained sound data to the data processing module in real time, the sound data being the third data; The data processing module performs instant processing and comprehensive processing on the first data, the second data and the third data after obtaining the first data, the second data and the third data, respectively, so as to output an instant sleep analysis report and a comprehensive sleep analysis report respectively; For real-time processing, the data processing module performs real-time analysis on the first data, the second data, and the third data, thereby making a timely response display or alarm, wherein: For the blood oxygen data, the data processing module analyzes the blood oxygen data of the first data. If the current blood oxygen data is within the normal blood oxygen range, the data processing module outputs a target normal blood oxygen signal. If the current blood oxygen data is lower than the first blood oxygen threshold, the data processing module outputs a target hypoxia signal. If the current blood oxygen data is lower than the second blood oxygen threshold, the data processing module outputs a target hypoxia danger signal and sends a blood oxygen alarm signal to the target or a third party. For the heart rate data, the data processing module analyzes the heart rate data of the first data, and if the current heart rate data is within the normal heart rate range, the data processing module outputs a target normal heart rate signal; if the current heart rate data is lower than the first heart rate threshold, the data processing module outputs a target heart rate bradycardia or stop danger signal and outputs a heart rate alarm signal to the target person or a third party; For the body motion data, the data processing module analyzes the first body motion data of the first data and the second body motion data of the second data to obtain various sleep states of the target, wherein: For bedtime, there is a target within the monitoring range of the radar acquisition unit; For the sleep start time, starting from the time of going to bed, if the number of micro-movements and large movements detected per unit time is lower than a certain threshold nth1, then sleep is considered to have started; For the end time of sleep, starting from the start time of sleep, if the number of micro-movements or large movements detected in a unit time is higher than a certain threshold nth2, then a sleep period is considered to have ended; For the time of leaving the bed, whether there is a target within the monitoring range of the radar acquisition unit; During the sleep from the start time to the end time of sleep, the number of subtle movements per unit time detected by the wristband collection unit and the radar collection unit is counted n1 and n2. When n1*α+n2*(1-α) is lower than a certain threshold nth3, the current sleep state is considered to be deep sleep and marked, where α is the weight of the first body motion data in signal evaluation, ranging from greater than 0 to less than 1. The larger the value, the greater the proportion of the first body motion data, and the greater the impact on the judgment of sleep depth; For the breathing data, the data processing module analyzes the first breathing data of the first data and the second breathing data of the second data to obtain various breathing states of the target, wherein: The data processing module gives priority to the second breathing data, checks the signal quality of the first breathing data and the second breathing data in real time, switches the breathing data in a timely manner, takes the breathing data with high signal quality as analysis, takes the second breathing data within a unit time, performs frequency domain transformation to obtain spectrum information, the frequency corresponding to the peak point of the spectrum is recorded as f0, the spectrum energy within (f0-delat_f, f0+delta_f) is recorded as P0, and the sum of all energies of the entire spectrum is recorded as Ps. When P0 / (Ps-P0) is lower than a certain threshold, it is determined that the second breathing data collected by the radar acquisition unit is of poor quality, and the data is switched to the first breathing data collected by the bracelet acquisition unit. If the signal peak of the first breathing data collected by the bracelet acquisition unit is lower than a certain threshold, it is determined that the first breathing data is too weak and the data reliability is low, thereby switching to the second breathing data; For the sound data, the data processing module analyzes the sound data of the third data, and calculates the frequency, intensity and peak value change of the sound data per unit time in real time. If the sound fluctuation is uniform, the corresponding peak is periodic and the intensity change is small, it indicates that the target is just snoring, and the snoring is marked at the corresponding time. If the sound fluctuation is large, the corresponding peak interval difference is large, and the peak value difference is large, the snoring needs to be marked and a snoring alarm signal is output to the target person. Through various data, the data processing module outputs an instant sleep analysis report.

2. A multi-device multi-sensor sleep monitoring system according to claim 1, characterized in that: For comprehensive processing, the data processing module performs comprehensive analysis on the first data, the second data and the third data, thereby making a breathing quality assessment and a sleep quality analysis, wherein: For respiratory quality assessment, when the respiratory waveform shows obvious fluctuations, the frequency or intensity shows a large change compared with the historical data, indicating that there is respiratory disorder or apnea. Through the whole night data monitoring, the number of apnea is counted. The more apnea and respiratory disorder occur, the worse the respiratory quality is. The respiratory quality score is set as Soc_resp, with a full score of 100 points. Each time there is abnormal breathing, a certain number of points will be deducted. The severity of apnea is comprehensively evaluated by sound waveform characteristics and blood oxygen data. At the same time, the sound data is compared, and the time period when apnea and snoring marks appear at the same time is checked and marked, and the number of times is counted; For sleep quality analysis, sleep quality is related to the proportion of deep sleep, the number of sleep interruptions and the number of tossing and turning. The comprehensive score of sleep quality is calculated. The data used are as follows: Deep sleep ratio a1: Count the sleep data of the whole night and calculate the ratio of deep sleep time to total sleep time, which is recorded as deep sleep ratio a1. a1 is a value greater than 0 and less than 1. Turning over number a2: Turning over is a large action that is continuous in a short time. The turning over action is accurately judged and marked through the first body motion data and the second body motion data; Awakening times a3: The more awake times you have at night, the worse your sleep is; Breathing quality score a4: If the breathing quality assessment score is low, it indicates that the breathing quality during sleep is poor; Based on the above four data, the sleep score Soc_sleep is calculated as follows: Soc_sleep=SocA1+SocA2+SocA3+SocA4; SocA1=100*α1*(0.5+a1); SocA2=100*α2*cur(a2); SocA3=100*α3*(1-0.2*a3); SocA4=100*α4*a4; Among them, cur is a centrally symmetrical distribution curve. The greater the deviation from the center, the smaller the value of cur(x). α1, α2, α3 and α4 are the weights of each parameter in sleep assessment. The larger the weight, the greater the influence of this parameter on sleep quality assessment. Finally, the breathing quality assessment and sleep quality analysis are combined to output a comprehensive sleep analysis report.

3. A multi-device multi-sensor sleep monitoring system according to claim 2, characterized in that: Before monitoring, the data acquisition module and the data processing module are self-checked respectively to ensure that the data acquisition module and the data processing module work normally during the monitoring period.

4. The multi-device multi-sensor sleep monitoring system according to claim 3, characterized in that: The data processing module transmits the timely sleep analysis report and the comprehensive sleep analysis report to the device terminal.

Images